JN Fuel your research with LabChart
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
 QUICK SEARCH:   [advanced]


     


J Neurophysiol 93: 2302-2317, 2005. First published November 10, 2004; doi:10.1152/jn.00806.2004
0022-3077/05 $8.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
93/4/2302    most recent
00806.2004v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (12)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Wu, C.
Right arrow Articles by Zhang, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Wu, C.
Right arrow Articles by Zhang, L.

INNOVATIVE METHODOLOGY

Size Does Matter: Generation of Intrinsic Network Rhythms in Thick Mouse Hippocampal Slices

Chiping Wu1,5, Wah Ping Luk1, Jesse Gillis1,3, Frances Skinner1,2,3,4,5 and Liang Zhang1,2,5

1Toronto Western Research Institute, University Health Network, 2Department of Medicine, Division of Neurology, 3Department of Physiology, 4Institute of Biomaterials and Biomedical Engineering, and 5Epilepsy Research Program, University of Toronto, Toronto, Ontario, Canada

Submitted 6 August 2004; accepted in final form 7 November 2004


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Rodent hippocampal slices of ≤0.5 mm thickness have been widely used as a convenient in vitro model since the 1970s. However, spontaneous population rhythmic activities do not consistently occur in this preparation due to limited network connectivity. To overcome this limitation, we develop a novel slice preparation of 1 mm thickness from adult mouse hippocampus by separating dentate gyrus from CA3/CA1 areas but preserving dentate–CA3-CA1 connectivity. While superfused in vitro at 32 or 37°C, the thick slice exhibits robust spontaneous network rhythms of 1–4 Hz that originate from the CA3 area. Via assessing tissue O2, K+, pH, synaptic, and single-cell activities of superfused thick slices, we verify that these spontaneous rhythms are not a consequence of hypoxia and nonspecific experimental artifacts. We suggest that the thick slice contains a unitary circuitry sufficient to generate intrinsic hippocampal network rhythms and this preparation is suitable for exploring the fundamental properties and plasticity of a functionally defined hippocampal "lamella" in vitro.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
The hippocampus is a limbic structure crucial for cognitive functions and highly susceptible to ischemic-hypoxic injuries and epileptic seizures. The rodent hippocampus exhibits distinct electroencephalographic rhythmic activities that are tightly linked with behavioral states. Particularly, hippocampal sharp waves that occur during slow wave sleep and consummatory behaviors have long been considered an intrinsic property of hippocampal networks (Buzsáki 1986Go; Buzsáki et al. 1983Go; O'Keefe and Nadel 1978Go; Suzuki and Smith 1987Go). However, our knowledge about cellular and neurochemical basis of intrinsic hippocampal network oscillations and their alterations by ischemia/hypoxia and epileptic seizures remains incomplete.

In the past, numerous studies have explored hippocampal rhythmic activities using conventional hippocampal slices (≤0.5 mm thickness) as a convenient in vitro model. Whereas neuronal rhythmic activities of various frequencies can occur in conventional slices after ionic/pharmacological manipulations or repeated afferent stimulation, spontaneous population rhythmic activities do not consistently arise in this preparation. This may be largely due to insufficient network connectivity preserved in vitro because CA3 and CA1 neuronal connections have a widespread distribution pattern along the ventral-dorsal hippocampal axis that is well beyond the boundary of conventional slices (Ishizuka et al. 1990Go; Li et al. 1994Go). One challenge is to have an innovative approach that preserves a large hippocampal circuitry in vitro and allows investigation of spontaneous network activities and their pathological alterations at both single-cell and large-scale network levels.

In this undertaking, Khalilov et al. (1997)Go have described a neonatal rat (at least postnatal day 11) hippocampal model that allows population neuronal activities to occur spontaneously. Because the morphological and electrophysiological maturation of rodent hippocampal neurons stabilize after the third postnatal week (Ben-Ari et al. 1989Go; Gomez-Di Cesare et al. 1997Go; Pokorny and Yamamoto 1981; Spigelman et al. 1992Go; Zhang et al. 1991Go), we have modified Khalilov's model and produced a hippocampal preparation from day 21–28 mice. Our approach is to isolate whole mouse hippocampus form the brain and remove the dentate gyrus while preserving CA3-to-CA1 connectivity (Wu et al. 2002Go). While superfused in vitro, the mouse hippocampal isolate exhibits spontaneous rhythmic field potentials (SRFPs, 1–4 Hz) that arise from the CA3 and spread longitudinally from the ventral hippocampus toward the dorsal hippocampus. The SRFPs are correlated with synchronous GABAA inhibitory postsynaptic potentials (IPSPs) in CA1 pyramidal neurons and rhythmic excitatory postsynaptic potentials (EPSPs)-discharges in inhibitory interneurons. As such, they are thought to represent an intrinsic GABAergic rhythm. However, our previous attempts to produce a similar preparation from adult mice were not successful likely due to insufficient in vitro oxygenation of densely packed mature hippocampal tissue. The issue remains as to whether a relatively large circuitry of adult rodent hippocampus can be reliably preserved in vitro that exhibits the IPSP-based SRFPs.

To address this issue, we produce a thick hippocampal slice preparation from adult mice. Considering that the dentate gyrus is structurally separated from CA2/CA1 areas by the hippocampal fissure, we separate the dentate gyrus from CA2/CA1 areas along the hippocampal fissure and obtain slices of 1-mm thickness along the hippocampal transverse plane. The thick hippocampal slice looks like a "C"-shaped tissue strip that preserves the dentate gyrus–CA3-CA1 connectivity and allows a direct exposure of its deeper layers to oxygenated perfusate while superfused in vitro. We verify the "healthiness" of the superfused thick slices via measuring tissue O2, K+, pH, and synaptic/intracellular activities. Our data show that SRFPs persist in nearly every thick slice successfully prepared and their properties are in keeping with a CA3-driven, IPSP-based intrinsic network rhythm. Therefore our thick slice preparation contains a unitary circuitry sufficient to stably generate intrinsic hippocampal network rhythms.


    METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Animals

C57Bl/6 mice (21 days to 12 mo, Charles River) were used in the present experiments. All experimentations were conducted after approval from the animal care committee of our institution.

Tissue preparations

Briefly, we decapitated the animal under halothane anesthesia, quickly dissected the brain and kept the brain in an ice-cold, oxygenated artificial cerebrospinal fluid (ACSF) for 4–5 min before further dissection. The hemi-sectioned brain was then glued onto a shallow container with its temporal cortex facing down (Fig. 1A), and brain stem-thalamic tissue was removed to expose the dorsal (or dentate gyrus) side of the hippocampus (Fig. 1B). When freshly prepared, a longitudinally orientated hippocampal artery could be readily recognized form the dorsal surface of the hippocampus under a dissecting microscope (Andersen et al. 1971Go). This artery attached to the entry area of hippocampal fissures and could be used as an anatomical marker for the underlying hippocampal fissure. To separate the dentate gyrus from the CA2/CA1 areas, we applied a fine glass probe along the hippocampal fissure and gently pushed the dentate gyrus toward the fimbra-fonix fiber bundle. By doing so, the dentate gyrus was detached from the distal portion of CA2/CA1 stratum radiatum but remained connected with the CA3 (Fig. 1C). The brain tissue was then glued onto an agar block, and 3–4 slices of 1-mm thickness were obtained via vibrotome cuts (Technical Precision Instruments, St. Louis, MO) along the transverse plane of the middle-dorsal hippocampus. The ventral 1/3 hippocampus was not used in the present experiments because of the curved ventral hippocampus and the difficulty of having a clean separation between dentate gyrus and CA3/CA1 areas. After slicing, adjacent cortical tissue was surgically removed. The removal of cortical tissue, together with separation of dentate gyrus from CA2/CA1 areas produced a C-shaped tissue strip (Fig. 1D), allowing a direct exposure of deeper layers of the thick slice to the oxygenated ACSF.



View larger version (76K):
[in this window]
[in a new window]
 
FIG. 1. Key procedures in preparation of thick mouse hippocampal slices. A: a hemi-sectioned mouse brain was kept in the ice-cold, oxygenated artificial cerebrospinal fluid (ACSF) for a few minutes before further dissection. B: an enlarged photo was taken after removal of cerebellum-brain stem-thalamic tissues from the hemi-sectioned brain, showing the dorsal (or dentate gyrus) side of the hippocampus. ->, the longitudinally orientated hippocampal artery that attached to the entry area of hippocampal fissure. C: the dentate gyrus was separated from the CA2/CA1 area by using a fine glass probe. The glass probe was applied along hippocampal fissure (or the longitudinal hippocampal artery), and the dentate gyrus was gently pushed toward the CA3 area (inner curve of the hippocampus). The separation was done under a dissecting microscope and the brain tissue was glued onto a shallow container with its temporal cortex facing down. An enlarged photo was taken after separation of the dentate gyrus and the glass probe (on right) was used to hold the separated dentate gyrus. D: after dentate gyrus separation, the brain tissue was glued onto an agar block and vibrotome slices of 1-mm thickness were obtained along the transverse plane of middle-dorsal hippocampus. An enlarged photo was taken from a 1-mm dorsal hippocampal slice that was placed on onto the stainless mesh of our recording chamber.

 
The thick slices were kept in the oxygenated ACSF at ~32°C for 1–6 h before transferring to a recording chamber. In some experiments, guided under a dissecting microscope, we made surgical cuts in the recording chamber to separate dentate gyrus from CA3 or CA3 from CA1 area. The ASCF contained (in mM) 3.5 KCl, 1.25 NaH2PO4, 125 NaCl, 25 NaHCO3, 2 CaCl2, 1.3 MgSO4, and 10 glucose. The pH of the ACSF was ~7.4 when aerated with 95% O2-5% CO2.

For preparing conventional brain slices, the mouse brain was glued on an aluminum block with its basal side facing down and a dorsal brain section of ~2.4-mm thickness was obtained via a vibrotome cut. Horizontal brain slices of 0.5-mm thickness that contained ventral hippocampal tissue at transverse plane were then obtained via cutting the remaining the basal brain (2–3 slices per half brain). The dorsal brain section was glued onto an agar block and vibrotome slices of 0.5-mm thickness were obtained along the transverse plane of dorsal hippocampus (2–3 slices per half brain). These slices were similarly maintained in vitro as described in the preceding text.

Perfusion apparatus

We used a submerged recording chamber with inner dimensions of 3.5 x 5 x 20 mm (depth x width x length) in the present experiments (Wu et al. 2002Go). The thick slice was held on a stainless steel fine mesh (0.015-in grid length) via six to eight mosquito pins or a frame made of fine stainless steel wires. The mesh was ~1.5 mm above the bottom of the chamber to allow the perfusion of the oxygenated ACSF to both sides of the slice. The oxygenated ACSF was warmed and superfused to the slice at ~32°C. A water bath underneath the recording chamber was set at ~32°C via an automatic temperature control unit, which allowed warmed and humidified 95% O2-5% CO2 passing over the perfusate to increase local oxygen tension in the recording chamber. By placing a fine temperature probe in the ACSF near the superfused slice, we verified that the perfusate temperature was at ~32°C. The slice was superfused at a rate of ~15 ml/min but at a minimal submerged level. The use of a high rate of superfusion was to achieve an effective exchange between the oxygenated ACSF and the thick slice.

A brief hypoxic episode was conducted by superfusing the slice with the ACSF that was aerated with 95% N2-5% CO2 for ~5 min. The humidified 95% O2-5% CO2 that passed over the perfusate was also replaced with 95% N2-5% CO2 during the hypoxic episode (Chung et al. 1998Go; Perez-Velazques and Zhang 1994Go).

Extracellular recordings and afferent stimulation

Extracellular recording electrodes were pulled from thin-wall glass pipettes (1.5 mm OD, World Precision Instrument, Sarasota, FL) via using a Narishige vertical puller. The resistance of these electrodes was 1–2 M{Omega} when filled with a solution containing 150 mM NaCl and 2 mM HEPES (pH 7.4). Extracellular signals were sampled via an Axoclamp-2B (Axon Instruments, Union City, CA) or an extracellular amplifier from A-M Systems (Model 3000, Carlsborg, WA). The input frequency of these amplifiers was set in the range of 0 to 3 kHz. Extracellular signals were amplified by 2,000–5,000 times before digitization (Digidata 1200 or 1300A, Axon Instruments).

Local afferent stimulation was conducted via placing a bipolar tungsten wire electrode (tip diameter of 50 µm) in the CA3 or CA1 stratum radiatum or dentate middle molecular layer. Constant current pulses of 0.1 ms were generated by a Grass stimulator and delivered through an isolation unit every 30 s.

Single-cell recordings

Patch-clamp recordings were conducted via a "blind" approach as previously described (Wu et al. 2002Go; L. Zhang et al. 1991Go, 1994Go; Y. Zhang et al. 1998Go). The recording electrodes were made with the thin wall glass pipettes described in the preceding text. The resistance of these electrodes was 4–5 M{Omega} when filled with an internal solution that contained 120 mM potassium gluconate, 2 mM HEPES, 0.1 mM EGTA and 0.5% neurobiotin (pH 7.25 and 280–290 mosM). For perforated-patch recordings, we used a patch pipette solution that contained 150 mM KCl, 2 mM HEPES, 0.1 mM EGTA, and gramicidin (≤50 µg/ml, Sigma) (Rhee et al. 1994Go). Intracellular responses were collected for analysis only if the access resistance of the perforated configuration was ≤60 M{Omega} (Wu et al. 2002Go; Zhang et al. 1996Go). In some experiments, we monitored neuronal discharges in cell-attached, voltage-clamp recordings. This recording configuration did not provide a full voltage control over the large currents associated with action potentials but allowed a reliable and noninvasive measurement of the timing of corresponding spike currents (Fricker et al. 1999Go; Verheugen et al. 1999Go; Wu et al. 2002Go).

Single-cell signals were sampled via using an Axopatch-200B, Axoclamp-2B, or dual-clamp 700A amplifier (Axon Instruments). Data acquisition, storage, and analyses were done using pCLAMP software (version 8 or 9, Axon Instruments).

To examine morphological properties of the neurobiotin-filled neurons, we fixed the slices with 4% paraformaldehyde/phosphate buffer after termination of the recording. Vibrotome sections of 50–100 µm in thickness were obtained and stained with the ABC kit (Vector Laboratory, Ontario, Canada) (Wu et al. 2002Go; Zhang et al. 1998Go). A Zeiss camera lucida drawing device was used to trace the neurobiotin-filled cellular processes.

Measurements of tissue O2

We constructed carbon fiber electrodes as per Jiang et al. (1991)Go and Mulkey et al. (2001)Go. Briefly, a single carbon fiber (diameter: ~10 µm) was sealed with epoxy (Epon 812, Fluka) in an electrode (tip diameter of 20–30 µm) pulled form the thin-wall glass pipette described in the preceding text. The carbon fiber protruded the glass electrode tip by ~20 µm. The other end of the carbon fiber was attached to a copper wire using a silver conductive paint adhesive (Silver Print, GC Electronics, Rockford, IL) and connected to the inputs of a polarographic amplifier (Chemical Microsensor 1201, Diamond General, Ann Arbor, MI). A voltage of –0.85 V was applied between the carbon fiber electrode and an Ag/AgCl reference electrode. The carbon fiber electrodes were calibrated at ~32°C before and after the measurements in slices. For calibration, ACSFs were aerated with 95% N2-5% CO2, 25% O2-5% CO2-70% N2, 50% O2-5% CO2-45% N2, or 95% O2-5% CO2 respectively, and the corresponding O2 measurements were 26.6 ± 2.2, 120.6 ± 3.7, 221.4 ± 21.3, and 310.8 ± 19.1 mmHg, respectively (n = 22 electrodes). These measurements could be fitted via a linear regression line with r = 0.994 and slope = 3.4. The measurement of 26.6 ± 2.2 mmHg that corresponded to 95% N2-5% CO2 reflected the background signals of these O2-sensitive carbon fiber electrodes, which did not affect the slope or O2 sensitivity of these electrodes.

To assess O2-depth profiles in the superfused slice, the carbon fiber electrode was advanced into the CA3 or CA1 proximal stratum radiatum area at an angle of 75° vertical to the horizontal plane (Fig. 2A). A micromanipulator was used to advance the electrode in a stepwise manner. The exact distance of the carbon electrode tip traveled in each movement could not be determined because of tissue dimpling associated with electrode movement. The estimated travel distance of the electrode tip was ~80 µm/step. Two observations were helpful in approximating the position of the carbon electrode in the slice. First, we observed a drop in measured O2 level (by 30–40 mmHg) as the electrode was advanced from the perfusate toward the slice surface. Others have noted a similar drop in previous studies (Bingmann and Kolde 1982Go; Fujii et al. 1982Go; Ganfield et al. 1970Go; Jiang et al. 1991Go) thought to reflect an O2 diffusion-limited area at the slice-ACSF interface hence symbolize the electrode at the superficial layer (≤50 µm) of the slice. Second, the tissue O2 measured at different depths of the slices exhibited an asymmetrical U-shaped profile (Fig. 2A), i.e., higher O2 levels at the top and bottom sides of the slice and the lowest level at the middle layer of the slice. The U-shaped profile likely resulted from an effective superfusion of both top and bottom sides of the thick slice in our recording chamber. Mulkey et al. (2001)Go have described a similar U-shaped O2-depth profile while superfusing both top and bottom sides of conventional brain slices in a submerged chamber.



View larger version (33K):
[in this window]
[in a new window]
 
FIG. 2. Measurements of tissue O2 and external K+. A: a representative trace (top) shows O2 measurements from a thick slice. The measurement was made by a carbon fiber electrode in the CA3 proximal stratum radiatum area at different depths of the slice. The carbon electrode was advanced in a stepwise manner at an angle of 75° vertical to the horizontal plane (see the schematic illustration). The O2 level measured from the top surface of the slice is indicated. B: tissue O2 levels were measured at different depths of superfused slices. Data (means ± SE) were normalized as percentage of O2 levels in the superfused ACSF, which were 202.1 ± 9.5 and 198.2 ± 7.6 mmHg for 1-mm slices with and without separation of dentate gyrus, respectively (n = 10 slices in each group) and 196.5 ± 9.1 mmHg for 0.5-mm slices (n = 6). *, P < 0.05, one-way ANOVA, 1-mm slices with vs. without separation of dentate gyrus. C: extracellular field potentials and tissue O2 measurements were collected simultaneously from the CA3 area of a 1-mm slice. O2 measurements were made at the middle layer of the slice, and baseline tissue O2 before hypoxia was ~49 mmHg. The extracellular trace was treated with a band-pass filter of 0.5–285 Hz for illustration purpose. D: CA1 somatic field potentials were evoked from another 1-mm slice before, at the end of a 5-min hypoxic episode and after re-oxygenation for ~10 min. Each trace was averaged from 3 consecutive measurements. Corresponding O2 levels measured from the middle layer of the slice are indicated above the traces. E: extracellular field potentials and tissue O2 measurements were collected simultaneously from the CA1 area of a thick slice. The baseline tissue O2 before high-frequency afferent stimulation (40 Hz for 3 s) was 55 mmHg. F: external K+ and nearby field potentials were recorded simultaneously from the CA3 area of a thick slice. The baseline external K+ before high-frequency stimulation (20 Hz, 1 s) was ~4.0 mM. G: extracellular K+ measured at different depths of thick slices. Data (means ± SE) were normalized as percentage of the K+ levels at slice surface (3.7 ± 0.2 mM).

 
K+ and pH measurements

Extracellular K+ and pH were measured using ion-selective electrodes (Ammann 1986Go; Amzica and Steriade 2000Go; Morris 1995Go), and the preparation of these electrodes have been described in detail previously (Wu et al. 2002Go). Briefly, the electrode was pulled from the thin-wall glass tubes described above (tip diameter of ~20 µm). For making the K+-selective electrodes, one barrel of the two-barrel electrode was filled at the tip with potassium ionophore I-cocktail A (Fluka, Buchs, Switzerland) and then back-filled with 200 mM KCl. The reference barrel was filled with 200 mM NaCl. For the pH-selective electrodes, the sensing barrel was filled with hydrogen ionophore-II cocktail A (Fluka) and then back-filled with a solution containing (in mM) 100 NaCl, 20 HEPES, and 10 NaOH (pH 7.25). Signals were recorded using a differential amplifier designed for ion-sensitive electrodes (input impedance of >1014{Omega}, A-M Systems). The calibration measurements of the K+ electrodes were: 5.0 ± 1.5 mV for 2.5 mM KCl, 23.8 ± 3.8 mV for 5 mM KCl, 42.8 ± 3.6 mV for 10 mM KCl, and 62.0 ± 5.8 mV for 20 mM KCl (n = 6 electrodes). These measurements could be fitted with a linear regression line with r = 0.998. The extrapolated sensitivity of the K+ electrodes was ~60 mV/10-fold increase in KCl concentration. The sensitivity of pH electrodes was 52.5 ± 5.0 mV for a change in pH from 6.8 to 7.8 (n = 4 electrodes). Calibration solutions were similar to the ACSF with either KCl or NaHCO3 substituted for corresponding moles of NaCl. The ion selective electrode was similarly advanced into the slice as described in the preceding text.

Data analyses

Evoked synaptic field potentials and basic intracellular parameters were measured as previously described (Wu et al. 2002Go; L. Zhang et al. 1994Go; Y. Zhang et al. 1998Go). Event detection function (threshold search method, pCLAMP software, version 9) was used to pick up the SRFPs and SRFP-correlated intracellular signals. The recognized events were then visually confirmed and false events were rejected. The peak amplitudes and rising and decay times were measured from averaged events.

To assess the presence or absence of SRFPs, a stable and continuous extracellular recording of ≥3 min was collected from each slice. The data collection was made after stabilizing the slice in the recording chamber for ≥20 min. For each slice, an averaged spectral plot was generated from a 30-s data segment (at frequency resolution of 0.1 Hz and averaged every 6 s). The highest peak in 0.5–5 Hz range of the spectral plot was considered as the main frequency of the SRFPs, which was then confirmed by visual inspection of the original data segment. The slices that failed to exhibit SRFPs were determined if extracellular traces showed no rhythmic activity by visual inspection and no evident peak in the 0.5- to 5-Hz range of the corresponding spectral plots was present. The general functionality of slices was determined by examining their responses to afferent stimulation and/or superfused carbachol (10 µM, 6–8 min). The slices that exhibited poor evoked responses, i.e., the amplitudes of CA3 or CA1 field EPSPs or population spikes were ≤0.5 mV in response to near maximal stimulation, were considered as unsuccessful preparations and they were excluded from the present data analyses.

For cross-correlation analyses, original data were treated with a band-pass filter (0.1–285 Hz) and plots were generated form a data segment of 5–10 s for each slice (Origin software, version 6, Origin Lab, Northampton, MA).

To assess the stability of SRFPs, time-frequency analyses of some of the extracellular recordings were made from data segments of a 10-min continuous record. No digital filtering was applied to the original data before the analyses.

Statistical significance was determined using Student's t-test or one-way ANOVA (SigmaStat). Mean ± SE were calculated and presented throughout the text except where indicated.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Verification of thick mouse hippocampal slices

The major concern when using thick slices is the potential influence of hypoxia. Previous studies using O2-sensitive microelectrodes have monitored tissue O2 in slices of different thickness while perfusing these slices in an interface chamber at 35–37°C (Bingmann and Kolde 1982Go; Fujii et al. 1982Go; Ganfield et al. 1970Go; Jiang et al. 1991Go). It has been suggested that anoxia can occur in deeper layers of adult brain slices if their thickness is >0.6 mm (Jiang et al. 1991Go). The consensus is that O2 tension at any depth of a slice is largely determined by O2 diffusion distance into the slice and metabolic rates of local cells. For a conventional slice of several square millimeters in surface area and placed onto a nylon mesh or solid base, O2 diffusion into the slice is mostly from the top surface, which renders a limited O2 delivery and hence anoxia in deep layers of thick slices. Therefore for thick slices to survive in vitro, a key issue is to reduce the O2 diffusion distance to the deeper layers of the slice. We thus designed our thick slice preparation accordingly.

Taking the advantage that the dentate gyrus and CA2/CA1 area are structurally divided by the hippocampal fissure, we separated the dentate gyrus from CA2/CA1 areas along hippocampal fissure while kept the dentate gyrus–CA3-CA1 connectivity (Fig. 1C). After removal of adjacent cortical tissue, the thick slice looked like a C-shaped tissue strip (Fig. 1D), and its deeper layers were in direct contact with the oxygenated ACSF while superfused in vitro. In addition, we superfused the thick slice in a submerged chamber that allowed an effective perfusion of both top and bottom sides of the slice at a high rate (~15 ml/min, see METHODS).

Measurements of tissue O2

The ultimate assessment of oxygenation is to directly measure tissue O2 in the superfused slices (Jiang et al. 1991Go). We thus constructed carbon fiber microelectrodes as per Jiang et al. (1991)Go and Mulkey et al. (2001)Go and measured tissue O2 at different depths of the superfused thick slices. The O2 depth profile was measured from the CA3 or CA1 proximal striatum radiatum area because it was the middle region of the thick slice and likely had a relatively long O2 diffusion distance as compared with other sub-regions. Thus the O2 measurement from this area would be sensitive to detect the potential hypoxia, if it occurred, in the superfused slice. We used a micromanipulator to advance the carbon fiber electrode into the thick slice at an angle of 75° vertical to the horizontal plane. The electrode was advanced in a stepwise manner. At each step, we monitored a stable O2 level for ≥10 s and moved the electrode into adjacent deeper area until the electrode tip reached to the bottom of the superfused thick slice. A schematic illustration of such measurements is shown in the lower panel of Fig. 2A.

We found an asymmetrical U-shaped O2-depth profile in all thick slices examined, i.e., higher O2 levels at the top and bottom sides of the slices and the lowest level at the middle layer (Fig. 2A). A similar U-shaped O2-depth profile was also observed from 0.5-mm slices that were similarly superfused. The O2 levels measured at the middle layers (the lowest O2 level) of 1- and 0.5-mm slices were comparable, i.e., 41.8 ± 2.9 and 46.0 ± 3.8% relative to the O2 levels at slice surface respectively (P = 0.27, one-way ANOVA; Fig. 2B). We used a submerged recording chamber with inner dimensions of 3.5 x 5 x 20 mm (depth x width x length), and the thick slice was held on a stainless steel fine mesh setting ~1.5 mm above the bottom of the chamber (see METHODS for details). Thus both top and bottom sides of the thick slices had effective exchanges with the oxygenated perfusate, and such U-shaped O2-depth profile might reflect a distance- and metabolism-dependent O2 delivery process to the slice under our experimental conditions.

We also measured the O2-depth profile in 1-mm hippocampal slices (n = 6) that were similarly prepared as described in the preceding text but without separation of dentate gyrus from CA2/CA1 areas. When similarly superfused, the O2 levels at deeper layers of these slices were significantly lower than those measured from the thick slices with separation of dentate gyrus (P < 0.05, one-way ANOVA; Fig. 2B). In addition, evoked synaptic field potentials were absent in the thick slices without dentate gyrus separation, likely reflecting the hypoxia-induced suppression of synaptic transmission. These observations indicate that the separation of dentate gyrus from CA2/CA1 areas is crucial for sufficient oxygenation and functional preservation of the thick slices. Therefore the thick slices with dentate gyrus separation were exclusively used in the following experiments.

The O2 levels measured from the superfused slices may be overestimated or underestimated due to potential drifts in background signals of O2-sensitive carbon microelectrodes. To control this, we exposed the thick slices to brief (~5 min) hypoxic episodes (Chung et al. 1998Go; Perez-Velazques and Zhang 1994Go) and measured the corresponding decrease in tissue O2. As measured at the middle layer of the thick slices, the tissue O2 was decreased by 31.1 ± 7.1 mmHg at the end of hypoxic episodes (n = 7 slices, P = 0.0048, paired t-test) and then gradually returned to the baseline level. In parallel to the drop in tissue O2, SRFPs and evoked synaptic potentials were reversibly suppressed by the hypoxic episodes (Fig. 3, C and D). The latter observation is consistent with our previous studies in conventional rat hippocampal slices (Chung et al. 1998Go; Ouanonou et al. 1999Go; Perez-Velazques and Zhang 1994Go) and in the mouse hippocampal isolate (Wu et al. 2002Go).



View larger version (33K):
[in this window]
[in a new window]
 
FIG. 3. Basic electrophysiological features of thick slices. A, left: superimposed traces show CA1 field EPSPs (F-EPSPs) evoked by paired stimuli from a thick slice (inter-pulse interval of 100 ms, stimulation intensities of 10–70 µA). Each illustrated traces were averaged from 3 consecutive recordings. Middle: CA1 F-EPSPs were evoked by paired stimuli (interval of 100 ms) from 6 thick slices. The amplitudes of the 1st ({blacksquare}) and 2nd ({bullet}) F-EPSPs (mean ± SE) were plotted vs. stimulation intensities. F-EPSPs that associated with population spikes after strong afferent stimulation were not included in the plot. Right: CA1 F-SPSPs were evoked paired stimuli from the same slice as shown at left, and the stimulation intensity was kept constant at 30 µA and the interpulse interval was 25, 50, 100, 200, 500, or 1,000 ms, respectively. The amplitude ratios of the 2nd vs. 1st F-EPSPs were calculated by taking the 1st F-EPSPs as 100%. B: CA1 somatic population spikes were evoked by paired stimuli from a thick slice, and each illustrated trace was averaged from 4 consecutive measurements. The amplitude ratios of the 2nd vs. 1st population spikes were calculated taking the 1st population spikes as 100%. C and D, left and middle: CA1 field potentials were recorded from superficial layers (left) and then from bottom layers (middle) of 2 separate thick slices, following similar stimulation at superficial layers of the CA2 area. Each trace was averaged from 4 consecutive measurements. Right: responses were collected from the bottom layers following super-maximal stimulation at superficial layer (D) or stimulation of deeper layers after inserting the stimulating electrode into the slice (C). E: evoked synaptic field potentials were recorded from a thick slice and the stimulation electrode was placed in the dentate gyrus middle molecular layer, CA3 or CA1 stratum radiatum, respectively. Each trace was averaged from 5 consecutive measurements. F: a CA3 pyramidal neuron and interneuron were recorded with a neurobiotin-containing internal solution, and repeated discharges were induced by intracellular injections of depolarizing current pulses. After termination of the recording, the slices were fixed, re-sectioned (100 µm thickness) and stained with the ABC kit. Photos were taken under a x20 (pyramidal neuron) or x40 (interneuron) objective lens, respectively. Camera lucida traces of the interneuronal processes were collected from 5 sections and then overlaid.

 
Fujii et al. (1982)Go have shown that in olfactory cortical slices (300–600 µM), repeated afferent stimulation (5–30 Hz) causes a rapid decrease in tissue O2, likely reflecting an activity-dependent increase in tissue O2 consumption. Recent studies have shown that sensory inputs or afferent stimulation cause similar decreases of tissue O2 in the cerebral cortex of intact animals (Ances et al. 2001Go; Masamoto et al. 2003Go; Padnick et al. 1999Go; Thompson et al. 2003Go). To test whether the activity-dependent changes in tissue O2 occurs in the thick slices, we stimulated CA3 or CA1 area (10–40 Hz, 1–2 s) and monitored tissue O2 at the middle layer of the slices. We found that the afferent stimulation caused a reversible decrease in tissue O2 by 8.9 ± 0.7 mmHg (n = 13, P < 0.0001, paired t-test). The decrease in tissue O2 appeared in ≤1 s after the onset of afferent stimulation and returned to baseline levels in 15 s (Fig. 2E).

Collectively, the preceding observations indicate that the 1-mm slices were adequately oxygenated under our experimental conditions and that there was no anoxic zone in the middle or deeper layers of the superfused thick slices.

Measurement of extracellular K+ and pH

We constructed ion selective electrodes (Ammann 1986Go; Amzica and Steriade 2000Go; Morris 1995Go) and measured extracellular K+ and pH from the superfused thick slices. As measured from the CA3 proximal striatum radiatum at the middle layer of thick slices, the basal extracellular pH was 7.23 ± 0.2 (n = 4 slices), which was lower than the perfusate pH (7.4 ± 0.1) but was of a similar magnitude to conventional rat hippocampal slices (Voipio and Kaila 1993Go; Xiong and Stringer 2000). The basal extracellular K+ was 3.9 ± 0.1 mM (8 slices), close to the KCl concentration (3.5 mM) in the perfusate. Unlike the U-shaped O2-depth profile described in the preceding text, the levels of extracellular K+ remained nearly unchanged when measured at different depths of superfused thick slice. The extracellular K+ levels at the middle layer of thick slices were only 1.5 ± 0.5% higher than that at slice surface (n = 4 slices, Fig. 2G). In addition, repeated afferent stimulation (5–40 Hz for 1 s) caused a rapid and reversible increase in extracellular K+. The increases of extracellular K+ were 1.1 ± 0.2 mM after stimulation at 5 Hz and 2.1 ± 0.3 mM after stimulation at 20–40 Hz (n = 8 and 10, respectively, P ≤ 0.045, paired t-test; Fig. 2F). Others have observed similar activity-dependent increases in extracellular K+ in conventional hippocampal slices or cultured hippocampal slices, and they suggested that both glial and neuronal ionic buffering systems play important roles in determining the amplitude and recovery rate of extracellular K+ (Kann et al. 2003Go; Walz and Wuttke 1999Go). Our observations are consistent with these studies, further suggesting that the superfused thick slices were adequately oxygenated and capable of supporting energy-dependent ionic buffer systems.

Measurements of synaptic and intracellular activities

Previously in conventional rat hippocampal slices, we showed that the evoked synaptic field potentials are largely mediated by AMPA glutamate receptors (Ouanounou et al. 1999Go; Shinno et al. 1997Go) and highly vulnerable to brief hypoxic episodes (Chung et al. 1998Go; Perez-Velazques and Zhang 1994Go). We thus evoked synaptic field potentials in thick slices to assess general functionality of glutamate synapses. CA1 field EPSPs were detectable after the afferent stimulation at ~20 µA (constant current pulses, 0.1 ms in duration), and their amplitudes were increased in a nearly linear fashion with the increase of stimulation intensity in the range of 20–70 µA (Fig. 3A). CA1 field EPSPs with evident population spikes could be induced by stronger stimuli (≥70 µA), and the amplitude of these responses appeared to be saturated as the stimulation intensities reached ≥100 µA. We observed paired-pulse facilitation of CA1 field EPSPs in a wide range of stimulation intensities tested, and the paired-pulse facilitation appeared to be the greatest when the inter-pulse interval was set at 100 ms (Fig. 3A). When evoked by paired stimuli with the inter-pulse interval of 100 ms and the intensity of 30–40 µA, the paired-pulse facilitation of CA1 field EPSPs were 133.8 ± 3.2% (n = 6 slices).

Similar afferent stimulation induced typical somatic field potentials when the extracellular recording electrode was placed near the cell body layer of the CA3 or CA1 area. Paired stimuli at intensities of ≥70 µA induced paired-pulse inhibition of CA1 somatic population spikes in a narrow interval timeframe. When stimulated at an interval of 5 ms, the second population spikes were reduced to 31.9 ± 13.1% of their preceding responses. There was no evident inhibition but facilitation in the CA1 population spikes when the inter-pulse intervals of paired stimuli were >20 ms. The observed paired pulse inhibition might largely reflect the effect of feedback IPSPs as the result of discharges of CA1 pyramidal neurons and subsequently the activation of local interneurons (Fig. 3B).

To test the depth profile of the CA1 somatic field potentials, we advanced the extracellular recording electrode from superficial layers toward the deeper layers in a stepwise manner as described in the preceding text while stimulating superficial layers at constant intensity. The CA1 population spikes exhibited a depth-dependent decrease, and their amplitudes were 1.72 ± 0.3 mV as measured from superficial layers and ≤0.1 mV when recorded from the bottom layer of the thick slices (n = 7 slices). However, CA1 population spikes were reliably recorded from the bottom layers of the thick slices (ranged from 0.5 to 1.5 mV) after super-maximal stimulation of superficial layers (Fig. 3C) or after stimulation of deeper layers by inserting the stimulating electrode into the thick slice (Fig. 3D). These observations suggest that the depth-related decrease in the CA1 population spikes was largely due to the limitation of local afferent stimulation and that pyramidal neurons at the bottom layers of the thick slices could generate synchronized discharges in response to effective afferent stimulation.

In response to the local afferent stimulation at near maximal strength (≥100 µA), the peak amplitudes of somatic population spikes and dendritic field EPSPs were 2.4 ± 0.3 and 1.8 ± 0.3 mV in the CA3 area and 2.8 ± 0.3 and 1.4 ± 0.2 mV in the CA1 area, respectively (n = 32 slices). The dentate gyrus somatic population spikes were 1.3 ± 0.2 mV, and the corresponding (presumably conveyed via the tri-synaptic pathway) CA1 dendritic field EPSPs were 0.2 ± 0.05 mV (n = 12 slices, Fig. 3E). The waveforms and amplitudes of these synaptic field potentials were comparable to our previous observations in conventional rat hippocampal slices (Ouanounou et al. 1999Go; Perez-Velazques and Zhang 1994Go; Zhang et al. 1998Go).

We then examined the activities of individual neurons via patch-clamp recordings (Zhang et al. 1991Go, 1994Go). Pyramidal neurons and putative inhibitory interneurons were recognized by their morphological and/or electrophysiological properties (Wu et al. 2002Go; Zhang et al. 1998Go). The basic intracellular parameters of CA3 pyramidal neurons were: resting potential –63.3 ± 0.8 mV, input resistance 100.0 ± 4.9 M{Omega}, spike peak amplitude 103.9 ± 1.6 mV, and spike half-width 1.1 ± 0.1 ms (n = 82). The parameters for CA1 pyramidal neurons were: resting potential –66.5 ± 1.0 mV, input resistance 111.8 ± 6.8 M{Omega}, spike peak amplitude 98.2 ± 2.6 mV and spike half-width 1.4 ± 0.1 ms (n = 26). The parameters for CA3 INs were: resting potential –55.0 ± 1.4 mV, input resistance 138.9 ± 15.7 M{Omega}, spike peak amplitude 84.3 ± 5.6 mV, and spike half-width 0.8 ± 0.1 ms (n = 15). Overall, these basic intracellular parameters are similar to measurements made from conventional mouse hippocampal slices (Shuttleworth and Connor 2001Go). Figure 3F shows examples of repeated discharges and neurobiotin-filled cellular processes of a CA3 pyramidal neuron and interneuron.

Consistent observation of SRFPs from thick slices

Having verified the general synaptic and neuronal activities of the thick slices, we then focused on the SRFPs previously observed in our whole mouse hippocampal preparation (Wu et al. 2002Go). Because the SRFPs were found to originate from the CA3 area, we particularly focused in the CA3 area in the following experiments.

Stable SRFPs were observed in 316 of 338 thick slices successfully prepared from middle-dorsal mouse hippocampus, which was in sharp contrast to the absence of SRFPs in 0.5-mm mouse dorsal hippocampal slices examined (n = 26 slices). The rate of observing SRFPs was also greater (P < 0.001, {chi}2 test) in the thick slices (94%) than in 0.5 mm ventral hippocampal slices (30 of 73 slices, 41%, see following text).

There was no clear relation between the SRFP frequencies and the ages of mice used to prepare the thick slices. The frequencies of CA3 SRFPs were 1.7 ± 0.1 Hz in the slices (n = 129) prepared from postnatal day 21–31 mice, 1.8 ± 0.1 Hz in the slices (n = 160) prepared from 2- to 3-mo-old mice and 1.5 ± 0.2 Hz in slices (n = 27) prepared from 4- to 12-mo-old mice, respectively (P > 0.05, ANOVA).

Stability and waveforms of SRFPs

SRFPs could persist for several hours in the thick slices while continuously superfused at ~32°C. In a set of five slices examined, the frequencies of CA3 SRFPs were 2.17 ± 0.42 and 1.67 ± 0.23 Hz as measured initially (after ~30 min of stabilization in the recording chamber) and ~4 h later respectively (P = 0.197, paired t-test; Fig. 4A). SRFPs could also be observed while superfusing the thick slices at ~37°C. The frequencies of CA3 SRFPs were 2.0 ± 0.3 Hz, and the amplitudes of evoked somatic population spikes were 1.9 ± 0.4 mV after ~2 h of superfusion at ~37°C (n = 5 slices; Fig. 4B).



View larger version (54K):
[in this window]
[in a new window]
 
FIG. 4. Stability of SRFPs. A: representative extracellular field potentials were collected from the CA3 somatic area initially (~30 min after placing the slice in the recording chamber) and 4 h later respectively. The traces at right were averaged from 177 SRFP events (dotted line, SD). B: SRFPs were collected the CA3 somatic area of another thick slice superfused at ~37°C. C: spectral plots were generated from extracellular data segments of 30 s including the traces shown in A and B. D: representative extracellular traces at top were collected simultaneously from the CA3 and CA1 areas of a thick slice. Corresponding spectrograms were generated from 10-min data segment including the traces shown at top. Plots are presented in 0–50 Hz range (top, log-amplitude and at 0.5-s resolution) and 1–8 Hz range (bottom, 5-s resolution), respectively.

 
The waveforms of CA3 SRFPs are in keeping with our previous study in the CA1 area of mouse hippocampal isolate (Wu et al. 2002Go). In general, the CA3 SRFPs were 30–300 µV in amplitude and 30–150 ms in duration, and their frequencies varied from slice to slice in the range of 0.5–4.5 Hz. The SRFPs exhibited positive (upward) waveforms as recorded from the peri-somatic areas (striatum pyramidale) where typical somatic population spikes were induced after local afferent stimulation (Fig. 4B). The SRFPs showed negative (downward) waveforms when recorded from the CA3 dendritic (striatum radiatum) areas where the afferent stimulation induced field EPSPs. When recorded from CA3 peri-somatic areas, the SRFPs could be superimposed with small-amplitude oscillatory activities or single-unit activities. As shown in the spectral plots, the frequencies of these oscillatory activities were in the 150- to 200-Hz range (Fig. 4C). Because the recordings of these fast oscillatory activities were position-dependent and because CA3 SRFPs with or without such oscillatory activities share similar intracellular correlates (see following text), we did not concentrate on these small oscillatory activities in the present study.

To assess variability of SRFPs in the frequency behavior over time, we collected continuous extracellular field potentials from a period of 10 min and examined these data using time-frequency analysis techniques. One example of such analyses is shown in Fig. 4D where spectrograms were generated from CA3 and CA1 extracellular records. Although the rhythmic signals exhibit some nonstationary character, a central band of activity over time (and harmonics) that corresponds to the SRFPs is evident in the spectrums, particularly for the CA1 recording (Fig. 4D). The CA3 signal exhibits transient (<0.1 s) high-frequency (>30 Hz) activities that are not evident in the CA1 signal (Fig. 4D). Similar activity patterns were obtained in five other thick slices examined. These preliminary analyses suggest that in addition to generating relatively stable SRFPs, the CA3 network of the thick slices produce other rhythmic activities at higher frequencies. This is likely via the interactions of CA3 recurrent circuitry and the targeted GABAergic interneurons. Further characterizations of the SRFPs and its associated rhythmic activities in the CA3 area, together with multi-channel and single-cell data, may help to reveal the network mechanisms of SRFPs.

Regional initiation and spread of SRFPs

We made simultaneous extracellular recordings from the dentate gyrus, CA3 and CA1 areas to assess the regional initiation and spread of SRFPs in the thick slices. In a set of four slices in which coherent SRFPs were recorded from these regions, CA3 SRFPs led dentate gyrus SRFPs by 14.5 ± 1.3 ms as measured from peak-to-peak time intervals from a data segment of 30 s for each slice. After a surgical cut that separated the CA3 and dentate gyrus recording sites, SRFPs remained coherent in the CA3 and CA1 areas, but absent in isolated dentate gyrus area as monitored up 40 min after the cut (Fig. 5A).



View larger version (37K):
[in this window]
[in a new window]
 
FIG. 5. Regional spread and pharmacological properties of SRFPs. A: extracellular recordings were made simultaneously from the dentate gyrus (DG), CA3 and CA1 areas of a thick slice. Representative traces were collected before and ~35 min after a surgical cut that separated the dentate gyrus and CA3 recording sites. The dentate gyrus and CA3 recording electrodes were re-positioned after the cut for convenience purpose. B: extracellular recordings were made from the CA3 and CA1 areas of another thick slice. Representative traces at left and middle panels were collected before and ~30 min after a surgical cut that separated these 2 recording sites. Traces at right were collected after perfusion of the cut slice with a high-K+ (15 mM) ACSF for ~8 min. C: CA3 extracellular potentials were recorded from a thick slice in the absence or presence of 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX; 5 µM), 2-amino-5-phosphonopentanoic acid (100 µM), or bicuculline (5 µM). Each agent was superfused for 6–8 min and washed for 10–25 min.

 
While SRFPs were simultaneously monitored from the CA3 and CA1 areas, the CA3 SRFPs led the CA1 SRFPs by 12.9 ± 2.6 ms (peak-to-peak time intervals, n = 14 slices), although a clear time separation between the CA3 and CA1 SRFPs was not always noticeable. After a surgical cut that disconnected the CA3 and CA1 recording sites, SRFPs remained in the CA3 but not in the isolated CA1 area (n = 4 slices, Fig. 5B). The loss of CA1 rhythmicity could not be simply attributed to the tissue damage associated with the surgical cut because the isolated CA1 area could elicit synaptic field potentials after local afferent stimulation or exhibited some rhythmic potentials after exposure to a high-K+ perfusate (15 mM KCl, n = 3, Fig. 5B). Collectively, these present observations are consistent with our previous study in the whole hippocampal isolate (Wu et al. 2002Go), further indicating that the SRFPs largely originate from the CA3 area and that the activities of dentate gyrus neurons are not necessary for the generation of SRFPs in vitro.

Pharmacological responses of SRFPs

Perfusion of thick slices with 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX; 5 µM, 6–8 min), but not the N-methyl-D-aspartate (NMDA) receptor antagonist 2-amino-5-phosphonopentanoic acid (100 µM for ~8 min), reversibly blocked the CA3 SRFPs (n = 5 slices). SRFPs were reversibly abolished by applications of bicuculline (2–5 µM, 6–8 min, n = 6 slices) but not by the GABAB receptor antagonist CGP55845(10 µM, 8–10 min, n = 4). An example of such pharmacological manipulations is shown in Fig. 5C.

SRFP-correlated activities in CA3 pyramidal neurons

Simultaneous extracellular and single-cell recordings were used to explore the cellular correlates of SRFPs. Of the 51 stable recordings made from CA3 pyramidal neurons, 19 neurons showed sporadic spiking currents (cell-attached voltage-clamp recordings) or action potentials (whole cell current-clamp recording) at resting potentials. However, these spike activities did not exhibit a clear temporal correlation with the local SRFPs. Repeated discharges of individual CA3 pyramidal neurons via intracellular injection of depolarizing current pulses (~0.5–2 s) did not induce detectable extracellular response or interrupt the ongoing local SRFPs.

In correlation with local SRFPs, CA3 pyramidal neurons exhibited periodic IPSPs, EPSPs, or mixed EPSP-IPSPs as monitored at the resting membrane potentials (ranging from –55 to –76 mV), and the IPSPs were the main component as these neurons were depolarized. We measured the reversal potentials of the SRFP-correlated synaptic currents via perforated voltage-clamp recordings, using a patch pipette solution containing gramicidin and 150 mM KCl (Wu et al. 2002Go). The gramicidin-perforation is thought to be impermeable to Cl (Rhee et al. 1994Go) thus minimally affecting intracellular Cl homeostasis of the recorded neurons. By holding the recorded neurons at different membrane potentials, we estimated that the reversal potentials for the SRFP-correlated synaptic currents were –66.3 ± 2.5 mV (n = 4, Fig. 6A). After the perforated measurement, we ruptured the membrane and dialyzed the neurons with the high-Cl patch pipette solution. The reversal potentials of SRFP-correlated synaptic currents were shifted positively by 30–35 mV after a few minutes of whole cell dialyses (n = 2 neurons, Fig. 6A). Collectively, these observations indicate that the SRFPs are correlated largely with Cl-dependent currents in CA3 pyramidal neurons (Zhang et al. 1991Go).



View larger version (40K):
[in this window]
[in a new window]
 
FIG. 6. SRFP-correlated activities in CA3 pyramidal neurons. Data were collected from 3 thick slices via simultaneous extracellular and patch recordings from the CA3 area. A: a CA3 pyramidal neuron was recorded initially via the perforated recording configuration (top) and then in the whole cell configuration after breakthrough the membrane (bottom). The patch pipette solution contained 150 mM KCl, 2 mM HEPES, 0.1 mM EGTA, and gramicidin (≤50 µg/ml). Correlated local SRFPs are not shown. The amplitudes of SRFP-correlated synaptic currents were plotted vs. holding voltages at right. Data (means ± SE) points at each voltage were calculated from 15 to 20 events. B: 2 CA3 pyramidal neurons were recorded in the whole cell voltage-clamped configuration (holding potentials of –50 mV), showing synchronous outward synaptic currents in correlation with local SRFPs. Representative events (->) are shown in fast sweeps. Cross-correlation plots were generated from 5-s data segments including the traces illustrated and presented at 2 time scales. Original data were treated with a band-pass filter (0.1–285 Hz) before the analyses. C: whole cell current-clamp recordings were made from other 2 CA3 pyramidal neurons. Representative traces show synchronous IPSPs and their blockade after perfusion of bicuculline (10 µM, ~6 min).

 
To assess the kinetics of SRFP-correlated synaptic currents, we voltage-clamped CA3 pyramidal neurons at –50 mV (n = 13) to sample the IPSC-dominated intracellular events. These IPSCs were then recognized off-line by the event detection function of Pclamp software (56–232 IPSCs per neuron collected from a stable recording period of 1–2 min; see METHODS) and averaged. The basic parameters of these IPSCs were: peak amplitude 124.9 ± 12.1 pA, rising time (baseline to peak) 16.9 ± 1.7 ms, and decay time constant 21.2 ± 1.6 ms (single exponential). The SRFP-correlated IPSCs were often superimposed with a few small components in their rising phases. In eight CA3 pyramidal neurons in which these small components could be readily recognized in the averaged traces, the time interval (peak to peak) between the first and second small components was 5.2 ± 0.4 ms (Fig. 6B). These observations suggest that CA3 pyramidal neurons might receive stable and coherent GABAergic inputs during the SRFPs.

The time interval between CA3 pyramidal neuronal IPSPs or IPSCs and local SRFP was 7.1 ± 1.6 ms (peak to peak), as measured from a data segment of 30 s per paired extracellular and patch recordings (n = 25 pairs). This time interval is quite small considering that the SRFP cycle was 250–2000 ms in duration and the two recording sites were ≥0.3 mm apart. Dual whole cell recordings revealed that CA3 pyramidal neurons displayed synchronous IPSPs or IPSCs (peak-to-peak time intervals of ≤5 ms, n = 4 pairs) in phase with local SRFPs. Collectively, these observations suggest that in the CA3 area, the SRFPs represent summed IPSPs originating from a large number of CA3 pyramidal neurons.

Perfusion of thick slices with bicuculline (5–10 µM for 6–8 min) reversibly abolished the SRFP-correlated IPSPs or IPSCs in all CA3 pyramidal neurons examined (n = 6). Frequent EPSPs or EPSCs occurred in clusters in these neurons during the bicuculline application, but they did not correlate with evident extracellular potential (Fig. 6C). Large field potentials (1–3 mV in amplitude and 300–600 ms in duration) superimposed with multiple spikes were found to occur spontaneously (1–2 events/min) during washing bicuculline and before the return of SRFPs. In correlation with these large field potentials, CA3 pyramidal neurons displayed prolonged EPSPs together with repeated discharges (data not shown).

SRFP-correlated activities in CA3 inhibitory interneurons

In 15 putative inhibitory interneurons recorded from CA3 peri-somatic areas, 12 interneurons exhibited periodic spikes or EPSPs/spikes in correlation with local SRFPs. The interneuronal discharges were recognizable in the cell-attached voltage-clamp recordings as uniform inward currents that represented not fully clamped action potentials. One example of this is shown Fig. 7A, where the recorded interneuron consistently exhibited two to three spike currents during the rising phase of local SRFPs. We measured the timing of spike currents relative to the peak of corresponding SRFP (time 0) and obtained mean values of spike timings over 30 consecutive SRFPs. A relatively stable firing pattern of this interneuron during the onset phase of local SRFPs was noticeable by plotting the spike timings versus the peak (time 0) of averaged SRFP (Fig. 7A). SRFP-correlated interneuronal discharges appeared to be more variable when monitored in the whole cell recordings. Nevertheless, rhythmic discharges of CA3 interneurons generally remained a close temporal relation with local SRFPs or the SRFP-correlated IPSPs in pyramidal neurons (Fig. 7B). In six CA3 interneurons that consistently exhibited two to three spikes or spike currents during the onset phase of local SRFPs, the mean time interval between the first and second interneuronal spike was 13.7 ± 0.1 ms (ranging form 4.6 ± 0.7 to 16.2 ± 2.8 ms as measured from 20 to 30 consecutive SRFP events). The time interval between the second interneuronal spike and the peak of local SRFPs was 15.9 ± 7.0 ms (ranging from 4.3 ± 0.5 to 17.1 ± 1.5 ms). Perfusions of slices with CNQX (5–10 µM, 6–8 min) completely abolished SRFP-correlated EPSPs/discharges in CA3 interneurons (n = 3).



View larger version (29K):
[in this window]
[in a new window]
 
FIG. 7. SRFP-correlated activities in CA3 interneurons. Data were collected from 2 thick slices via simultaneous extracellular and patch recordings. A: a CA3 interneuron was recorded via the cell-attached voltage-clamp configurations at the resting potential (0 holding voltage). Representative traces at top show repeated interneuron (IN) spike currents in correlation with local SRFPs. Superimposed interneuronal spike currents below (left) were collected from 30 consecutive SRFP-correlated events (n = 100, amplitude of 127.5 ± 0.01 pA and half-width of 0.16 ± 0.01 ms, mean ± SD). Right: trace was averaged from 30 consecutive SRFPs. The short bars below represent the timings (mean ± SE) of interneuronal spike currents in relation to the SRFP peak (time 0, · · · ). B: representative traces were collected via whole cell current-clamp recordings from a CA3 interneuron and nearby CA3 pyramidal neuron. The pyramidal neuron was depolarized from –58 mV to –45 mV to illustrate the SRFP-correlated IPSPs.

 
Collectively, these limited observations are in keeping with the hypothesis that the SRFPs represent an IPSP-based rhythm and result from networked activities of interneurons. However, further studies examining the coherent activities of interneurons in the same or different CA3 subregions are necessary to delineate the generation mechanisms of the SRFPs.

Rhythmic field potentials observed from hippocampal slices of 0.5 mm thickness

In parallel to our study in the mouse hippocampal isolate (Wu et al. 2002Go), several studies have reported the presence of spontaneous field rhythms of 2–4 Hz in conventional hippocampal slices of adult rats (Colgin et al. 2004Go; Kubota et al. 2003Go; Papatheodoropoulos and Kostopoulos 2002Go) and mice (Maier et al. 2002Go, 2003Go). These spontaneous field rhythms share several common features with the SRFPs, i.e., they are of CA3 origin, abolished by the pharmacological antagonism of AMPA glutamate receptors or GABAA receptors and correlated largely with GABAA IPSPs in CA1 pyramidal neurons (Maier et al. 2003Go; Papatheodoropoulos and Kostopoulos 2002Go). However, these spontaneous rhythms appear to occur only in horizontal brain slices that contain ventral hippocampal circuitry (Kubota et al. 2003Go; Maier et al. 2002Go, 2003Go). By preparing transverse slices (0.55 mm thickness) from isolated rat hippocampus, Papatheodoropoulos and Kostopoulos (2002)Go have reported that these spontaneous field rhythms occur in ~70% of ventral hippocampal slices but they are absent in the dorsal hippocampal slices. Colgin et al. (2004)Go have reported the lack of spontaneous field rhythm in conventional mouse dorsal hippocampal slices (0.35-mm thickness). The reasons for the lack of spontaneous field rhythms in the conventional dorsal hippocampal slices are not clear at present. However, the ventral hippocampal circuitry is thought to be more compact than the dorsal hippocampal circuitry (cf. Borck and Jefferys 1999Go; Papatheodoropoulos and Kostopoulos 2002Go). Thus insufficient network connectivity in the dorsal hippocampal slices may explain their inability to produce spontaneous field activities.

We conducted a set of experiments to further compare field rhythmicity in conventional ventral and dorsal mouse hippocampal slices. We prepared transverse slices of 0.5-mm thickness from dorsal and ventral hippocampus of 3-mo-old mice. The presence or absence of spontaneous field activities in these slices was determined by visual inspection of extracellular traces together with spectral analyses (see METHODS). In a set of 26 dorsal hippocampal slices examined, extracellular recordings from the CA3/CA1 areas revealed no persistent spontaneous field rhythm, in keeping with the study in rat and mouse dorsal hippocampal slices (Colgin et al. 2004Go; Papatheodoropoulos and Kostopoulos 2002Go). In contrast, spontaneously occurring field potentials were detectable in 30 of 73 ventral mouse hippocampal slices examined. The observed rhythms were reversibly blocked following applications of the GABAA receptor antagonist bicuculline (2 µM) or the AMPA glutamate receptor antagonist CNQX (5 µM, n = 4 and 5, respectively, Fig. 8A). We term these spontaneous rhythms as SRFPs because their similarity to those we observed from the mouse hippocampal isolate (Wu et al. 2002Go) and the thick slices regarding waveforms, frequencies, and pharmacological responses.



View larger version (29K):
[in this window]
[in a new window]
 
FIG. 8. Presence or absence of SRFPs in conventional mouse hippocampal slices (0.5-mm thickness). A: representative traces were collected from a ventral hippocampal slice showing baseline CA3 SRFPs and their reversible blockade by bicuculline (2 µM, ~5 min). B: extracellular traces were collected from another ventral hippocampal slice. Top trace shows quiescent baseline recording and induced rhythmic field potentials after high-frequency afferent stimulation (80 Hz, 1 s). The square areas are shown in fast sweeps below. C: each spectral plot was generated from a data segment of 30 s with SRFPs and without SRFPs. The data segments include extracellular records shown in A (before bicuculline) and B (before stimulation). The original data were treated with a low-pass filter of 1,000 Hz before spectral analyses. The inserted spectral plot was generated from a 5-s data segment after the high-frequency afferent stimulation including the squared area. D: averaged spectral plots were generated from 2 sets of data segments with or without SRFPs (30 s in each data segment, n = 10 slices in each group). *, P < 0.001, one-way ANOVA. The statistical comparison was made between the signals in the 1- to 10-Hz range.

 
The main frequency of SRFPs in 0.5 mm ventral hippocampal slices was 1.21 ± 0.11 Hz (n = 30). In addition, rhythmic signals of a wide frequency band (~1 to ~50 Hz) were noticeable in the SRFP-exhibiting data segments (Fig. 8, C and D). When averaged spectral plots were compared (Fig. 8D), the overall amplitude of rhythmic signals in the 1- to 10-Hz range was significantly greater in the SRFP-containing extracellular records than those without SRFPs (P < 0.001, one-way ANOVA).

The absence of SRFPs in some 0.5-mm slices could not be attributed to nonspecific tissue damage because local afferent stimulation reliably induced synaptic field potentials in these slices (somatic population spikes of 1.7 ± 0.3 mV and dendritic field EPSPs of 0.9 ± 0.1 mV, n = 17 and 10). In addition, high-frequency stimulation (80 Hz, 1 s) induced some transient, fast rhythmic field potentials (3–30 Hz, 5–10 s, n = 3 slices, Fig. 8B). Moreover, applications of the cholinergic agonist carbachol (10 µM, 6–10 min) induced rhythmic field potentials with mixed frequencies (3–20 Hz, n = 4 slices), and this observation is in keeping with the previous study in conventional rat hippocampal slices (Fellous and Sejnowski 2000Go).

Collectively, our results are supportive of recent studies (Colgin et al. 2004Go; Kubota et al. 2003Go; Maier et al. 2002Go, 2003Go; Papatheodoropoulos and Kostopoulos 2002Go), further suggesting that spontaneous population rhythmic activities can occur in conventional mouse ventral hippocampal slices. However, the occurrence of SRFPs in the slices of 0.5 mm thickness is not consistent in our experiments, i.e., they were detectable in ~41% of the ventral hippocampal slices and absent in the dorsal hippocampal slices. In contrast, we found that SRFPs occurred in nearly every thick slice successfully prepared. Thus a relatively large hippocampal circuitry than that provided by the 0.5 mm slice may allow consistent generation of SRFPs in vitro.


    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Functional implications of using thick mouse hippocampal slices

Morphological studies have shown that the Schaffer collateral pathway and the perforant pathway project to their targeted areas with a widespread distribution pattern along the longitudinal axis of rat hippocampus (Dolorfo and Amaral 1998Go; Ishizuka et al. 1990Go; Li et al. 1994Go). In addition, GABAergic interneurons have extensive axon and dendrite arbors traveling longitudinally in the hippocampus (Freund and Buzsáki 1996Go). To preserve such extensive hippocampal connectivity in vitro, Khalilov et al. (1997)Go have established the entire neonatal rat hippocampal model, and we have produced a hippocampal isolate preparation from young mice (Wu et al. 2002Go). However, these models cannot be applied to adult rodent hippocampus because of insufficient oxygenation delivery to densely packed mature hippocampal tissue during in vitro perfusion. We thus developed the thick slice preparation as an alternative approach. The thick slice contains only a small portion (~15%) of the adult mouse hippocampus (~8 mm in length), but its thickness is double that of the conventional slice. Considering that the areas with the highest density of recurrent axons of CA3 pyramidal neurons are always several hundred micrometers anterior or posterior to the cell body (Li et al. 1994Go), one major advantage of the thick slice over the conventional slice lies in the preservation of more CA3 recurrent axons. The CA3 circuitry plays crucial roles in the generation of synchr