Olfactory bulb networks revealed by lateral olfactory tract stimulation in the in vitro isolated guinea-pig brain

Neuroscience 142 (2006) 567–577

OLFACTORY BULB NETWORKS REVEALED BY LATERAL OLFACTORY TRACT STIMULATION IN THE IN VITRO ISOLATED GUINEA-PIG BRAIN L. UVA,a B. W. STROWBRIDGEb AND M. DE CURTISa*

1996). Intracellular recordings in salamander OBs demonstrate that the sensory-evoked excitation of mitral cells is interrupted by large inhibitory potentials. The resulting irregular temporal pattern of action potentials presumably codes for qualities of the sensory input. The inhibitory local circuits that generate these IPSPs are dominated by granule cells, small GABAergic interneurons that lack axons and form dendrodendritic synapses with the lateral dendrites of glutamatergic mitral cells (Price and Powell, 1970; Shepherd et al., 2004). Reciprocal dendro-dendritic synapses represent the morphological entities that control most recurrent and lateral inhibition onto mitral cells. They were first inferred from analyses of the sequence of field potentials activated antidromically following stimulation of the lateral olfactory tract (Phillips et al., 1963; Nakashima et al., 1978; Mori, 1987) and by electron microscopic examination of the OB (Price and Powell, 1970; Wellis and Kauer, 1994). While local circuitry in the OB is well understood (Shepherd, 1972; Shepherd et al., 2004; Shipley and Ennis, 1996), the functional organization of the synaptic pathways that reciprocally interconnect piriform cortex and the OB have not been extensively explored. The isolated guinea-pig brain preparation, commonly used to study field activity in limbic structures (Uva and de Curtis, 2005; Paré et al., 1992; de Curtis et al., 1991, 1998; Biella et al., 2002; Biella and de Curtis, 1995, 2000), is well suited to define the components of field potentials generated within the OB and synaptic activities that may result from long-range interactions with cortical olfactory structures. Since the entire brain is exposed, there is excellent access to position stimulating electrodes on the lateral olfactory tract (LOT) and multiple recording electrodes in the OB and piriform cortex. Pharmacological agents, including receptor antagonists that would be lethal if applied systemically in whole-animal preparations, can be applied intra-arterially and readily affect the entire brain. Utilizing this method, we have reproduced in the in vitro isolated guinea-pig brain the distribution of field potentials observed in the OB in vivo (Rall and Shepherd, 1968; Phillips et al., 1963; Neville and Haberly, 2003; Nakashima et al., 1978) and in in vitro slice preparations (Aroniadou-Anderjaska et al., 1999). Using current source density (CSD) analysis of multi-electrode array laminar profiles, we now report three distinct current sinks following LOT stimulation corresponding to (1) antidromic mitral cell body activation, (2) mitral cell dendrite activation and (3) granule cell synaptic excitation. We also observed following both LOT and direct piriform cortex stimulation a fourth component with a

a

Department of Experimental Neurophysiology, Istituto Nazionale Neurologico, via Celoria 11, 20133 Milano, Italy b Department of Neurosciences, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, OH 44106, USA

Abstract—Olfactory information processing is mediated by synaptic connections between the olfactory bulbs (OBs) and piriform–limbic cortices. Limited accessibility using common in vivo and in vitro preparations has hindered previous attempts to define these synaptic interactions. We utilized the isolated guinea-pig brain preparation to overcome these experimental limitations. Previous studies demonstrated extensive functional preservation in this preparation maintained in vitro by arterial perfusion. Field potential laminar profiles were performed with multichannel probes in the OB following stimulation of both the lateral olfactory tract (LOT) and the anterior piriform cortex (APC). Current-source density analysis was carried out on laminar profiles to reconstruct current sinks/sources associated with intrinsic synaptic activities. LOT stimulation induced sequentially i) an antidromic population spike (at 2.66ⴞ0.39 ms) located in the mitral cell layer that was resistant to 100 Hz high-frequency stimulation (HFS) and 6-cyano7-nitroquinoxaline-2,3-dione (CNQX) (10 ␮M), ii) a component located in the external plexiform layer at 3.85ⴞ0.63 ms that was unaffected by HFS, iii) a large amplitude potential (peak amplitude at 5.84ⴞ0.58 ms) generated in the external plexiform layer, abolished by HFS and CNQX, but not by bicuculline (50 ␮M), iv) a late response (onset at 20.00ⴞ2.94 ms) abolished by CNQX and enhanced by bicuculline. Stimulation of the APC also induced a late potential abolished by HFS and CNQX. Both APC-evoked and late LOT-evoked responses were abolished by a transverse cut to separate OB from APC. These results demonstrate in an isolated mammalian brain preparation the presence of reciprocal synaptic interactions between the OB and piriform cortical structures. © 2006 IBRO. Published by Elsevier Ltd. All rights reserved. Key words: olfactory bulb, isolated guinea-pig brain, current source density.

Synaptic circuits in the olfactory bulb (OB) play a critical role in patterning the discharge of mitral cells, the principal neurons in OB, following sensory stimulation (Kauer, 1991; Shepherd, 1972; Shepherd et al., 2004; Shipley and Ennis, *Corresponding author. Tel: ⫹39-02-23942280; fax: ⫹39-02-70600775. E-mail address: [email protected] (M. de Curtis). Abbreviations: APC, anterior piriform cortex; APV, D-2-amino-5-phosphonovaleric acid; BMI, bicuculline methiodide; CNQX, 6-cyano-7nitroquinoxaline-2,3-dione; CSD, current source density; EPL, external plexiform layer; GCL, granule cell layer; HFS, high-frequency stimulation; LOT, lateral olfactory tract; MCL, mitral cell layer; OB, olfactory bulb.

0306-4522/06$30.00⫹0.00 © 2006 IBRO. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.neuroscience.2006.06.047

567

568

L. Uva et al. / Neuroscience 142 (2006) 567–577

long delay maybe associated with centrifugal activation of granule cells by projection neurons in olfactory cortex. These results have been presented previously in abstract form (Uva et al., 2005).

EXPERIMENTAL PROCEDURES Guinea-pig brains were isolated and maintained in vitro using a variation on a standard procedure developed for examining electrical activity in limbic cortical areas (Muhlethaler et al., 1993; de Curtis et al., 1998; Singer et al., 1998). Female Hartley guineapigs (150 –200 g, Charles River, Calco, Italy) were anesthetized with sodium thiopental (125 mg/kg i.p, Farmotal, Pharmacia, Italy) and were transcardially perfused with a cold (4°–5 °C) and oxygenated (95% O2, 5% CO2) saline solution that included dextran as a plasma expander (composition: 126 mM NaCl, 3 mM KCl, 1.2 mM KH2PO4, 1.3 mM MgSO4, 2.4 mM CaCl2, 26 mM NaHCO3, 15 mM glucose, 2.1 mM Hepes and 3% dextran M. W. 70,000, pH⫽7.1). The brain was rapidly removed and transferred into the recording chamber. A cannula made from heat-pulled PE60 tubing was inserted in the basilar artery to perfuse the brain with the same saline solution (15 °C, pH⫽7.3). The cannula tip was placed at the junction of the basilar artery and the circle of Willis. Small knots made from filaments isolated from silk suture thread were used to tie the basilar artery around the plastic cannula and to close major arteries severed when the brain was removed from the skull. OB field potentials described in this report were reliably observed when the perfusion to anterior brain structures was increased relative to the protocol used previously to study activity in temporal lobe structures. Perfusion of the OBs was obtained by a flow rate into the basilar artery of 7.5 ml/min (from 5.5 ml/ml used in previous studies) and in some experiments by closing the circle of Willis on the non-experimental hemisphere between posterior cerebral and basilar arteries (see Fig. 1B). Recordings were restricted to the olfactory region on the perfused hemisphere. In this protocol, two other vessels (limbic and posterior cerebral arteries) were closed on the experimental hemisphere to increase perfusion to the OB. The temperature of the perfusate was slowly increased (0.2 °C/min) to a final temperature of 32 °C. Recordings began when the brain temperature stabilized (1.5 h after the initial cardiac perfusion) and typically lasted for 4 – 6 h. The viability of the preparation was assessed periodically by recording field responses in the anterior piriform cortex (APC) following single shocks to the LOT (Fig. 1A). These responses were biphasic in healthy preparations (see Fig. 1C), reflecting mono- and di-synaptic excitatory responses within piriform cortex (Rodriguez and Haberly, 1989; Ketchum and Haberly, 1993; Biella and de Curtis, 1995), and had a peak amplitude ⬎1 mV. We observed similar LOT-evoked OB and APC field potentials in experiments in which only one hemisphere was perfused and experiments in which the circle of Willis was intact. These experimental protocols were reviewed and approved by the Committees on Animal Care and Use of the Istituto Nazionale Neurologico and Case Western Reserve University and by Ethics Committee of the Istituto Nazionale Neurologico. All experiments conformed to international guidelines on the ethical use of animals. Efforts were made to minimize the number of animals used and their suffering. Synaptic responses were evoked using bipolar electrodes connected to a stimulus isolation unit driven by a Grass-Telefactor S88 pulse generator (Warwick, RI, USA) and were positioned on the LOT and in some experiments also on the APC (Fig. 1A). Extracellular evoked potentials were recorded in the OB with 16-channel linear silicon probes (50 ␮m contact separation, kindly provided by the Center for Neural Communication Technology, University of Michigan, Ann Arbor, MI, USA; Fig. 1A). Recordings in the APC were also performed with glass pipettes filled with 1 M NaCl (10 ␮m tip diameter, 5–10 M⍀ input resistance). The record-

ing electrodes were inserted perpendicular to the lamination of the OB and APC with a ventral to dorsal approach, under visual control using a stereoscopic microscope. Electrophysiological signals were amplified with a multichannel amplifier (Biomedical Engineering, Thornwood, NY, USA) and were digitized via an AT-MIO-63E3 National Board (National Instruments, Milano, Italy) at a sampling rate of 3–5 kHz. Custom software (ELPHO©) written in Labview (National Instruments) by Vadym Gnatkovsky was used to acquire and analyze the electrophysiological data. CSD analysis of the field potential profiles was performed using a 200 ␮m separation grid, to reveal the local OB and APC activities evoked by stimulation of the LOT (Mitzdorf, 1985; Ketchum and Haberly, 1993; Biella and de Curtis, 1995). Laminar profiles were obtained by averaging 10 –12 evoked responses. Pharmacological agents were applied by switching perfusion reservoirs. The NMDA receptor blocker, D-2-amino-5-phosphonovaleric acid (APV, 100 ␮M; RBI, Milano, Italy), the non-NMDA receptor blocker, 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, 10 ␮M, RBI) and the GABAA receptor antagonist, bicuculline methiodide (BMI; 50 ␮M, Sigma, Milano, Italy), were dissolved in the perfusate and were arterially applied for 10 min (CNQX and APV) or 3 min (BMI). As shown previously, these concentrations of glutamate receptor antagonists are sufficient to block evoked field EPSPs in entorhinal (Dickson et al., 2000) and piriform (de Curtis et al., 1999) cortices. Electrolytic lesions were made by passing a 30 ␮A current for 30 s between the two contacts on the silicon probe at the conclusion of most experiments. Brains were fixed for at least 1 week in paraformaldehyde (4%) and parasagittal sections (75 ␮m thick) were cut at the vibratome and stained with Thionine to reveal the location of the multielectrode probe. To identify sub-components of the OB response, cuts of the LOT and transverse cut of the APC were performed using microscissors. A thin plastic spacer foil was positioned in the cut lesion to prevent passive spread of activity.

RESULTS Experiments were performed on 17 in vitro isolated guineapig brains. In all preparations studied, low intensity stimulation of the LOT (4 –5 V; 200 ␮s stimulus duration; 0.1– 0.3 Hz stim rate) evoked a single positive field potential in APC (lower traces in Fig. 1C1, 5 V). At higher stimulus intensities (⬎5.5 V), LOT stimulation evoked a characteristic biphasic positive potential previously reported in piriform cortex (Rodriguez and Haberly, 1989; Ketchum and Haberly, 1993; Biella and de Curtis, 1995). Responses recorded simultaneously in the OB revealed a stereotyped sequence of short-latency responses evoked by LOT stimulation described below (see also Nakashima et al., 1978; Stripling and Patneau, 1999). Unlike APC potentials, the OB response in the granule cell layer (GCL) was preceded by a fast negative-going spike (Fig. 1C, top traces; see also Fig. 1D). Field responses in the OB are graded (see superimposed responses to 5 and 6 V stimuli in Fig. 1C2) and do not contain an obvious potential that correlates with the second positive-going peak recorded in APC at higher stimulus intensities. This result suggests that field responses in the OB are likely generated locally and do not simply reflect distant activity neighboring cortex (i.e. “far field” responses). We analyzed laminar profiles recorded with 16-channel silicon probes (average of 10 –12 evoked responses) to define the components in the OB field response following LOT stimulation. As shown in Fig. 1D, LOT stimulation

L. Uva et al. / Neuroscience 142 (2006) 567–577

569

Fig. 1. Field potentials evoked by LOT stimulation in the OB of the isolated guinea-pig brain preparation. (A) Diagram of the ventral surface of the isolated guinea-pig brain. Outlined box shows the region depicted in stereomicroscope image on right, showing a 16-channel recording silicon probe in the OB and two bipolar stimulating electrodes positioned in the APC and on the LOT. Compass inset indicates rostral– caudal and lateral–medial axes. (B) Diagram of the perfusion pattern (black filling) used in most experiments. The limbic and posterior cerebral arteries were closed on the experimental hemisphere, to enhance perfusion to the OB. (C1) Simultaneously recorded field responses in the OB and the APC to increasing intensity stimulation of the ipsilateral LOT (0.1 Hz), as indicated above each pair of traces. (C2) Superposition of amplitude-normalized responses to 5 and 6 V stimulations. (D1) OB responses to LOT stimulation recorded using a 16-channel linear microelectrode array. Three components (a at 1.9 ms onset latency, b at 3.1 ms and c at 4.2 ms) are reliably isolated using this protocol, though all three responses are not evident on all channel recordings. (D2) Responses from the same 16-channel silicon probe in D1 shown at a slower time base. A fourth component (d, onset at 20 ms) can be resolved on most channels.

evoked a short latency, fast potential positive in the surface and negative in depth, termed a (onset delay 1.88⫾0.37ms; time-to-peak 2.66⫾0.39 ms; n⫽17 experiments). Component a was followed by a larger wave that represents the summation of multiple events. Component b appeared as a hump on the rising phase of this large wave, negative at surface and positive at depth (onset 3.08⫾0.44 ms; peak amplitude 3.85⫾0.63 ms). Component c corresponded to

the peak of the rising phase of the large wave, negative in the surface and positive in depth (onset 4.22⫾0.36 ms; peak amplitude 5.84⫾0.58 ms). In a minority of experiments we also recorded a fast inflection on the decaying phase following c, but the basis of this response was not addressed in this study. Finally, a slow wave (d), positive in surface and negative in depth, characterized by a delayed onset (20.00⫾2.94 ms) and a slow decay time (circa 100 ms)

570

L. Uva et al. / Neuroscience 142 (2006) 567–577

was observed in eight experiments with recording electrode contacts in the GCL (Fig. 1D2). Next we used CSD analysis to define the position of the locally generated events with corresponding histological lesions that marked the position of the recording probe (Fig. 2). Component a corresponded with a current sink in the mitral cell layer (MCL) and a passive source in the

external plexiform layer (EPL). Component b coincided with a sink diffusely distributed in the EPL coupled with a source in MCL/GCL. Component c corresponded to a complex sink located in the EPL and a source in the EPL/MCL (n⫽13). In addition, LOT stimulation evoked a fourth component (termed d) with a long onset latency (20.00⫾2.94 ms) that was readily recorded in deep layers of the OB

Fig. 2. Laminar distribution of current sinks a, b and c in the OB following LOT stimulation. Superimposed responses to ipsilateral LOT stimulation in the OB recorded using a 16-channel linear microelectrode array (50 ␮m electrode spacing) are shown in the top panel on the left. Laminar profiles were formed by averaging 10 evoked responses. Components a, b and c are marked by the vertical lines. The asterisk marks the stimulus artifact. The CSD profile obtained from the traces is shown in the middle panel. Current sinks are indicated by upward deflections. The contour plot of the CSD profile is shown in the lower panel. Sinks and sources are illustrated by continuous and dotted lines, respectively. Contour line intervals⫽0.0025 mV/mm2. Reconstruction (right) of the electrode track following electrolytic lesions at three different levels showed that the top trace was superficial to the glomerular layer and the bottom trace was in the GCL. The location of the recording sites (3–11) utilized to build the CSD contour plot is indicated by the arrows. Glom, glomerular layer.

L. Uva et al. / Neuroscience 142 (2006) 567–577

Fig. 3. Laminar distribution of current sink d in the OB following LOT stimulation. Superimposed field potential responses to LOT stimulation (0.1 Hz) recorded from a 16-channel microelectrode array that spans from the superficial GCL to the EPL. Laminar profiles were formed by averaging 12 evoked responses. The corresponding CSD contour plot shown below demonstrates the current sink for the late component d in the MCL–GCL. Components a and c also are shown in the CSD plot. Sinks and sources are illustrated by continuous and dotted lines, respectively. Contour line intervals⫽0.01 mV/mm2.

(granule and mitral cell body layers). In all experiments, the current sink for component d included both the superficial GCL and MCL (Figs. 2 and 3). The current sink associated with d was always smaller in amplitude than largest response in EPL (component c) but was longer in duration. We used high-frequency (100 Hz) LOT stimulation to distinguish between antidromic and synaptically-mediated field responses. As shown in Fig. 4, both components a and b were maintained during high-frequency stimuli and were present at the end of prolonged (1–5 s) stimulus trains. Similar results were observed in 10 experiments. By contrast, component c was attenuated on the second response and was abolished after four to five stimuli. These results are consistent with the hypothesis that component a represents the mitral cell antidromic spike and suggest that component b reflects dendritic activation of mitral cells triggered by the antidromically invading spike. The abolition of component c by repetitive stimulation (observed in all experiments) suggests that this response reflects synaptic activation of bulbar neurons. To define the synaptic circuits that mediate component c, the largest of the three fast responses evoked by LOT stimulation, we performed intra-arterial perfusion of ionotropic glutamate and GABAA receptor antagonists. As expected, arterial perfusion with the non-NMDA receptor antagonist, CNQX (10 ␮M; n⫽5 experiments), or with both CNQX (10 ␮M) and APV (100 ␮M; n⫽3) greatly reduced the amplitude of component c (to 41.1⫾7.1% of control in CNQX, P⬍0.001, and 49.0⫾21.4% of control in CNQX⫹ APV, P⬍0.05) without affecting components a (87.3⫾6.1% in CNQX) and b (90.6⫾5.8% in CNQX). The late compo-

571

nent d was abolished by arterial application of CNQX (Fig. 5A and B) and by high-frequency stimulation (HFS) (data not shown). The effects of these receptor antagonists on the field responses in the OB were completely reversed upon washout (Fig. 5A and B). The failure of glutamate receptor antagonists to block the initial two components is consistent with the other lines of evidence presented in this study, suggesting that these components are intrinsic and are triggered directly by the antidromic spike. Arterial application of the GABAA receptor antagonist BMI at a concentration high enough to induce spontaneous epileptic discharges in piriform cortex and the OB (see Fig. 5C) did not affect the amplitude of component c (85.1⫾7.2% of control; not significantly different from control; n⫽7 experiments; Fig. 5C). Blockade of GABAA receptors strongly affected the kinetics of the field potential response after the peak of component c, increasing the amplitude of the negative-going late response d to 217.6⫾41.2% of control (P⬍0.05; n⫽5 experiments). This effect was reversible upon washout of BMI. These results, summarized in Fig. 5E, suggest that the rising phase and peak of component c and d reflect excitatory synaptic activation of processes in EPL and GCL, respectively. While the rising phase of component c is not altered by GABAergic neurotransmission, GABAergic synaptic potentials strongly modulate decay of component c and the amplitude of component d. We next used APC stimulation and focal lesions to define the synaptic pathways responsible for components c and d. As shown in Fig. 6A, APC stimulation effectively triggered a late response d that had similar or greater magnitude than LOT stimulation (8 of 10 experiments), with little, if any, short-latency response a. We were able to isolate current sinks corresponding to component c in some experiments using multielectrode arrays (but at approximately 2 ms longer latency than LOT-evoked responses) in 5 of 10 experiments. The lack of correlation between the amplitude of the short-latency MCL sink a (presumed mitral cell antidromic population spike) and the late component d argues against an intrinsic origin for the late response d in mitral cells. Finally, we tested whether cortical activation could trigger the late response d when the direct antidromic pathway to mitral cells was eliminated by focal lesions of the LOT. In these experiments we initially verified all four components evoked by LOT stimulation (Fig. 6B, upper panel). We next transected the LOT using micro-scissors and placed a thin plastic spacer across the lesion to prevent passive spread of activity through the cut. LOT stimulation caudal to the cut evoked a large late component but did not evoke the initial a/b/c complex (second trace from top in Fig. 6B). A similar potential was evoked by direct stimulation of APC (Fig. 6B, middle trace) in six of eight experiments with focal LOT lesions, suggesting that the late component do not require synchronous antidromic activation of mitral cells, and may be due to a cortical input generated caudal to the LOT cut. This was confirmed by performing a second transverse cut lesion through APC; this dual lesion eliminated all evoked responses (Fig. 6B, fourth panel from top; n⫽3), suggesting that the delayed potential illustrated in

572

L. Uva et al. / Neuroscience 142 (2006) 567–577

Fig. 4. Separation of intrinsic and synaptic components using high-frequency stimulation. (A) Responses to a 1.5 s train of LOT shocks (12 V) at 100 Hz, recorded with a 16-channel linear electrode array positioned in the EPL, MCL and GCL of the OB. Only the first six and the last five responses are shown. Responses to the initial shock consist of both fast components (a and b) as well as potential c, which is progressively abolished on subsequent shocks. The large positive wave at the onset of each response marked by the asterisks is the stimulus artifact. (B) CSD contour plots illustrating the current density distribution of the sinks associated to components a, b and c, evoked by the first four LOT stimuli (left panel) and by a stimulus near the end of the train (right panel). Stimulus artifacts are blanked in B. Contour line intervals were 0.03 mV/mm2 and 0.01 mV/mm2, respectively, for left and right panels.

the second and third traces in Fig. 6B are mediated by polysynaptic projections fibers to the OB that travel in the APC. At the conclusion of each experiment we verified that the initial response pattern could still be evoked by stimulation of the LOT proximal to the first lesion (Fig. 6B, lower panel). The partial recovery of component d (lower trace of Fig. 6B) may reflect the smaller volume of APC activated by the LOT stimulus. After LOT and APC lesion, indeed, only a small rostral part of APC was activated. Similar results were observed in two other experiments in which we generated the same sequence of focal lesions.

DISCUSSION The present study demonstrates that the isolated guineapig brain is well suited to investigate synaptic interactions and network circuits in olfactory cortices. While the existence of reciprocal dendro-dendritic microcircuits in the OB

is now widely accepted (Shepherd, 1972; Shepherd et al., 2004), relatively few subsequent studies have attempted to define the antidromic and related synaptic field potentials directly. In the present study we confirmed the basic findings of Rall and Shepherd (1968) and also provide functional evidence for reciprocal interactions between OB and surrounding cortical areas. Our study is the first attempt to apply modern pharmacology and simultaneous multi-channel recordings to define the pathways responsible for the primary field potentials in the OB, using an intact brain preparation. We used LOT stimulation to analyze OB–APC interactions in the three-dimensionally preserved olfactory region. The simplified network activated by LOT stimulation is summarized in the diagram illustrated in Fig. 7. LOT stimuli induced an early component a in the OB, mediated by the antidromic activation of neurons located in the MCL, as

L. Uva et al. / Neuroscience 142 (2006) 567–577

573

Fig. 5. Pharmacology of excitatory and inhibitory synaptic components of the OB field response. (A1) Effect of intra-arterial perfusion of ionotropic glutamate receptor antagonists (CNQX, 10 ␮M and APV 100 ␮M) on the LOT-evoked field response recorded in the EPL. Co-application of APV and CNQX reversibly blocked the late component c, while not affecting the early components a and b. Traces are averages of four to five consecutive responses. (A2) Enlargement and superposition of the three traces. Similar results were obtained when only CNQX was perfused (as shown in B1 and B2). (C1) Intra-arterial perfusion of BMI (50 ␮M) did not affect the early components a and b and only slightly reduced the peak amplitude of component c. Blockade of GABAA receptors with BMI reversibly enhanced the negative-going response d. Averages of four to five consecutive responses. (C2) Superposition of the three traces shown in B1. (D) Long-duration field potential recordings showing the presence of interictal discharges in OB and APC before (upper traces) and after exposure to BMI (lower traces). (E) Summary of the experiments using CNQX (n⫽7) and BMI (n⫽9).

demonstrated by CSD profile analysis. Response a, indeed, showed a very short time delay (⬍2 ms) from stimulus artifact, was not affected by drugs that block excitatory

and inhibitory synapses and resisted HFS at 100 Hz. A second fast component (b) located in EPL followed the antidromic spike in MCL. Component b also persisted

574

L. Uva et al. / Neuroscience 142 (2006) 567–577

Fig. 6. (A) Comparison of the field responses recorded in OB to stimulation of the LOT (upper trace) and the APC (lower trace). (B) OB responses to LOT stimulation in another experiment in which both the LOT and APC were sequentially lesioned. Electrodes are illustrated on diagrams on the left of each panel. Delayed potentials were evoked in OB when stimulation was performed in LOT and APC caudal to the LOT cut (arrows). Lesions are represented by black bars. See text for details. Please, note that component d is not completely restored in the lower trace.

during HFS and was not abolished by synaptic blockers. The laminar distribution and depth reversal of components a and b were similar to those described in vivo in the rabbit OB (Phillips et al., 1963; Rall and Shepherd, 1968), in the rat OB in vivo (Nakashima et al., 1978; Patneau and Stripling, 1992; Stripling and Patneau, 1999) and in slices (Aroniadou-Anderjaska et al., 1999). We attribute response b to the back-propagation of the antidromic spike to mitral cell dendrites. Because of its small amplitude and possibly

because of its extended spatial distribution in EPL, component b was not detected in all experiments, but could be reliably identified after prolonged high-frequency tetanic stimulation. A large amplitude potential (component c) followed the antidromic MCL responses (Fig. 7, middle panel). Previous studies proposed that such surface-negative/depth positive potential is mediated by the activation of excitatory synaptic potentials in the dendrites of OB granule cells (Rall and Shepherd, 1968; Nakashima et al., 1978; Martinez and Freeman, 1984; Mori, 1987; Aroniadou-Anderjaska et al., 1999). In our experiments the excitation of granule cell dendrites is likely due to glutamate released at dendrodendritic synapses as the antidromically-evoked action potential propagates along the lateral dendrites. In many experiments, component c was associated with more than one sink located at different depths in EPL, suggesting that multiple populations of interneurons may be synaptically activated following LOT stimulation. The demonstration that component c and the associated sinks were abolished both during HFS and during pharmacological blockade of excitatory synaptic transmission strongly suggests that this potential is dependent on synaptic excitation in the EPL. Our results also are consistent with the principal findings of Rall and Shepherd (1968) that component c (roughly equivalent to their period III) reflects excitation of dendritic processes of granule cells. A major consequence of this granule cell excitation postulated by Rall and Shepherd (1968) is strong mitral cell inhibition, a process that we show begins during component c using intra-arterial perfusion of GABAA receptor antagonists. BMI strongly affected the field potential response following the peak of component c (Fig. 5B). The changes induced by BMI could reflect either the presence of synaptic inhibition or the sculpting of excitatory potentials by polysynaptic activity. Unfortunately, CSD analysis of laminar profiles in BMI proved difficult because of the large variability in the LOT-evoked responses that were strongly affected by spontaneous interictal spiking, and possibly also because of the non-laminar distribution of the underlying sink/sources. The mitral cell– granule cell complex potential was followed by a late component, d, that showed an onset at approximately 20 ms, lasted about 100 ms and may be due to several mechanisms. Our experiments suggest that a major contribution to this response is mediated by an excitatory synaptic input into granule cells via associative fiber terminals that originate either in the piriform cortex or in other areas caudal to the APC (lower panel in Fig. 7). Component d is abolished both by glutamate receptor blockade. A delayed potential that showed depth reversal in OB similar to component d was observed when LOT and APC were stimulated in positions caudal to a LOT cut. This potential was abolished by surgical cut between APC and OB. However, we cannot exclude the possibility that response d is due to the excitation of an intermediate cortical region located rostral to the APC. Our findings confirmed that associative projections that originate caudal to the OB terminate in MCL–GCL (Nakashima et

L. Uva et al. / Neuroscience 142 (2006) 567–577

575

al., 1978), where APC projection fibers are known to target granule cell dendrites (Luskin and Price, 1983; Shepherd et al., 2004). A previous study conducted in vivo on the rat (Neville and Haberly, 2003) reached similar conclusions on the delayed OB component generated in response to direct APC stimulation. This study was focused on the generation of odor-induced fast oscillatory activities and demonstrated the dependence of oscillations in the beta range on a feedback loop originating from the piriform cortex. The amplitude of component d was enhanced by application of BMI, which induces disinhibition of the olfactory cortex and promotes spontaneous interictal spikes (de Curtis et al., 1994; Librizzi and de Curtis, 2003). It is likely that the increase in component d during BMI application is due to increased associative input from the disinhibited APC. The OB plays a critical role in processing olfactory sensory information. This brain structure converts slow monophasic excitatory input from receptor neurons into reproducible temporal patterns of action potentials that are conveyed by through the LOT to the piriform cortex and the olfactory tubercle (Dennis and Kerr 1975, 1968; Shepherd et al., 2004; Shipley and Ennis, 1996). Intracellular recordings from mitral cells (Hamilton and Kauer, 1988) revealed that inhibition plays a central role in patterning these action potential trains. Presumably, most of the inhibitory potentials evoked in mitral cells by sensory stimulation arise from GABA released by local bulbar interneurons (Jahr and Nicoll, 1980; Isaacson and Strowbridge, 1998; Schoppa et al., 1998); extrinsic GABAergic fibers appear to innervate primarily neurons with processes in the GCL (Kunze et al., 1992). However, bulbar interneurons may be activated through multiple pathways, either locally within the bulb through dendrodendritic synapses made between granule cell dendrites and the lateral dendrites of mitral cells (Rall et al., 1966) or through “backward” connections made by axons of deep layer pyramidal cells in the piriform cortex (Nakashima et al., 1978; Neville and Haberly, 2003). Utilizing the guinea-pig isolated brain preparation, we were able to demonstrate directly the current sinks associated with these two pathways (components c and d, respectively) and show different laminar organization. Finally, we were able to demonstrate that both excitatory pathways rely on AMPA receptors and thus are likely to be glutamatergic in nature. The large amplitude and robustness of the late current sink d in our study suggests that inter-regional pathways may account for a large fraction of the inhibition onto mitral cells following sensory stimulation. It is also possible that both local and inter-regional pathways work synergistically. Halabisky and Strowbridge (2003) showed that near-coincident electrical stimulation of glutamatergic fibers in the GCL potentiated recurrent dendrodendritic inhibition onto mitral cells. Thus, piriform cortical neurons may play a dual Fig. 7. Diagram of the circuit hypothesized to generate OB field potentials (dotted lines). The first two responses recorded in the OB probably reflect antidromic population spikes in mitral cells (a) and the back-propagation of such spikes in mitral cell dendrites (b; upper panel). During the next phase, granule cells likely receive an excitatory synaptic input leading to a large, complex sink the EPL (c; middle panel). Pyramidal cells in APC also are excited during this phase. The

last phase may reflect polysynaptic activation of granule cells by centrifugally projecting neurons locate in APC or caudal to APC. This polysynaptic feedback circuit may mediate the late sink d recorded in the GCL.

576

L. Uva et al. / Neuroscience 142 (2006) 567–577

role in modulating mitral cell discharges. First by directly activating bulbar interneurons, causing them to release GABA onto mitral and tufted cells, and secondarily by depolarizing the distal processes of granule cells and thereby relieving the Mg block of NMDA receptors that normally attenuates dendrodendritic inhibition. Halabisky and Strowbridge (2003) showed that there was a relatively narrow (60 – 80 ms) window in which GCL stimulation can “gate” dendrodendritic inhibition. The relative timing of discharges in mitral cells and projection neurons in piriform cortex following sensory stimulation may govern the strength of mitral cell inhibition and contribute to patterning of the output of the OB. Acknowledgments—Silicon probes were provided by the Center for Neural Communication Technology, University of Michigan. The study was supported by NIH grants NS33590 and DC04285 to B.W.S.

REFERENCES Aroniadou-Anderjaska V, Ennis M, Shipley MT (1999) Current-source density analysis in the rat olfactory bulb: laminar distribution of kainate/AMPA- and NMDA-receptor-mediated currents. J Neurophysiol 81:15–28. Biella G, de Curtis M (1995) Associative synaptic potentials in the piriform cortex of the isolated guinea-pig brain in vitro. Eur J Neurosci 7:54 – 64. Biella G, de Curtis M (2000) Olfactory inputs activate the medial entorhinal cortex via the hippocampus. J Neurophysiol 83:1924 – 1931. Biella G, Uva L, de Curtis M (2002) Propagation of neuronal activity along the neocortical-perirhinal-entorhinal pathway in the guinea pig. J Neurosci 22:9972–9979. de Curtis M, Biella G, Buccellati C, Folco G (1998) Simultaneous investigation of the neuronal and vascular compartments in the guinea pig brain isolated in vitro. Brain Res Protoc 3:221–228. de Curtis M, Biella G, Forti M, Panzica F (1994) Multifocal spontaneous epileptic activity induced by restricted bicuculline ejection in the piriform cortex of the isolated guinea pig brain. J Neurophysiol 71:2463–2475. de Curtis M, Pare D, Llinas RR (1991) The electrophysiology of the olfactory-hippocampal circuit in the isolated and perfused adult mammalian brain in vitro. Hippocampus 1:341–354. de Curtis M, Radici C, Forti M (1999) Cellular mechanisms underlying spontaneous interictal spikes in an acute model of focal cortical epileptogenesis. Neuroscience 88:107–117. Dennis BJ, Kerr DI (1968) An evoked potential study of centripetal and centrifugal connections of the olfactory bulb in the cat. Brain Res 11:373–396. Dennis BJ, Kerr DI (1975) Olfactory bulb connections with basal rhinencephalon in the ferret: an evoked potential and neuroanatomical study. J Comp Neurol 159:129 –148. Dickson CT, Biella G, de Curtis M (2000) Evidence for spatial modules mediated by temporal synchronization of carbachol-induced gamma rhythm in medial entorhinal cortex. J Neurosci 20:7846 – 7854. Halabisky B, Strowbridge BW (2003) Gamma-frequency excitatory input to granule cells facilitates dendrodendritic inhibition in the rat olfactory bulb. J Neurophysiol 90:644 – 654. Hamilton KA, Kauer JS (1988) Responses of mitral/tufted cells to orthodromic and antidromic electrical stimulation in the olfactory bulb of the tiger salamander. J Neurophysiol 59:1736 –1755. Isaacson JS, Strowbridge BW (1998) Olfactory reciprocal synapses: dendritic signaling in the CNS. Neuron 20:749 –761.

Jahr CE, Nicoll RA (1980) Dendrodendritic inhibition: demonstration with intracellular recording. Science 207:1473–1475. Kauer JS (1991) Contributions of topography and parallel processing to odor coding in the vertebrate olfactory pathway. Trends Neurosci 14:79 – 85. Ketchum KL, Haberly LB (1993) Membrane currents evoked by afferent fiber stimulation in rat piriform cortex. I. Current source-density analysis. J Neurophysiol 69:248 –260. Kunze WA, Shafton AD, Kem RE, McKenzie JS (1992) Intracellular responses of olfactory bulb granule cells to stimulating the horizontal diagonal band nucleus. Neuroscience 48:363–369. Librizzi L, de Curtis M (2003) Epileptiform ictal discharges are prevented by periodic interictal spiking in the olfactory cortex. Ann Neurol 53:382–389. Luskin MB, Price JL (1983) The topographic organization of associational fibers of the olfactory system in the rat, including centrifugal fibers to the olfactory bulb. J Comp Neurol 216:264 –291. Martinez DP, Freeman WJ (1984) Periglomerular cell action on mitral cells in olfactory bulb shown by current source density analysis. Brain Res 308:223–233. Mitzdorf U (1985) Current source-density method and application in cat cerebral cortex: Investigation of evoked potentials and EEG phenomena. Physiol Rev 65:37–100. Mori K (1987) Membrane and synaptic properties of identified neurons in the olfactory bulb. Prog Neurobiol 29:275–320. Muhlethaler M, de Curtis M, Walton K, Llinas R (1993) The isolated and perfused brain of the guinea-pig in vitro. Eur J Neurosci 5: 915–926. Nakashima M, Mori K, Takagi SF (1978) Centrifugal influence on olfactory bulb activity in the rabbit. Brain Res 154:301–306. Neville KR, Haberly LB (2003) Beta and gamma oscillations in the olfactory system of the urethane-anesthetized rat. J Neurophysiol 90:3921–3930. Paré D, de Curtis M, Llinás RR (1992) Role of the hippocampalentorhinal loop in temporal lobe epilepsy: extra- and intracellular study in the isolated guinea pig brain in vitro. J Neurosci 12: 1867–1881. Patneau DK, Stripling JS (1992) Functional correlates of selective long-term potentiation in the olfactory cortex and olfactory bulb. Brain Res 585:219 –228. Phillips CG, Powell TP, Shepherd GM (1963) Responses of mitral cells to stimulation of the lateral olfactory tract in the rabbit. J Physiol 168:65– 88. Price JL, Powell TP (1970) The morphology of the granule cells of the olfactory bulb. J Cell Sci 7:91–123. Rall W, Shepherd GM (1968) Theoretical reconstruction of field potentials and dendrodendritic synaptic interactions in olfactory bulb. J Neurophysiol 31:884 –915. Rall W, Shepherd GM, Reese TS, Brightman MW (1966) Dendrodendritic synaptic pathway for inhibition in the olfactory bulb. Exp Neurol 14:44 –56. Rodriguez R, Haberly LB (1989) Analysis of synaptic events in the opossum piriform cortex with improved current source-density techniques. J Neurophysiol 61:702–718. Schoppa NE, Kinzie JM, Sahara Y, Segerson TP, Westbrook GL (1998) Dendrodendritic inhibition in the olfactory bulb is driven by NMDA receptors. J Neurosci 18:6790 – 6802. Shepherd GM (1972) Synaptic organization of the mammalian olfactory bulb. Physiol Rev 52:864 –917. Shepherd GM, Chen WR, Greer CA (2004) Olfactory bulb. In: The synaptic organization of the brain (Shepherd GM, ed), pp 165–216. Oxford, UK: Oxford University Press. Shipley MT, Ennis M (1996) Functional organization of olfactory system. J Neurobiol 30:123–176. Singer MS, Hughes TE, Shepherd GM, Greer CA (1998) Identification of olfactory receptor mRNA sequences from the rat olfactory bulb glomerular layer. Neuroreport 9:3745–3748.

L. Uva et al. / Neuroscience 142 (2006) 567–577 Stripling JS, Patneau DK (1999) Potentiation of late components in olfactory bulb and piriform cortex requires activation of cortical association fibers. Brain Res 841:27– 42. Uva L, de Curtis M (2005) Polysynaptic olfactory pathway to the ipsiand contralateral entorhinal cortex mediated via the hippocampus. Neuroscience 130:249 –258.

577

Uva L, Strowbridge BW, de Curtis M (2005) Olfactory bulb/piriform cortex interactions in the in vitro isolated guinea-pig brain. Soc Neurosci Abstract 279:2. Wellis DP, Kauer JS (1994) GABAergic and glutamatergic synaptic input to identified granule cells in salamander olfactory bulb. J Physiol 475:419 – 430.

(Accepted 19 June 2006) (Available online 2 August 2006)