Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Excitatory Amino Acid Receptors at a Feedback Pathway in the Electrosensory System: Implications for the Searchlight Hypothesis NEIL J. BERMAN, JAMES PLANT, RAY W. TURNER, AND LEONARD MALER Department of Cellular and Molecular Medicine, University of Ottawa, Ottawa, Ontario K1H 8M5, Canada Berman, Neil J., James Plant, Ray W. Turner, and Leonard Maler. Excitatory amino acid receptors at a feedback pathway in the electrosensory system: implications for the searchlight hypothesis. J. Neurophysiol. 78: 1869–1881, 1997. The electrosensory lateral line lobe (ELL) of the South American gymnotiform fish Apteronotus leptorhynchus has a laminar structure: electroreceptor afferents terminate ventrally whereas feedback input distributes to a superficial molecular layer containing the dendrites of the ELL principle (pyramidal) cells. There are two feedback pathways: a direct feedback projection that enters the ELL via a myelinated tract (stratum fibrosum, StF) and terminates in the ventral molecular layer (VML) and an indirect projection that enters as parallel fibers and terminates in the dorsal molecular layer. It has been proposed that the direct feedback pathway serves as a ‘‘searchlight’’ mechanism. This study characterizes StF synaptic transmission to determine whether the physiology of the direct feedback projection is consistent with this hypothesis. We used field and intracellular recordings from the ELL to investigate synaptic transmission of the StF in an in vitro slice preparation. Stimulation of the StF produced field potentials with a maximal negativity confined to a narrow band of tissue dorsal to the StF. Current source density analysis revealed two current sinks: an early sink within the StF and a later sink that corresponded to the anatomically defined VML. Field potential recordings from VML demonstrated that stimulation of the StF evoked an excitatory postsynaptic potential (EPSP) that peaked at a latency of 4–7 ms with a slow decay ( Ç50 ms) to baseline. Intracellular recordings from pyramidal cells revealed that StF-evoked EPSPs consisted of at least two components: a fast gap junction mediated EPSP (peak 1.2–1.8 ms) and a chemical synaptic potential (peak 4–7 ms) with a slow decay phase ( Ç50 ms). The amplitudes of the peak and decay phases of the chemical EPSP were increased by depolarizing current injection. Pharmacological studies demonstrated that the chemical EPSP was mainly due to ionotropic glutamate receptors with both N-methyl-D-aspartate (NMDA) and non-NMDA components. NMDA receptors contributed substantially to both the early and late phase of the EPSP, whereas non-NMDA receptors contributed mainly to the early phase. Stimulation of the StF at physiological rates (100–200 Hz, 100 ms) produced an augmenting depolarization of the membrane potential of pyramidal cells. Temporal summation and a voltage-dependent enhancement of later EPSPs in the stimulus train permitted the compound EPSP to reach spike threshold. The nonlinear behavior of StF synaptic potentials is appropriate for the putative role of the direct feedback pathway as part of a searchlight mechanism allowing these fish to increase the electrodetectability of scanned objects. INTRODUCTION Vertebrate sensory transmission consists of ascending pathways leading from receptors to higher ‘‘integrative’’ regions and eventually to motor areas of the brain. At each level, local neuronal interactions generate receptive fields that characteristically increase in selectivity and complexity at succeeding levels of the neuraxis. Sensory systems also have extensive feedback projections and long-range horizontal interactions. These connections may permit regions outside the cell’s classic receptive field to modulate responses to receptive field input and may underlie ‘‘higher level’’ effects such as attention and adaptation to sustained input. This paper focuses on the synaptic physiology of a feedback projection in the electrosensory system. Electric fish generate electric organ discharges (EODs) used for electrocommunication (Heiligenberg 1991) and electrolocation (Bastian 1986a). Distortions in the fish’s EOD are detected by electroreceptors, which project to the medullary electrosensory lobe (ELL) (Carr and Maler 1986). The ELL is a laminar structure and consists of four segments (medial, centromedial, centrolateral, and lateral segments) optimized for processing of different features of the electrosensory input (Maler 1989; Shumway 1989). Pyramidal cells are the principle output neurons of the ELL and project topographically to the midbrain (torus semicircularis) and the isthmic nucleus praeminentialis (Pd) (Carr and Maler 1986). The ascending electrosensory projections have been mapped in detail up to sensorimotor interfaces in the optic tectum and diencephalon (tectum: Bastian 1982; diencephalon: Heiligenberg 1991; Rose and Heligenberg 1988), and recordings from periphery to diencephalon reveal the elaboration of receptive fields leading to outputs suitable for driving motor responses (Heiligenberg 1991). There are two excitatory feedback pathways to the ELL, both emanating from at least three cell types in Pd (Carr and Maler 1986; Sas and Maler 1983, 1987): stellate, multipolar, and tufted cells. Stellate cells project directly and topographically back to the ventral molecular layer (VML) of the ELL (direct feedback) via a myelinated fiber bundle (tractus stratum fibrosum, tSF; within the ELL this tract is designated the stratum fibrosum or StF) and terminates densely in the ventral-most portion of the ELL molecular layer (VML; Fig. 1) (Maler 1979; Maler et al. 1981b). Multipolar and tufted cells project to cerebellar granule cells overlying the ELL, and these cells in turn project parallel fibers to the dorsal molecular layer (DML) of the ELL (indirect feedback). There is also a direct inhibitory feedback projection from GABAergic bipolar cells in the hilar region of Pd, via fibers lying at the ventral aspect of tSF, to pyramidal cell somata (Maler and Mugnaini 1994). Bastian and coworkers have used a variety of physiological electrosensory inputs to characterize the responses of the cells giving rise to the direct (Bratton and Bastian 1990) 0022-3077/97 $5.00 Copyright q 1997 The American Physiological Society / 9k1d$$oc22 J277-7 09-05-97 11:58:59 neupa LP-Neurophys 1869 1870 N. J. BERMAN, J. PLANT, R. W. TURNER, AND L. MALER FIG . 1. Sites for recording and stimulating the stratum fibrosum (StF) projection in an electrosensory lateral line lobe (ELL) transverse section (level T-7) (Maler et al. 1991). Injection of wheat germ agglutinin (WGA)-horseradish peroxidase in isthmic nucleus praeminentialis (Pd) anterogradely labeled the stratum fibrosum, which terminated in the ventral molecular layer (VML) and retrogradely labeled pyramidal cells (Maler 1979; Maler et al. 1992). One basilar pyramidal cell ( right of – – – ) shows the orthogonal orientation ( – – – ) of their dendritic arbors to the ELL laminae. Inset: simultaneous recordings obtained from a pyramidal cell proximal apical dendrite ( bottom traces) and the mid-VML field potential ( top traces) in the centromedial segment during low- (thick trace) and high-intensity (thin traces) StF stimulation in the medial segment (*). AS, antidromic spike; BS, brain stem; EGp, eminentia granularis posterior (cerebellar granule cells); CLS, centrolateral segment; CMS, centromedial segment; DFL, deep fiber layer; DML, dorsal molecular layer; DNL, deep neuropil layer; GCL, granule cell layer; LS, lateral segment; MS, medial segment; PA, primary (electroreceptor) afferents; pf, parallel fiber; PCL, pyramidal cell layer; and PlL, plexiform layer. and indirect (Bastian and Bratton 1990) feedback pathways. Stellate cells have small receptive fields with phasic responses and appear to code for the movement of objects across their receptive fields; their response properties and the topographic nature of their feedback projections prompted Bratton and Bastian (1990) to hypothesize that the direct feedback pathway was involved in focusing the electrosensory system, i.e., a ‘‘searchlight’’ mechanism (Crick 1984; see also Maler and Mugnaini 1993, 1994). The indirect feedback pathway appears to regulate more global changes in electrosensory processing (Bastian 1986b,c; Bastian and Bratton 1990). Both direct and indirect feedback projections to the ELL molecular layer probably use the excitatory amino acid (EAA) glutamate as a neurotransmitter (Wang and Maler 1994). Studies using receptor binding (Maler and Monaghan 1991), in vivo pharmacological analysis (Bastian 1993), and in situ hybridization (Bottai et al. 1995–1997) have suggested that both N-methyl-D-aspartate (NMDA) and aamino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors are associated with the excitatory feedback input to the ELL. The development of selective ionotropic EAA receptor agonists and antagonists has facilitated analysis of their role in sensory systems; however, the interpretation of these studies has been hampered by the complexity of sensory processing in mammals. The simpler electrosensory system is more tractable for the study of low-level sensory processing. This study focuses on the synaptic physiology of the StF / 9k1d$$oc22 J277-7 (direct) feedback projection to the ELL and on the properties of the ionotropic EAA receptors involved in the putative electrosensory searchlight. METHODS Preparation of ELL slices A total of 87 weakly electric fish of the species Apteronotus leptoryhchus (Brown Ghost Knife Fish) were used. Transverse slices of the ELL were prepared as previously described (Mathieson and Maler 1988; Turner et al. 1994). Fish were anesthetized (MS-222), transferred to a foam-rubber–lined holder, and respirated with 11–14 ml/min of oxygenated water containing anesthetic during surgery. The brain stem was transected, glued to an aluminum block and 350–500 mm transverse ELL slices cut on a Vibratome into oxygenated ice-cold artificial cerebrospinal fluid [ACSF, which contained (in mM) 124 NaCl, 24 NaHCO4 , 10 D-glucose 1.25 KH2PO4 , 2 KCl, 2 MgSO4 , and 2 CaCl2 ; all chemicals from Sigma unless otherwise noted]. Slices were positioned rostral side up in an interface slice chamber and maintained at room temperature in oxygenated ACSF for a minimum of 1 h before recordings were begun. The StF was clearly evident (Mathieson and Maler 1988) and served as a landmark for the placement of stimulating and recording electrodes. Recordings were obtained exclusively from the centromedial segment of the ELL (Figs. 1 and 2). Stimulation of StF Stimulating electrodes were either bipolar (65 mm nichrome wire) or a monopolar tungsten electrode ( õ5-mm tip diameter). 09-05-97 11:58:59 neupa LP-Neurophys EXCITATORY AMINO ACID RECEPTORS AT A FEEDBACK PATHWAY 1871 ACSF) were 2 kynurenic acid, 1 3-((RS)-2-carboxypiperazin4-yl)-propyl-1-phosphonic acid (CPP, Research Biochemicals International, MA), 2 DL-2-amino-5-phosphonopentanoic acid (APV, Tocris, UK), and 1 6-cyano-7-nitroquinoxaline-2,3-dione [CNQX, Tocris; predissolved in 100 ml dimethyl sulfoxide]. Control microdroplet applications of ACSF alone had no effect on StFevoked responses. Recording of synaptic potentials Potentials recorded by glass micropipettes (1 M NaCl; 2–10 MV ) were filtered (DC-10 kHz), amplified (Axoclamp-2A, Axon Instruments), digitized (10–25 kHz), and stored on disk for off-line analysis (CLAMPAN, Axon Instruments; A/Dvance; IgorPro, Wavemetrics, OR). ANALYSIS OF FIELD POTENTIALS. Under each stimulus condition, 10–15 consecutive StF-evoked field potentials were averaged. Field potentials were mapped along the dendro-somatic axis of ELL pyramidal cells (dashed line in Fig. 1) by two methods. 1) StF field potentials were recorded from positions corresponding to visually identifiable anatomic structures within the ELL slice preparation, allowing for the construction of low-resolution spatial maps of StF-evoked field potentials (n Å 8). 2) High-resolution maps constructed by recording field potentials (50 mm depth) every 25 mm along most of the ELL pyramidal cell axis (n Å 3) were used for one-dimensional current source density (CSD) analysis. Standard methods for CSD analysis in slices were followed (BodeGreul et al. 1987; Richardson et al. 1987; Taube and Schwartkroin 1988) using the equation (Bode-Greul et al. 1987) d 2 (P)/dz 2 Å P(z / n ∗ Dz) 0 2 ∗ P(z) / P(z 0 n ∗ Dz)/(n ∗ Dz) 2 , where P is the evoked field potential, z is the spatial location, Dz is the sampling interval (25 mm), and n is the integration grid (n Å 3). INTRACELLULAR RECORDING. Intracellular recording pipettes (3 M potassium acetate, 70–200 MV ) were advanced into the CMS pyramidal cell layer using a microdrive (Burleigh Instruments, NY). Recordings from pyramidal cell somata or proximal apical dendrites (Turner et al. 1994) were amplified by an Axoclamp 2A preamplifier, filtered (DC-10 or 25 kHz) and digitized for off-line analysis. Results are given as means { SD, and statistical analysis is by Student’s t-test. FIELD POTENTIAL RECORDING. FIG . 2. Laminar profile and current source density (CSD) analysis of StF-evoked currents. A: schematic diagram of a tissue slice at the level of the ELL. Recordings were made in the centromedial segment along a track parallel to the pyramidal cell soma-dendritic axis (rrr). Refer to Fig. 1 for abbreviations. B: field potentials recorded at several key sites from a laminar profile of StF-evoked responses with distance noted with respect to the StF recording site; r, stimulus time (in this and subsequent figures, artifacts were digitally suppressed). The VML response ( – – – ) is displayed at half the gain of other potentials (4 mV on calibration bar). *, peak negativity in StF (presumed fiber volley). C: superimposed CSD measurements from 3 locations shown in B. D: a spatial profile of CSD over the pyramidal cell axis at a latency corresponding to the peak of the negative-going field potential in VML shown in B. Profile is aligned with a schematic of a basilar pyramidal cell, shown with StF afferents contacting its apical dendrites in the VML. The stimulating electrode was positioned at the dorsal aspect of the StF within the medial segment to prevent direct electrical activation of centromedial segment interneurons, pyramidal cell efferent axons within the plexiform layer (Turner et al. 1994) or GABAergic axons from Pd bipolar cells (Maler and Mugnaini 1994). Stimulus timing was computer-controlled (pClamp, Axon Instruments, CA; A/Dvance, McKellar Designs, BC; Master-8, AMPI, Israel) and delivered via constant voltage or current stimulus isolation units (Digitimer, UK; 10–80 V or 30–800 mA; 0.1–0.5 Hz; tetanic stimulation: 100–200 Hz for 100 ms). Drug application Drugs were applied using two methods. 1) Bath applications for field potential experiments. After 1 h of normal ACSF, perfusate was switched for ú1 h to either 4 mM Mn 2/ -ACSF solution [containing (in mM) 129 NaCl, 10 D-glucose, 3.25 KCl, 0.2 CaCl2 , 11.4 tris(hydroxymethyl)aminomethane, and 20 N-2-hydroxyethylpiperazine-N *-2-ethanesulfonic acid] or Mg 2/ -free ACSF. 2) Micro-droplet applications: Drugs were applied by brief pressure ejection ( õ10 psi; 80–190 ms) from pipettes (3- to 7-mm tips) placed at the surface of the slice centered in the VML dorsal to the recording site (Turner et al. 1994) (NeuroPhore BH-2, Medical Systems, NY). For microdroplet application, drug concentrations (in mM in / 9k1d$$oc22 J277-7 RESULTS The ELL is a laminar structure with separate layers for electrosensory afferent (deep fiber layer) and feedback input (VML, DML). The majority of efferent neurons are within the pyramidal cell layer, whereas most interneurons are located in the granular cell layer. The StF is a myelinated fiber tract the unmyelinated terminal fibers of which branch densely in the VML, which occupies Ç100–150 mm (15– 20%) of the total molecular layer width (measured along the pyramidal cell dendritic axis: dashed line, Fig. 1). Pyramidal cell axons run in the plexiform layer to exit at the medial aspect of the ELL. As seen in Fig. 1, the dendritic axis of pyramidal cells is orthogonal to the ELL cellular and fiber laminae (Maler 1979); this geometric arrangement allows field potential recordings of activity evoked by stimulation of input or output fiber tracts. Because the StF is in close proximity to both the plexiform layer (pyramidal cell efferents) and the DML, it was important to establish a means of selectively stimulating StF afferent inputs. As shown by Fig. 1, inset, StF stimulation at relatively low intensities ( õ70 V) selectively activated a prominent EPSP in a recording from a pyramidal cell proxi- 09-05-97 11:58:59 neupa LP-Neurophys 1872 N. J. BERMAN, J. PLANT, R. W. TURNER, AND L. MALER mal apical dendrite. Stimulation at intensities ú 70 V evoked both a short-latency (antidromic) spike and a reduced excitatory postsynaptic potential (EPSP). These two potentials are reflected in field recordings in the mid-VML as a shortlatency negativity (0.96 { 0.42 ms; n Å 8) and a later, longer lasting field EPSP (3.2 { 0.56 ms; n Å 44). The first response results from current spread to the plexiform layer, which elicits an antidromic spike that backpropagates into dendritic regions (Turner et al. 1994). The second response represents synaptic depolarization of pyramidal cell apical dendrites in the VML (see below). These identifying characteristics allowed stimulus intensity in all subsequent experiments to be set at a level that selectively activated StF afferent inputs. StF-evoked EPSPs FIELD POTENTIAL RECORDINGS. Laminar profiles of StFevoked field potentials were constructed, and CSD analysis carried out to examine the site for termination and the nature of synaptic inputs activated by StF stimulation (Fig. 2). The largest field potential response evoked by StF stimulation always was recorded in the VML as a biphasic potential consisting of a short-duration positivity followed by a rapid negative-going potential (peak latency 3.2 { 0.56 ms, 03.02 { 1.50 mV; n Å 44) with a long duration ( ú30 ms; Fig. 2B). The initial positivity in the VML was correlated temporally with a sharp, short-duration negativity in the StF (1.7 { 0.30 ms; n Å 35; Fig. 2B, StF trace, *). The next negative peak in the StF response was lower in amplitude and longer lasting, decaying to baseline over Ç20 ms. A small broad field positivity was recorded at the level of the pyramidal cell body layer and the proximal DML with a peak latency slightly delayed to that recorded in the VML (4–7 ms; Fig. 2B, PCL trace, *). In the pyramidal cell body layer, single-unit spike discharge was sometimes superimposed on the peak of the positivity (not shown). Much smaller potentials were recorded in the mid-DML and granular cell body layers, most often as a triphasic potential or monophasic positivity, respectively; no field potentials could be detected in the deep fiber layer (DFL) or distal DML. CSD analysis indicated that the shortest latency current sink (net inward current) was located in the StF at a latency similar to a current source in the VML (Fig. 2C). This was followed by current sinks of longer duration in both the StF and VML, which was balanced by a current source of similar duration in the pyramidal cell body layer. Figure 2D illustrates the sink-source relationships over the pyramidal cell axis at the latency of the peak current sink in VML. This reveals that a current sink extended for Ç100 mm dorsal to the pyramidal cell body layer, a location consistent with the anatomic definition of the VML (125 mm thick in the example of Fig. 1). This current sink was flanked by prominent current sources in both the pyramidal cell layer and the proximal DML. These results verify anatomic reports that the termination field of StF afferents is restricted to the StF and VML region (Fig. 1) (Maler 1979; Maler et al. 1982) and that our stimulus did not substantially activate the excitatory parallel fiber input in the adjacent DML. The primary StF response is one of excitatory synaptic drive, with the shortest latency current / 9k1d$$oc22 J277-7 sink located in the StF, followed by that in the VML. This is consistent with anatomic reports that StF fibers branch sharply to course dorsally and form synaptic contacts with pyramidal cell apical dendrites in the StF and then VML (Maler 1979; Maler et al. 1981a). This ventro-dorsal axonal projection and synaptic termination pattern may account for the coincidence of an initial sharp current sink in the StF with a sharp current source in the VML. The latter then may arise in part as a passive current source for active spike discharge and/or synaptic depolarizations of dendritic membrane in the more ventral StF (see below). PHARMACOLOGY OF STF-EVOKED FIELD POTENTIAL. The above identification of StF-evoked responses was further tested using blocking agents of synaptic transmission (Fig. 3). The StF-evoked VML potential was decreased significantly by bath application of 4 mM Mn 2/ in low Ca 2/ (0.2 mM)-containing medium, decreasing from 3.0 { 1.5 to 1.73 { 1 mV (P õ 0.01; n Å 6; Fig. 3A). The negative peak of the remaining triphasic potential had a latency of 2.55 { 0.73 ms and was the expected response for a fiber volley in StF afferent fibers (Fig. 3A; note that subsequent FIG . 3. Pharmacology of StF-evoked extracellular field potentials in the VML. Ai: superimposed recordings of the VML field potential under control (1, con) conditions and in the presence of 4 mM Mn 2/ and 0.2 mM Ca 2/ medium (2, Mn 2/ ) to block synaptic transmission. Aii: synaptic component as the difference between the control and test potentials (1 and 2). B–D: superimposed recordings of the VML field potential under control (con) conditions and after exposure to kynurenic acid (B, Kyn), 3-((RS)-2carboxypiperazin-4-yl)-propyl-1-phosphonic acid (CPP; C), and 1 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; D). Note the partial reduction by kynurenic acid and CPP of both an early ( B and C; õ10 ms) and late phase (*) of the field response, whereas CNQX (D) completely blocks the synaptic response. E: superimposed recordings of the VML field potential under control (con) conditions and after exposure to nominally Mg 2/ -free medium for 1 h. Note the pronounced augmentation of both the peak and late phase of the synaptic response by low Mg 2/ medium. F: superimposed recordings of a VML field potential augmented under 0 Mg 2/ medium and test responses to subsequent application of CNQX and CNQX with CPP. Under these conditions, CNQX only partially blocks the synaptic response, and the remaining component was blocked almost entirely by CPP. Field recordings recovered to near control levels after washout (n Å 2) of CNQX, kynurenic acid, or Mg 2/ -free medium after 1–2 h. 09-05-97 11:58:59 neupa LP-Neurophys EXCITATORY AMINO ACID RECEPTORS AT A FEEDBACK PATHWAY experiments described below indicated an additional contribution by an electrotonic synaptic component). Subtraction of the remaining response in Fig. 3A revealed a long-lasting, synaptically mediated component of the StF-evoked response with a peak latency of 4.3 { 1.3 ms (n Å 6), a value similar to the peak latency of EPSPs recorded with intradendritic impalements (see below) and significantly greater than the latency of the fiber volley (P õ 0.05). These data indicated that the StF fiber volley slightly overlapped the early component of the field synaptic response, contributing to this aspect of field potential recordings in normal medium. Nevertheless, there was sufficient separation between presynaptic fiber and synaptic potentials to pharmacologically distinguish an early and late phase of the EPSP (see below). Fiber volley potentials recorded during Mn 2/ perfusion at 100-mm intervals along the StF were used to calculate a StF conduction velocity of 0.31 m/s ( Ç237C). An additional conduction delay of Ç0.28 ms was introduced by the unmyelinated collaterals of the StF fibers, which ascend and synapse in the VML. Previous studies of receptor ligand binding (Maler and Monaghan 1991) and application of EAA agonists and antagonists in vivo (Bastian 1993) indicated that EAA receptors are present in the ELL molecular layer. Bastian (1993) in particular demonstrated that pressure ejection of the glutamate receptor agonists AMPA or NMDA in the ELL molecular layer increased the excitability of pyramidal cells (although his injections were not specifically restricted to either the DML or VML). He also was able to antagonize the response to these agonists with EAA antagonists [6,7-dinitroquinoxaline-2,3-dione (DNQX), APV]. We therefore tested several antagonists of ionotropic EAA receptors to identify the contribution of glutamate receptor subtypes to the StF EPSP. Pressure ejection of the broad-spectrum antagonist kynurenic acid into the VML resulted in a significant (P õ 0.001; n Å 7) and reversible reduction in the StF-evoked field EPSP (Fig. 3B). The small residual late component of the VML response (Fig. 3B, *) proved sensitive to subsequent perfusion of Mn 2/ -containing medium (not shown). Note that kynurenic acid partially reduced the amplitude of the early field positivity in the VML, again suggesting a relationship between this potential and inward synaptic current in the underlying StF (Fig. 3B). Pressure ejection of the NMDA-receptor antagonist CPP also significantly (P õ 0.001; n Å 12) and reversibly reduced the peak amplitude and late phase of the StF-evoked EPSP (early positive and negative components) to about half of control values (Fig. 3C). In contrast, pressure ejection of the non-NMDA–receptor antagonist CNQX into the VML completely blocked the synaptic component of the StFevoked EPSP, leaving only the biphasic presynaptic fiber potential (Fig. 3D; n Å 5). Subsequent application of CPP had no effect on the response left after exposure to CNQX (data not shown). These studies indicated that the StF-evoked EPSP is mediated by ionotropic glutamate receptor subtypes but produced two unexpected results. First, the ability of CPP to reduce the amplitude of both an early ( õ10 ms) and late phase of the EPSP suggested the unusual contribution of a fast / 9k1d$$oc22 J277-7 1873 NMDA-receptor component to the StF-evoked EPSP. Second, a complete blockade of the EPSP by the non-NMDA– receptor antagonist CNQX suggests a possible cross-reactivity with NMDA receptors in this preparation. We therefore used a perfusate nominally free of Mg 2/ to test these drugs under conditions expected to augment NMDA-receptor transmission. Perfusion with 0 Mg 2/ medium (Fig. 3E) significantly increased the peak of the StF-evoked VML EPSP (133 { 23.9%; P õ 0.02; n Å 5), as well as the late phase of the EPSP. This is consistent with the idea that NMDA receptors contribute substantially to both the peak and late phase of the EPSP. In 0 Mg 2/ medium, CNQX substantially reduced the amplitude of both the peak and late phases of the 0 Mg 2/ -enhanced EPSP (n Å 7; Fig. 3F). However, unlike the result found in normal medium, subsequent application of CPP further reduced both the peak and late phases of the StF-evoked EPSP (Fig. 3 F). This result suggested that the ability of CNQX to block the NMDA-receptor component of the StF-evoked EPSP may be partially due to the Mg 2/ -dependent voltage sensitivity of the NMDA receptor. The 0 Mg 2/ experiments support the conclusion that NMDA receptors mediate both an early and a late phase of the StF EPSP. They also suggest that CNQX exhibits some cross-reactivity with NMDA receptors, although a CNQXresistant, CPP-sensitive component can be identified in the presence of 0 Mg 2/ . Interestingly, Bastian (1993) also reported that, although APV was a selective antagonist of pyramidal cell responses to NMDA in vivo, DNQX (a nonNMDA receptor antagonist similar to CNQX) also antagonized the response of ELL pyramidal cells to both AMPA and NMDA. Intracellular recording Recordings were obtained from ú100 somatic and 13 proximal apical dendritic penetrations of pyramidal cells (somatic vs. dendritic recordings were distinguished using criteria established by Turner et al. 1994). Resting membrane potential (RMP) and input resistance (Rinput ) were similar in somatic (RMP Å 072.3 { 7.9 mV, n Å 40; Rinput Å 18.8 { 7.6 MV, n Å 42) and dendritic recordings (RMP Å 072.6 { 8.7 mV; Rinput Å 20.1 { 8.7 MV ). Stimulation of StF always evoked EPSPs and often also evoked inhibitory postsynaptic potentials (IPSPs) of variable amplitude; data on IPSPs generated by StF stimulation will be presented elsewhere. Consistent with the VML field recordings (Fig. 3), StFevoked EPSPs consisted of a rapid depolarization followed by a prolonged decay phase (Fig. 4, A and C; additional examples in Fig. 6A). Because the late phase did not reverse on depolarization, it was an EPSP rather than a reversed IPSP. There were three components of the StF-evoked EPSP: a rapid electrotonic EPSP followed by the transmitter mediated peak and late phase. In cases where the stimulus artifact was relatively small (n Å 17), StF stimulation resulted in a small short-latency EPSP (1.55 { 0.59 ms; Fig. 4A) with an amplitude of 3.1 { 1.8 mV. This fast component of the EPSP could not be blocked by excitatory amino acid antagonists (CNQX / CPP; Fig. 4, A and B) or low Ca 2/ , high Mn 2/ application (not shown). Its amplitude was voltage insensitive but in- 09-05-97 11:58:59 neupa LP-Neurophys 1874 N. J. BERMAN, J. PLANT, R. W. TURNER, AND L. MALER FIG . 4. Intracellular electrophysiology of pyramidal cell responses to StF stimulation. A: responses to a series of graded stimulus intensities in a different cell before (top trace in each set, Control) and after application of CNQX and CPP (bottom trace in each set). Amplitudes of 2 clearly separable peaks (1, 1.1 ms; 2, 5.6 ms) are plotted in B. Traces are connected by dashed line where artifact has been blanked (this and subsequent figures). B: early peak (top) scaled with stimulus intensity but was not significantly affected by CNQX and CPP. Later peak (bottom) similarly scaled with stimulus intensity but was strongly sensitive to CNQX and CPP. C: somatic excitatory postsynaptic potentials (EPSPs) with multiple peaks (r ) recorded while cell was current clamped to various indicated prestimulus (vertical – – – ) membrane potentials. Late phase of the EPSP evoked at depolarized potentials lasted 40–60 ms. Note that, in this case, at depolarized potentials ( 066 and 062 mV) there is a clear separation of the EPSP into early (3 and 6 ms; arrows 1 and 2) and late (arrow 3) peaks. D: traces ( 078, 066, and 062 mV) from C overlaid to show clearly the enhancement of EPSPs at depolarized potentials. E: amplitude plots of EPSPs in C vs. prestimulus membrane potential. Earliest (3 ms) EPSP is insensitive to membrane potential, whereas the later components of the EPSP show strong inward rectification near and above resting membrane potential ( 068 mV). F: membrane potential during the EPSPs in C–E can be reset by the occurrence of a spike; note that the potential decays to resting levels ( – – – , 062 mV) after a spike (2 of the 3 current injections evoked spikes). creased directly with stimulating current (Fig. 4B), as would be expected by the recruitment of additional fibers with increasing current. In the presence of CNQX / CPP, this potential was observed to rapidly decay and was negligible after 5–8 ms (Fig. 4A; see also Fig. 8A, *). The CNQX/ CPP-sensitive peak of the EPSP had a much longer latency of 4.6 { 1.5 ms and an amplitude of 4.3 { 2.2 mV (data from the same recordings). The fast depolarization is therefore probably an electrotonic EPSP transmitted through the morphologically identified gap junctions between StF fibers and the proximal apical dendrites of ELL pyramidal cells (Maler et al. 1981b). Because the main EPSP peaked considerably later than the early electrotonic component, our measurements below reflect predominantly chemical transmission in the VML. When evoked at RMP, the somatic EPSP had a peak latency of 6.1 { 2.4 ms (n Å 86) and a peak amplitude of 5.89 { 1.27 mV for intensities just subthreshold for spike discharge. The latency to peak was highly variable (e.g., Fig. 4, A and C; range, 2.8–15 ms; typically 4–7 ms), a result attributable to three factors: the precise location of the stimulating electrode, the membrane potential (see below), and the amplitude of StF-evoked IPSPs (strong IPSPs trun- / 9k1d$$oc22 J277-7 cated the EPSP and shifted its apparent peak to shorter latencies). The EPSP recorded from the apical dendrite peaked at a slightly shorter latency of 4.32 { 1.71 ms and amplitude of 5.70 { 2.06 mV (n Å 13). Both the early ( Ç6 ms) and late ( Ç20 ms) phases of StF-evoked EPSPs were clearly voltage dependent (n Å 5 dendritic; n Å 18 somatic). Depolarization to more than 068 mV typically produced a dramatic increase in the amplitude of the peak and late phase ( ú10-ms latency) of the EPSP, with the peak shifting to the late phase as it became enhanced at depolarized levels (Fig. 4, C–E; see also Fig. 5B). The combination of these factors could result in a plateau potential that persisted for °60 ms. Because we did not block IPSPs that are evoked by StF stimulation (Berman and Maler, unpublished observations), this may underestimate the duration of the plateau potential. At depolarized levels (more than -65 mV), action potentials often were evoked at a latency of 15–20 ms (Fig. 4F). The action potential also reset the membrane potential, again suggesting that the StF-evoked EPSP is highly voltage sensitive. In some recordings (Fig. 4, C and D), depolarizing the cell with current injection revealed an early (3 ms) and late (5–7 ms) peak. The early peak had a considerably longer 09-05-97 11:58:59 neupa LP-Neurophys EXCITATORY AMINO ACID RECEPTORS AT A FEEDBACK PATHWAY latency than the putative electrotonic response, was not voltage dependent (Fig. 4E), and could be blocked by CNQX (see below). These results suggest that there may be at least three components of the chemically mediated StF-evoked EPSP: an early (3–4 ms) voltage-independent phase, a slightly slower (5–7 ms) voltage-dependent phase, and a voltage-dependent late phase. In most cases, however, there was only a single clear peak of the chemical EPSP (Fig. 4 A; other examples also can be seen Figs. 6A and 8A below). StF fibers originate from stellate cells of Pd (Bratton and Bastian 1990; Maler et al. 1982; Sas and Maler 1983); Bratton and Bastian (1990) have demonstrated that, with physiological sensory input, Pd stellate cells discharge in short bursts ( Ç100 ms) at rates of 50–200 Hz. We therefore attempted to mimic natural stimulus patterns in vitro. Field potential recordings revealed an apparent strong paired pulse facilitation (PPF; Fig. 5A) at a 10-ms interpulse interval (180%); this facilitation declined to baseline by 120 ms. Intracellular recording also revealed PPF (Fig. 5B). Because this could be seen at hyperpolarized membrane potentials after the EPSP had almost returned to baseline (40 ms after stimulus onset), it suggests that there may be classic presynaptic PPF (Zucker 1989) at StF synapses. At rest or when the cell is depolarized (Fig. 5B), the facilitation of the peak of the second EPSP mostly can be accounted for by simple summation of the second EPSP with the decaying late phase of the first EPSP. FIG . 5. Paired pulse facilitation (PPF) of StF synaptic potentials. A: field recordings in VML (12 overlaid traces). After the initial stimulus (first EPSP), a second stimulus was delivered at 10-ms intervals °120 ms. Inset: summary (n Å 10) of the decay of PPF from 180% at 10 ms to baseline at intervals ú120 ms. – – – , prestimulus baseline and control (first) EPSP peak. B: intracellular recording with paired pulses at an interval of 40 ms. At hyperpolarized potentials ( 079 mV, lower dashed line), the EPSP decayed to 0.6 mV above baseline by the time of the second stimulus onset and the second EPSP is 1.2 mV greater than the first (PPF Å 0.6 mV). At resting membrane potential (RMP; 068 mV) or depolarized potentials (not shown), the second EPSP was evoked on the late phase of the first EPSP (1.9 mV ú prestimulus baseline); the increase in the second EPSP (2.1 mV) can be attributed mainly to summation with the late phase of the first EPSP. Note that in this recording both the early and late phase of the EPSP are voltage sensitive, and there is only one clear peak. – – – , prestimulus baseline and control EPSP peak for hyperpolarized case. / 9k1d$$oc22 J277-7 1875 As apposed to single pulse stimuli (Fig. 6 A), stimulation of StF at frequencies of 50–200 Hz produced an augmenting depolarization of the membrane potential of pyramidal cells that plateaued after 50–100 ms (Fig. 6, B–D). In cases where inhibition was more prominent, the plateau depolarization was reached earlier (Fig. 6C) or replaced by a compound IPSP (e.g., Fig. 8 B) (Berman and Maler, unpublished observations of cells depolarized with current injection). When inhibition was not prominent (IPSPs not seen when cell was depolarized), stimulus trains produced long depolarizing tails that could exceed 500 ms (Fig. 6, B and D). When the cell was depolarized slightly, the slow depolarization generated repetitive spiking (Fig. 6B, inset), indicating that it was not a reversed IPSP. Within the stimulus train window, although the second EPSP was typically larger than the first, the relative amplitude (from the immediately preceding membrane potential) of subsequent individual EPSPs in the train generally remained stable (Fig. 6, C and D). Unlike the situation after single shocks (cf. Fig. 4F), spikes evoked during a train were not able to reset the membrane potential (Fig. 6D). Presumably this is at least partly due to the inward currents produced by the summating late phases of the EPSPs greatly exceeding the spike repolarizing K / currents; a time-dependent attenuation of repolarizing K / currents during the train (Turner et al. 1994) also may contribute to this observed effect. The augmenting depolarization produced by tetanic stimulation was due to both temporal summation of the evoked EPSPs and the voltage-dependent increase in the late phase of the EPSP. For single EPSPs during the train (Fig. 7, A and B), if the cell became sufficiently depolarized during the tetanus (despite the hyperpolarizing holding current: Fig. 7B), the voltage dependence of the EPSP peak became evident. When stimulation occurred with the cell at resting potential ( –65 mV), the enhanced (due to their voltage dependence) and summated EPSPs crossed threshold and triggered spikes (Fig. 7C). The posttetanic depolarization (the tail described above) was also voltage dependent (data not shown) and was presumably due to nonlinear summation of the late, voltage-dependent phase of single EPSPs. Thus the voltage dependence of StF-evoked EPSPs is an important determinant of the response of ELL pyramidal cells to physiologically patterned inputs. INTRACELLULAR RECORDING: PHARMACOLOGY. Consistent with the results from the extracellular studies, pressure application of APV or CPP into the region of the VML ( n Å 44 ) consistently reduced the earliest chemical component of the StF-evoked EPSPs by 40 – 50% ( Fig. 8 A ) . The effect on the slow component of the evoked responses was variable and ranged from no effect to complete blockade; Fig. 8 A illustrates a typical case with a reduction of the late phase of the EPSP. These results therefore suggest that Ç50% of the peak EPSP ( õ10 ms ) can be accounted for by NMDA-receptor activation. The slow voltage-dependent component of StF-evoked EPSPs also is mediated partially by NMDA receptors, although its lability and overlap with IPSPs makes it difficult to quantify the contribution of NMDA receptors. Similar to the results of the field potential studies, applica- 09-05-97 11:58:59 neupa LP-Neurophys 1876 N. J. BERMAN, J. PLANT, R. W. TURNER, AND L. MALER FIG . 6. Intracellular electrophysiology of StF-evoked EPSPs. A: single shock evoked EPSP. Note initial fast EPSP, followed by a slowly decaying tail depolarization (65 ms duration). B: same cell as A, but with train (10 pulses, 100 Hz) stimulation. Summating EPSPs during the train evoke several spikes (truncated), followed by a slow tail that lasts for hundreds of milliseconds after the last stimulus. When cell is depolarized (inset, 0.2 nA, 800 ms), the slow tail evokes repetitive spikes. C: different cell: summation of train stimulation-evoked EPSPs caused the slow depolarizing wave (shaded region, top). Examination of the individual EPSPs (bottom trace) shows that the profile of each EPSP is similar, but the initiation of each EPSP rides on the depolarized tail of the previous EPSP, causing significant summation. Markers indicate stimulus timing. D: effect of spiking on the depolarizing wave underlying the train response. With (gray trace) or without (dark trace) a spike the membrane potential followed a similar trajectory during repetitive StF stimulation. Same cell as A and B. tion of CNQX (n Å 30) completely blocked the StF-evoked EPSP remaining after CPP or APV treatment; a small, putatively gap junction, component remained in most cases (Fig. 8A). Combined application of CNQX and APV (or CPP) often unmasked a StF-evoked IPSP (Fig. 8, A and B); this IPSP probably emanates from Pd bipolar cells (Maler and Mugnaini 1994) and will be described elsewhere (Berman and Maler, unpublished observations). As already shown in field recordings, when applied first, CNQX blocked the StFevoked EPSP, and subsequent application of CPP or APV had no further effect (n Å 6). The small depolarization that remained in some cases after application of CPP / CNQX (e.g., Fig. 8A) was resistant to further block by Mn 2/ (not shown) and therefore presumably reflects an electrotonic component of the EPSP. We attempted to assess whether depolarizing the cell after application of CNQX would reveal a voltage-dependent NMDA component. However, depolarization induced numerous spontaneous transient depolarizations with a time course similar to that of StF-evoked EPSPs (as reported by Mathieson and Maler 1988); this precluded visualization of a possible residual NMDA component of the EPSP. Application of APV or CPP also attenuated the augmenting potential seen during stimulus trains (Fig. 8B) and greatly reduced the number of spikes produced by tetanic stimulation (Fig. 8B, inset). Subsequent application of CNQX completely blocked train-evoked spiking (n Å 28; Fig. 8B, inset) and left only the electrotonic EPSP. This electrotonic EPSP is likely to cause the summation during the tetanus that survived APV / CNQX, probably via an interaction with intrinsic currents (e.g., INaP ) (Mathieson and Maler 1988; Turner et al. 1994) because the EPSP decayed rapidly ( õ5 ms, Fig. 8A, *). The long posttetanic depolarization was diminished greatly by APV or CPP (Fig. 8B). / 9k1d$$oc22 J277-7 Subsequent CNQX applications eliminated the remaining slow depolarization (Fig. 8B). DISCUSSION In this study, we have shown that the direct feedback pathway from the rhombencephalic n. praeminentialis remains intact in ELL brain slices. Although the StF is close to pyramidal cell efferent fibers in the plexiform layer, correct placement of the stimulating electrode and limiting stimulation voltages prevented antidromic activation. Physiological (CSD) and anatomic estimates of the thickness of the VML coincide, suggesting that StF stimulation did not recruit DML fibers. The interpretation of our results presented below therefore is based on specific activation of the StF, the direct feedback projection to the ELL. We have demonstrated that StF stimulation evokes a mixed electrotonic and chemical EPSP; this is consistent with a previous morphological study (Maler et al. 1981b). The role of the electrotonic component of the StF-evoked EPSP is unclear, and the discussion below focuses on the larger chemical component of the StF-evoked EPSP. The pharmacology of the StF-evoked EPSP suggests that it is mediated mainly by NMDA and non-NMDA EAA receptors. These results are consistent with the identification of glutamate as the StF transmitter (Wang and Maler 1994), the presence of NMDA and AMPA binding sites in the VML [kainic acid binding sites are not present in the VML (Maler and Monaghan 1991)], and Bastian’s (1993) demonstration that pyramidal cells discharge vigorously in response to application of AMPA and NMDA in vivo. The CNQX-sensitive component of the EPSP had a rapid onset (latency to peak of Ç3 ms) and decay and was not voltage sensitive, consistent with the physiology of the mammalian AMPA receptor. 09-05-97 11:58:59 neupa LP-Neurophys EXCITATORY AMINO ACID RECEPTORS AT A FEEDBACK PATHWAY 1877 to more than 060 mV before it contributes appreciable synaptic current (Hestrin et al. 1992). Our CNQX results are perhaps not surprising as DNQX greatly attenuates the response to application of NMDA in the ELL molecular layer in vivo (Bastian 1993; see also Drejer and Honore 1988). Indeed NMDA receptor (NR1 subunits) of ELL pyramidal cells contain a 5 * end insertion (Bottai et al. 1996) that has been shown to confer marked CNQX sensitivity to mammalian NMDA receptors (Hollmann et al. 1993). Additionally we have shown that the CNQX block of NMDA receptors in the ELL is due partly to the voltage sensitivity of the NMDA receptor, so that depolarization by the non-NMDA–receptor component of the EPSP is required to relieve their Mg 2/ block; similar results have been reported in the hippocampal slice (Blake et al. 1988). The ability of CNQX to antagonize the NMDA-receptor component of synapses in the ELL molecular layer is there- FIG . 7. Voltage dependence of train-evoked EPSPs. A: current injection ( 00.3–0.1 nA) during tetanic stimulation of StF (100 Hz, 10 pulses, shaded region). Note characteristic ramp of summating EPSPs at membrane potentials ranging from 074 to 065 (rest) mV; at rest 3 spikes are initiated by the StF-evoked EPSPs. Shaded region expanded in B and C. B: when normalized for prestimulus membrane potential, the 074 and 071 mV traces, including EPSPs, superimpose; at these membrane potentials, the ramp is due entirely to temporal summation of the slow tail of the EPSPs. At 068 mV (thick trace) individual EPSPs still superimpose, but the underlying depolarizing wave shows inward rectification. C: at 065 mV (thick trace), the peaks of individual fast EPSPs show inward rectification. Three EPSPs evoke spikes (truncated) with prominent afterhyperpolarizations. Stimulus timing is indicated ( m ). The presence of a NMDA-receptor component of the StFevoked EPSP additionally is supported by the sensitivity of the EPSP to Mg 2/ and voltage. Further, Bottai et al. (1995– 1997) recently have cloned the A. leptorhynchus NMDA receptor (NR1 subunit) and demonstrated by in situ hybridization that NR1 mRNA is enriched highly in ELL pyramidal cells, including their proximal apical dendrites. There are, however, three apparently anomalous results with respect to the physiology of NMDA-receptor–mediated transmission in the VML. 1) CNQX appears to antagonize completely the StF-evoked EPSP, leaving no evident residual NMDA component unless augmented by perfusion with 0 Mg 2/ ACSF. 2) The NMDA-receptor component of the EPSP is prominent at the peak of the EPSP ( õ10-ms latency) as well as during its slow decay phase. It commonly is asserted that NMDA receptors contribute mainly a slow component to EAA synaptic transmission (Collingridge et al. 1988; Edmonds et al. 1995; Forsythe and Westbrook 1988; Hestrin et al. 1992). 3) The NMDA channels contributing to the StF-evoked EPSP appear to conduct at more negative membrane potentials (less than -60 mV) than the hippocampal NMDA receptor, which requires depolarization / 9k1d$$oc22 J277-7 FIG . 8. Pharmacology of the StF-evoked EPSPs recorded intracellularly. Effects of a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and N-methyl-D-aspartate (NMDA) antagonists on the EPSPs evoked by single and train pulses. A: staggered drug applications showing NMDA and non-NMDA components of the EPSP. DL-2-amino-5-phosphonopentanoic acid (APV) blocked about half of the EPSP peak amplitude and the tail depolarization. Subsequent addition of CNQX (APV / CNQX) blocked the remaining potential, leaving a small depolarized ‘‘hump’’ (*) followed by a hyperpolarization. Artifact blanked for clarity. B: train-evoked depolarizations with spike rates plotted in inset. Train stimulation (20 pulses, 200 Hz) evoked a strong excitatory response (Con), spiking (Con, inset), and depolarization, which outlasted the stimulus. APV reduced the evoked spike rate (inset) and blocked a major portion of the depolarization after the train. CNQX further reduced the slow depolarizing wave, but its major effect was to further reduce the excitatory response (and eliminate spiking, inset) during the train. These recordings also illustrate that the large inhibitory postsynaptic potentials (IPSPs) evoked by StF stimulation are blocked only partially by CNQX; a detailed analysis of the various components of these IPSPs will be presented elsewhere. Traces are averages of 3 trials. 09-05-97 11:58:59 neupa LP-Neurophys 1878 N. J. BERMAN, J. PLANT, R. W. TURNER, AND L. MALER fore probably due to both the molecular characteristics of the NR1 subunit expressed by pyramidal cells and the voltage dependence of the NMDA receptor. The slow time course of the NMDA-receptor component of EAA transmission has been attributed to NMDA channels opening after a long delay (reviewed in Edmonds et al. 1995). Recent kinetic experiments, however, have demonstrated that NMDA channels open, on average, Ç10 ms after agonist binding (Dzubay and Jahr 1996); this is only slightly longer than the apparent peak of the NMDA-receptor component of the StF-evoked EPSP. In addition, fast ( õ10 ms) NMDA-receptor–mediated transmission has been demonstrated in mammalian sensory systems: retino-geniculate EPSPs (Esguerra et al. 1992), somatosensory cortex (Armstrong-James et al. 1993), and visual cortex (Shirokawa et al. 1989). Electrophysiological analysis of cloned NMDA receptors further demonstrates that combinations of NR1 with different NR2 subunits can result in functional receptors with markedly different kinetics (Monyer et al. 1992). StF-evoked EPSPs decay far more rapidly ( õ60 ms) than the NMDA component of cortical EPSPs (time constant of 60–150 ms) (Hestrin et al. 1992). This is likely due to the activation of ELL inhibitory interneurons because application of g-aminobutyric acid-A (GABAA ) antagonists prolongs the EPSPs to ú100 ms (Berman and Maler, unpublished observations). Stimulus trains do cause prolonged EPSPs ( ú200 ms in cases where inhibition is weak, e.g., Fig. 6B), which are APV and CPP sensitive, suggesting that the physiology of the NMDA receptors associated with the StF input is similar to that reported for mammalian neurons. The late component of the StF-evoked EPSP ( ú20 ms) is voltage dependent and therefore might be attributed entirely to the activation of NMDA receptors. However this late phase was often not completely blocked by APV or CPP. In addition, preliminary experiments have shown that intracellular injection of lidocaine N-ethyl bromide (QX314) greatly reduces the late component of single-shock StFevoked EPSPs, the slow depolarization that follows tetanic stimulation of StF and the depolarizing ramp during tetanic stimulation (Plant 1994). This suggests that voltage-dependent ion channels also might contribute to the late phase of the StF evoked EPSPs (see Hirsch and Gilbert 1991). Because QX-314 can block both Na / and Ca 2/ inward currents (Talbot and Sayer 1996), the channels contributing to the voltage-sensitive late phase still must be elucidated. Initial studies emphasized the requirement of depolarization for NMDA-receptor activation and gave rise to the idea that NMDA receptors are activated primarily during tetanic stimulation (Collingridge and Lester 1989). However, recent in vivo studies have suggested that NMDA receptors contribute to the response of cortical neurons to even weak sensory input (Armstrong-James et al. 1993; Fox et al. 1990; Kwon et al. 1992), suggesting that these receptors may operate at near RMPs. Biophysical analysis of cloned NMDA receptors also has demonstrated a wide variation in susceptibility to Mg 2/ blockade, and that some subunit combinations can pass appreciable current at less than 070 mV (Kutswada et al. 1992). Thus it appears that the properties of NMDA receptors in the StF pathway are not unusual when compared with those in mammalian sensory systems and that, as in the mammal, they contribute to normal sensory processing (reviewed in / 9k1d$$oc22 J277-7 Nelson and Sur 1992). The NR1 subunit of the NMDA receptor of A. leptorhynchus is highly homologous to the mammalian NR1 subunit (Bottai et al. 1995–1997); it therefore will be important to determine which specific combinations of NR1 splice variants and NR2 subunits determine the voltage threshold, time to peak, and time constant of decay of NMDA receptors within the VML. Our present picture of the StF evoked EPSP is summarized in Fig. 9. There are at least four components to this EPSP: a very early (1.5 ms) electrotonic synapse; an early ( Ç3 ms), voltage-insensitive AMPA-receptor component; an early ( Ç4–7 ms) voltage-sensitive NMDA-receptor component, which also contributes to the late ( ú10 ms) phase of the EPSP; and a late, voltage-sensitive component ( Ç50 ms after single pulse and °500 ms in duration after tetanic stimulation) dependent on NMDA receptors and perhaps on voltage-sensitive ion channels as well. The frequency and duration of our tetanic stimuli mimic natural firing patterns of StF (Bratton and Bastian 1990). FIG . 9. A: summary diagram of the relevant contributions of gap junction and different ionotropic excitatory amino acid receptors to the StFevoked EPSP. A small electrotonic EPSP (Gap) precedes a CNQX-sensitive, voltage-insensitive EPSP mediated by an AMPA receptor component. NMDA receptors contribute to both the early peak and the late phase of the EPSP; the voltage sensitivity of these components is indicated by the double arrows. B: summary of how the receptive field sizes of pyramidal and stellate cells, time delays of reciprocal ELL-Pd connections, and voltage sensitivity of StF-evoked EPSP might contribute to the hypothesized searchlight function of the StF feedback pathway. As the fish scans past an object, changes in the firing rate of electroreceptors first will drive ELL pyramidal cells with receptive fields on the left (a). The axons of these cells travel in the lateral lemniscus to terminate on stellate cells of the contralateral Pd. Stellate cells have larger receptive fields than ELL pyramidal cells; in this diagram, all the indicated pyramidal cells are supposed to project to the Pd stellate cell. The stellate cell is activated phasically and emits a burst of spikes; this activity reaches the ELL pyramidal cell with receptive field (RF) ‘‘b’’ (gray) with a delay due to slow conduction in the tractus stratum fibrosum and synaptic delay; these delays are matched to the scan rate of the fish. Feedback input therefore arrives at pyramidal cell with RF b at the same time as does electroreceptor input generated by the object entering its receptive field. 09-05-97 11:58:59 neupa LP-Neurophys EXCITATORY AMINO ACID RECEPTORS AT A FEEDBACK PATHWAY The various components of the StF-evoked EPSPs interact in a dynamic manner to stimulation that mimics natural activation (i.e., 100–200 Hz, 100 ms) (Bratton and Bastian 1990) of this feedback pathway. At least four mechanisms may be involved in the typical ramp-like response to tetanic stimulation: presynaptic facilitation may increase the amplitude of succeeding EPSPs; temporal summation of the late phase of the EPSP will generate a rising depolarization; the voltage sensitivity of peak and late phase of the EPSP increase the response to later stimuli in the train and thus contribute to the ramp-like responses; and IPSPs can attenuate the response to later EPSPs in the train (Berman and Maler, unpublished observations). It is clear that the response to StF stimulation can be regulated dynamically at many potential sites (see below) and that it will take more detailed physiological studies as well as modeling to understand transmission at this feedback synapse. Relating in vitro properties of Stf-evoked EPSPs to electroreception The ELL is connected reciprocally and topographically to the n. praeminentialis (Maler et al. 1982, 1991; Sas and Maler 1983); the connections are bilateral in both directions with contralateral projections dominating. Because connections are excitatory in both directions, this represents an example of positive feedback. Bratton and Bastian (1990) have recorded from stellate cells in Pd, the neurons that give rise to the StF (Sas and Maler 1983). Stellate cells respond, with high sensitivity and gain, to AM (AMs õ16 Hz) of the EOD (contralateral electroreceptors). A. leptorhynchus scans its environment with stereotyped movements, the velocity of which (10–15 cm/s) (Lannoo and Lannoo 1992) would be expected to generate AMs of õ10 Hz (Bastian 1981). As expected from these considerations, stellate cells respond vigorously and phasically to movement of objects over their receptive fields on the contralateral side of the fish’s body (Bratton and Bastian 1990). On the basis of these data, Bratton and Bastian hypothesized that stellate cells are optimized to detect small moving objects. Taking into account the excitatory reciprocal connectivity between ELL and Pd, they further hypothesized that the StF feedback pathway might act as a searchlight to enhance the response to salient features of the environment. A recent behavioral study in fact has shown that lesions of Pd do affect the electrodetection of objects (Green 1996). The original searchlight hypothesis was enunciated by Crick (1984) with regard to the reciprocal connections between mammalian thalamus and cortex. This theory contained three essential ingredients: reciprocal excitatory connectivity, a relatively diffuse parallel inhibitory feedback system that can keep the feedback excitation confined to a limited spatial domain, and a nonlinearity in the responsiveness of the lower order cells (thalamic relay neurons in Crick’s thesis) that amplifies stronger inputs. The identification of a diffuse inhibitory (GABAergic) feedback pathway from cells in medial Pd to ELL pyramidal cells suggests that the second of Crick’s criteria also is met by the direct feedback system to the ELL (Maler and Mugnaini 1993, 1994). As discussed below, the data presented in this paper suggest that the third criteria is met as well. / 9k1d$$oc22 J277-7 1879 We propose that the voltage dependence of StF synapses in VML causes them to behave as a nonlinear thresholding device. StF activation will only produce large, suprathreshold EPSPs when the pyramidal cell is depolarized by other inputs. Pyramidal cells can be depolarized by input from electroreceptors (Bastian 1981) or by indirect feedback input to the DML. The latter alternative is certainly plausible, however, other than its involvement in gain control (Bastian 1986b,c), little is known about the physiology of the indirect feedback pathway. We therefore hypothesize that when electroreceptor (primary afferents) and StF feedback inputs arrive concurrently at a pyramidal cell, the StF input is enhanced greatly and therefore is very effective at bringing that cell above spike threshold. From this viewpoint, it is the voltage dependence of the NMDA-receptor component of StF synapses that is critical for their function; a similar proposal has been made for the role of NMDA receptors associated with corticothalamic feedback fibers (to lateral geniculate nucleus) (Esguerra and Sur 1992). The spatial aspects of this theory are discussed below (Fig. 9). The hypothesis discussed above depends on nonlinear summation of EPSPS evoked by primary afferents and StF fibers; the nonlinearity is due to the voltage dependence of the StF-evoked EPSPs. Recent work suggests a second nonlinearity might be present in this system. Turner et al. (1994) found that the proximal apical dendrites of pyramidal cells conduct Na / spikes and that inward current associated with these spikes sourced back to the soma where they appeared as depolarizing afterpotentials that could generate the spike bursts seen in these cells in vitro (Turner et al. 1996). Gabbiani et al. (1996) recently have shown that ELL pyramidal cells in vivo can signal the occurrence of temporal electrosensory ‘‘features’’ by spike bursts. They also suggest that the feature extraction occurs in the proximal dendrites rather than the soma (Gabbiani et al. 1996). It is thus possible that StF-evoked EPSPs, in addition to directly triggering spikes, can enhance antidromic spike invasion of the proximal apical dendrite of ELL pyramidal cells and thus increase the probability of burst initiation. A recent study (Magee and Johnston 1997) has demonstrated in hippocampal pyramidal cells that dendritic depolarization due to EPSPs can in fact increase the amplitude of antidromic dendritic action potentials. In this case, the slow summating component of the StF-evoked EPSP might increase the electrodetectability of moving objects for hundreds of milliseconds by enabling primary afferent input to trigger spike bursts. Pyramidal cells and stellate cells have receptive field (RF) centers with diameters of 11 and 18 mm, respectively (Bastian 1981; Bratton and Bastian 1990); the greater size of the stellate cell’s receptive field is presumably due to convergence of pyramidal cell input (Maler et al. 1982). Because pyramidal and stellate cells are connected reciprocally (Maler et al. 1982), a stellate cell’s receptive field will extend Ç3.5 mm beyond that of its concentrically matched pyramidal cell. This suggests that an object moving past electroreceptors can cause Pd stellate cells to discharge and that these stellate cells can in turn excite pyramidal cells that have not yet been activated by that object (Fig. 9). The relative spread of the ELL pyramidal cell and Pd stellate cell reciprocal connections will determine how far ahead the descending input will prime ELL pyramidal cells the 09-05-97 11:58:59 neupa LP-Neurophys 1880 N. J. BERMAN, J. PLANT, R. W. TURNER, AND L. MALER receptive fields of which lie in the object’s future trajectory. Further details on the precision of this connectivity is required to accurately quantify the scope of the searchlight. Bratton and Bastian (1990) have noted that Pd stellate cells respond to electrosensory inputs with a long latency (11 vs. 5.2 ms for ELL pyramidal cells). The tSF projection from Pd to ELL is Ç3,500 mm long (estimated from the atlas of Maler et al. 1991). Our data suggests a delay of Ç17 ms for the Pd-ELL projection (conduction velocity of 0.31 m/s produces a conduction delay of 11 ms and a delay to EPSP peak of 6 ms). Data from Bastian suggest a shorter delay of 10 ms. Thus activation of ELL pyramidal cell with RF ‘‘a’’ (Fig. 9) will produce a feedback EPSP in the pyramidal cell with RF ‘‘b’’ after Ç20–28 ms. If the fish is scanning at 10 cm/s (Lannoo and Lannoo 1992), it will traverse Ç2–3 mm in this period, placing the object over RF b. The duration of the late EPSP, under this hypothesis, would determine the minimal scanning rate. Therefore, primary afferent input to a pyramidal cell may coincide with feedback input from stellate cells representing adjacent regions of skin; given the voltage dependence of the StFevoked EPSP discussed above, this will result in large feedback EPSPs and an enhancement of the response to the moving object. These considerations suggest that this feedback system will create a traveling beam (searchlight) of enhanced responsiveness in pyramidal cells, priming them to detect scanned objects. Correlative biophysical and systems studies of the StF feedback pathway may elucidate the cellular basis of an elementary form of spatially and temporally localized attention. We thank Dr. Rob Dunn for insightful discussion on the molecular biology of NMDA receptors and W. Ellis for technical support. This work was supported by Medical Research Council grants to L. Maler and R. Turner, who is a Medical Research Council and Alberta Heritage Foundation for Medical Research Scholar. J. Plant was supported by a National Science and Engineering Research Council Fellowship. Present addresses: J. Plant, Dept. of Psychology, University of Victoria, PO Box 3050, Victoria, British Columbia V8W 3P5; R. W. Turner, Dept. of Anatomy, University of Calgary, Calgary, Alberta T2N 1N4, Canada. Address for reprint requests: L. Maler, Dept. of Anatomy and Neurobiology, University of Ottawa, Ottawa, Ontario K1H 8M5, Canada. Received 7 April 1997; accepted in final form 18 June 1997. REFERENCES ARMSTRONG-JAMES, M., WELKER, E., AND CALLAHAN, C. A. The contribution of NMDA and non-NMDA receptors to fast and slow transmission of sensory information in the rat S1 barrel cortex. J. Neurosci. 13: 2149– 2160, 1993. BASTIAN, J. Electrolocation. II. The effects of moving objects and other electrical stimuli on the activities of two categories of posterior lateral line cells in Apteronotus albifrons. J. Comp. Physiol. [A] 144: 481– 494, 1981. BASTIAN, J. Vision and electroreception. Integration of sensory information in the optic tectum of the weakly electric fish Apteronotus leptorhynchus. J. Comp. Physiol. [A] 147: 287–297, 1982. BASTIAN, J. Electrolocation: Behaviour, anatomy and physiology. In: Electroreception, edited by T. H. Bullock and W. Heiligenberg, New York: Wiley, 1986a, p. 577–612. BASTIAN, J. Gain control in the electrosensory system mediated by descending inputs to the electrosensory lateral line lobe. J. Neurosci. 6: 553– 562, 1986b. BASTIAN, J. Gain control in the electrosensory system. A role for descending projections to the lateral electrosensory lateral line lobe. J. Comp. Physiol. [A] 158: 505–515, 1986c. / 9k1d$$oc22 J277-7 BASTIAN, J. The role of amino acid neurotransmitters in the descending control of electroreception. J. Comp. Physiol. [A] 172: 409–423, 1993. BASTIAN, J. AND BRATTON, B. Descending control of electroreception. I. Properties of nucleus praeeminentialis neurons projecting indirectly to the electrosensory lateral line lobe. J. Neurosci. 10: 1226–1240, 1990. BLAKE, J. F., BROWN, M. W., AND COLLINGRIDGE, G. L. CNQX blocks acidic amino acid induced depolarizations and synaptic components mediated by non-NMDA receptors in rat hippocampal slices. Neurosci. Lett. 89: 182–186, 1988. BODE-GREUL, K. M., SINGER, W., AND ALDENHOFF, J. B. A current source density analysis of field potentials in slices of visual cortex. Exp. Brain. Res. 69: 213–219, 1987. BOTTAI, B., ELLIS, B., MALER, L., AND DUNN, R. NMDA receptor 1 splice variants: differential cellular expression in the central nervous system of the electric fish Apteronotus leptorhynchus. Soc. Neurosci. Abstr. 22: 1996. BOTTAI, B., MALER, L., AND DUNN, R. Molecular characterization of NMDA receptors in the electric fish Apteronotus leptorhynchus. Soc. Neurosci. Abstr. 21: 187, 1995. BOTTAI, D., DUNN, R., ELLIS, W., AND MALER, L. NMDA R1 mRNA distribution in the central nervous system of the weakly electric fish Apteronotus leptorhynchus. J. Comp. Neurol. In press. BRATTON, B. AND BASTIAN, J. Descending control of electroreception. II. Properties of nucleus praeeminentialis neurons projecting directly to the electrosensory lateral line lobe. J. Neurosci. 10: 1241–1253, 1990. CARR, C. E. AND MALER, L. Electroreception in gymnotiform fish. Central anatomy and physiology. In: Electroreception, edited by T. H. Bullock and W. Heiligenberg, New York: Wiley, 1986, p. 319–373. COLLINGRIDGE, G. L., HERRON, C. E., AND LESTER, R. A. J. Synaptic activation of N-methyl-D-aspartate receptors in the Schaffer collateral-commissural pathway of the hippocampus. J. Physiol. (Lond.) 399: 283–300, 1988. COLLINGRIDGE, G. L. AND LESTER, R. A. J. Excitatory amino acid receptors in the vertebrate central nervous system. Pharmacol. Rev. 40: 145–210, 1989. CRICK, F. Function of the thalamic reticular complex. The searchlight hypothesis. Proc. Natl. Acad. Sci. USA, 81: 4586–5490, 1984. DREJER, J. AND HONORE, T. New quinoxalindiones show potent antagonism of quisqualate responses in cultured mouse cortical neurons. Neurosci. Lett. 87: 104–108, 1988. DZUBAY, J. A. AND JAHR, C. E. Kinetics of NMDA channel opening. J. Neurosci. 16: 4129–4134, 1996. EDMONDS, B., GIBB, A. J., AND COLQUHOUN, D. Mechanisms of activation of glutamate receptors and the time course of excitatory synaptic currents. Annu. Rev. Physiol. 57: 495–519, 1995. ESGUERRA, M., KWON, Y. H., AND SUR, M. Retino-geniculate EPSPs recorded intracellularly in the ferret lateral geniculate nucleus in vitro. Role of NMDA receptors. Vis. Neurosci. 8: 545–555, 1992. ESGUERRA, M. AND SUR, M. Corticogeniculate feedback gates retinogeniculate transmission by activating NMDA receptors. Soc. Neurosci. Abstr. 16: 159, 1992. FORSYTHE, I. D. AND WESTBROOK, G. L. Slow excitatory postsynaptic currents mediated by N-methyl-D-aspartate receptors on mouse cultured central neurons. J. Physiol. (Lond.) 396: 515–533, 1988. FOX, K., SATO, H., AND DAW, N. The effect of varying stimulus intensity on NMDA-receptor activity in cat visual cortex. J. Neurophysiol. 64: 1413–1428, 1990. GABBIANI, F., METZNER, W., WESSEL, R., AND KOCH, C. From stimulus encoding to feature extraction in weakly electric fish. Nature 384: 564– 567, 1996. GREEN, R. L. How lesioning the nucleus praeeminetialis affects electrolocation behaviour in the weakly electric fish, Apteronotus leptorhynchus. J. Comp. Physiol. [A] 179: 353–361, 1996. HEILIGENBERG, W. Neural Nets in Electric Fish, Cambridge, MA: MIT Press, 1991. HESTRIN, S., NICOLL, R. A., PERKEL, D. J., AND SAH, P. Analysis of excitatory synaptic action in pyramidal cells using whole-cell recording from rat hippocampal slices. J. Physiol. (Lond.) 422: 203–225, 1992. HIRSCH, J. AND GILBERT, C. Synaptic physiology of horizontal connections in the cat’s visual cortex. J. Neurosci. 11: 1800–1809, 1991. HOLLMANN, M., BOULTER, J., MARON, C., BEASLEY, J., PECHT, G., AND HEINEMANN, S. Zinc potentiates agonist-induced currents at certain splice variants of the NMDA receptor. Neuron 10: 943–954, 1993. KUTSWADA, T., KASHIWABUCHI, T., MORI, N. H., SAKIMURA, K., KUSHIYA, 09-05-97 11:58:59 neupa LP-Neurophys EXCITATORY AMINO ACID RECEPTORS AT A FEEDBACK PATHWAY E., ARAKI, K., MEGURO, H., MASAKI, H., KUMANISHI, T., ARAK AWA, M., AND MISHINI, M. Molecular diversity of the NMDA receptor channel. Nature 358: 36–41, 1992. KWON, Y., NELSON, K., TOTH, K., AND SUR, M. Effect of stimulus contrast and size on NMDA-receptor activity in cat lateral geniculate nucleus. J. Neurophysiol. 68: 182–196, 1992. LANNOO, M. J. AND LANNOO, S. J. Why do electric fish swim backwards? An hypothesis based on gymnotiform behavior, interpreted through sensory constraints. Environ. Biol. Fishes 36: 157–165, 1992. MAGEE, J. C. AND JOHNSTON, D. A synaptically controlled associative signal for Hebbian plasticity in hippocampal neurons. Science 275: 209–213, 1997. MALER, L. The posterior lateral line lobe of certain gymnotiform fish. Quantitative light microscopy. J. Comp. Neurol. 183: 323–363, 1979. MALER, L. The role of feedback pathways in the modulation of receptive fields: an example from the electrosensory system. In: Neural Mechanisms of Behaviour, edited by J. Erber, R. Menzel, H.-J. Pfluger, and D. Todt. Stuttgart, Germany: Thieme, 1989, p. 111–115. MALER, L., COLLINS, M., AND MATHIESON, W. B. The distribution of acetylcholinesterase and choline acetyl transferase in the cerebellum and posterior lateral line lobe of weakly electric fish (Gymnotidae). Brain Res. 226: 320–325, 1981a. MALER, L. AND MONAGHAN, D. The distribution of excitatory amino acid binding sites in the brain of an electric fish. Apteronotus leptorhynchus. J. Chem. Neuranat. 4: 39–61, 1991. MALER, L. AND MUGNAINI, E. Organization and function of feedback to the electrosensory lateral line lobe of gymnotiform fish, with emphasis on a searchlight mechanism. J. Comp. Physiol. [A] 173: 667–670, 1993. MALER, L. AND MUGNAINI, E. Correlating gamma-aminobutyric acidergic circuits and sensory function in the electrosensory lateral line lobe of a gymnotiform fish. J. Comp. Neurol. 345: 224–252, 1994. MALER, L., SAS, E., CARR, C., AND MATSUBARA, J. Efferent projections of the posterior lateral line lobe in a gymnotiform fish. J. Comp. Neurol. 21: 154–164, 1982. MALER, L., SAS, E., JOHNSTON, S., AND ELLIS, W. An atlas of the brain of the weakly electric fish. Apteronotus Leptorhynchus. J. Chem. Neuranat. 4: 1–38, 1991. MALER, L., SAS, E. K., AND ROGERS, J. The cytology of the posterior lateral line lobe of high frequency weakly electric fish ( Gymnotoidei): differentiation and synaptic specificity in a simple cortex. J. Comp. Neurol. 195: 87–139, 1981b. MATHIESON, W. B. AND MALER, L. Morphological and electrophysiological properties of a novel in vitro preparation: the electrosensory lateral line lobe brain slice. J. Comp. Physiol. [A] 163: 489–506, 1988. / 9k1d$$oc22 J277-7 1881 MONYER, H., SPRENGEL, R., SCHOEPFER, R., HERB, A., HIGUCHI, M., LOMELI, H., BURNASHEV, N., SAKMAN, B., AND SEEBURG, P. Heteromeric NMDa receptors: molecular and functional distinction of subtypes. Science 256: 1217–1221, 1992. NELSON, S. AND SUR, M. NMDA receptors in sensory processing. Curr. Opin. Neurobiol. 2: 484–488, 1992. PLANT, J. R. Fast NMDA Transmission and Non-Linear Synaptic Interactions in a Sensory Feedback Pathway (PhD thesis). Ottowa, Ontario: Univ. of Ottawa, 1994. RICHARDSON, T. L., TURNER, R. W., AND MILLER, J. J. Action potential discharge in hippocampal CA1 pyramidal neurons: a current source density analysis. J. Neurophysiol. 58: 981–986, 1987. ROSE, G. J. AND HELIGENBERG, W. Neural coding of difference frequencies in the midbrain of the electric fish Eigenmannia. Reading the sense of rotation in an amplitude-phase map. J. Comp. Physiol. [A] 158: 613– 624, 1988. SAS, E. AND MALER, L. The nucleus praeeminentialis: a golgi study of a feedback center in the electrosensory system of gymnotid fish. J. Comp. Neurol. 221: 127–144, 1983. SAS, E. AND MALER, L. The organization of afferent input to the caudal lobe of the cerebellum of the gymnotid fish Apteronotus leptorhynchus. Anat. Embryol. 177: 55–79, 1987. SHIROK AWA, T., NISHIGORI, A., KIMURA, F., AND TSUMOTO, T. Actions of excitatory amino acid antagonists on synaptic potentials of layer II/III neurons of the cat’s visual cortex. Exp. Brain. Res. 78: 489–500, 1989. SHUMWAY, C. Multiple electrosensory maps in the medulla of weakly electric Gymnotiform fish. I. Physiological differences. J. Neurosci. 9: 4388– 4399, 1989. TALBOT, M. J. AND SAYER, R. J. Intracellular QX-314 inhibits calcium currents in hippocampal CA1 pyramidal neurons. J. Neurophysiol. 76: 2120– 2124, 1996. TAUBE, J. S. AND SCHWARTKROIN, P. A. Mechanisms of long-term potentiation: a current source density analysis. J. Neurosci. 8: 1645–1655, 1988. TURNER, R. W., MALER, L., DEERINCK, T., LEVINSON, S. R., AND ELLISMAN, M. H. TTX-sensitive dendritic sodium channels underlie oscillatory discharge in a vertebrate sensory neuron. J. Neurosci. 14: 6453–6471, 1994. TURNER, R. W., PLANT, J. R., AND MALER, L. Oscillatory and burst discharge across electrosensory topographic maps. J. Neurophysiol. 76: 2364–2382, 1996. WANG, D. AND MALER, L. The immunocytochemical localization of glutamate in the electrosensory system of the gymnotiform fish, Apteronotus Leptorhynchus. Brain Res. 653: 215–222, 1994. ZUCKER, R. S. Short-term synaptic plasticity. Annu. Rev. Neurosci 12: 13– 31, 1989. 09-05-97 11:58:59 neupa LP-Neurophys