| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1 Douglas Hospital Research Center, Department of Psychiatry, McGill University, 6875, Lasalle Boulevard, Montreal, Quebec, Canada H4H 1R3
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
(Received 1 May 2005;
accepted after revision 25 May 2005;
first published online 26 May 2005)
Corresponding author S. Williams: Douglas Hospital Research Center, Department of Psychiatry, McGill University, 6875, Lasalle Boulevard, Montreal, Quebec, Canada H4H 1R3. Email: wilsyl{at}douglas.mcgill.ca
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
It is interesting that there is increasing evidence that a third population of neurones is present in the MS/DB. It has been acknowledged for some time that a portion of MS/DB neurones cannot be identified with cholinergic and GABAergic markers and could in fact be glutamatergic (Kiss et al. 1997; Gritti et al. 1997). Indirect evidence for glutamate neurones in the MS/DB was obtained using phosphate-activated glutaminase (the enzyme thought to be involved in the synthesis of the neurotransmitter glutamate) or glutamate itself, and by the retrograde transport of [3H]aspartate in septal neurones (Gonzalo-Ruiz & Morte, 2000; Manns et al. 2001; Kiss et al. 2002). We have recently provided strong evidence for the presence of glutamate neurones in the MS/DB using single-cell multiplex RT-PCR by identifying a subpopulation of septo-hippocampal neurones that express mRNAs for the vesicular glutamate transporters type 1 and/or 2 (VGLUT1/2) (Sotty et al. 2003; Danik et al. 2003, 2005). Together with the vesicular glutamate transporters type 3 (VGLUT3), these transporters are presumed to be contained in neurones that have the capacity to use glutamate as neurotransmitter (Moriyama & Yamamoto, 2004). In agreement with our results, the expression of VGLUT2 transcripts was also reported by other groups (Hisano et al. 2000; Fremeau et al. 2001; Lin et al. 2003) whereas the presence of the VGLUT2 protein in the cell body of numerous neurones of the MS/DB was demonstrated in colchicine-treated rats (Hajszan et al. 2004).
These glutamatergic neurones could potentially play an important physiological role in the activity of the MS/DB and septo-hippocampal pathway as earlier studies have shown that intraseptal injection of glutamate receptor antagonists produces significant reduction in hippocampal learning and memory (Izquierdo et al. 1992; Izquierdo, 1994). However, whether glutamatergic neurones of the MS/DB can provide synaptic input within this area remains to be determined. Whereas electrically evoked or spontaneous EPSPs (Segal, 1986; Armstrong & MacVicar, 2001) are known to be present in the MS/DB, it is not known whether these responses originate from local MS/DB glutamatergic neurones and/or from the significant glutamatergic input coming from outside the MS/DB (Leranth & Kiss, 1996; Leranth et al. 1999; Jaskiw et al. 1991; Kiss et al. 2000; Bokor et al. 2002). In this study, we have investigated whether MS/DB glutamatergic neurones can mediate local excitation to GABAergic and cholinergic neurones. The present experiments were performed using a variety of preparations including cultured organotypic MS/DB slices devoid of extrinsic glutamatergic afferents in order to investigate exclusively local synaptic input from MS/DB glutamatergic neurones. Moreover, using an intact preparation of the septum in vitro, we have examined if MS/DB glutamatergic neurones were connected and could be activated synchronously, a property potentially important for pacemaker activity and oscillations. We present novel evidence that glutamatergic neurones within the MS/DB can generate powerful excitatory inputs to cholinergic and GABAergic neurones, are organized into a local excitatory network and can be activated by cholinergic agonists. These findings suggest that intrinsic glutamatergic neurones may play an important role in the MS/DB, and subsequently by modulating septo-hippocampal neurones, in hippocampal activity.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
All experiments were performed using Sprague-Dawley rats (Charles River, St-Constant, Quebec, Canada). Animal care was provided according to protocols and guidelines approved by the McGill University Animal Care Committee and the Canadian Council on Animal Care. Bicuculline, 4-aminopyridine (4-AP), carbachol, glucose and all inorganic salts were obtained from Sigma (Oakville, Ontario, Canada). 6,7-Dinitroquinoxaline-2,3-dione (DNQX) and DL-2-amino-5-phophonopentanoic acid (DL-APV) were from Tocris Cookson (Bristol, UK). All drugs were diluted in artificial cerebrospinal fluid (ACSF) from previously prepared stock solutions that were prepared in deionized H2O and stored at –20°C. All drugs were delivered by bath application.
Acute MS/DB slice preparation
Rats (8–17 days (d) old) were killed by decapitation and the brain was rapidly removed and placed in ice-cold ACSF, pH 7.4, equilibrated with 95% O2–5% CO2 and containing (mM): NaCl 126, NaHCO3 24, glucose 10, KCl 3, MgSO4 2, NaH2PO4 1.25, ascorbic acid 0.4, thiourea 0.8 and CaCl2 1.2. Coronal slices (350 µm thick) containing the MS/DB were cut with a Vibroslice (Campden Instruments Ltd, UK) and placed in a Petri dish containing oxygenated ACSF. Slices were further trimmed to include either the MS/DB and lateral septum (whole-septum slice) or the MS/DB alone (MS/DB mini-slice). After 1–1.5 h, single slices were transferred to a Plexiglas recording chamber on the stage of a Nikon microscope (Eclipse E600FN) equipped with Nomarski optics, and continuously perfused with oxygenated ACSF (containing 2 mM CaCl2; room temperature, 
24°C) at a rate of 1–2 ml min–1 for electrophysiological recordings. Cells were visualized using a 40 x water immersion objective.
MS/DB organotypic slices
Coronal 350 µm-thick slices of septum were cut as above from the brains of 6- to 9-day-old rats in Hanks' buffered salt solution (HBSS) supplemented with 25 mM Hepes (both from Invitrogen, Burlington, Ontario, Canada) and 25 mM glucose. The MS/DB area of the septum was dissected out with a scalpel blade and grown on patches of polytetrafluoroethylene membrane placed in Millicell-CM culture inserts (both from Millipore, Ontario, Canada) according to the interface culture method (Stoppini et al. 1991). The culture medium (50% minimal essential medium, 25% HBSS, 25 mM Hepes, 25% horse serum (all from Invitrogen) and 25 mM glucose) was replaced 24 h after dissection and every 2 days thereafter. For electrophysiological experiments, slices cultured for 4–6 days in vitro were transferred to the recording chamber and continuously perfused with oxygenated ACSF as described above. For immunohistochemistry, we used 15 slices at 4 days in vitro. Colchicine was used to disrupt axoplasmic transport in order to visualize the perikarya of VGLUT2-containing neurones (Kreutzberg, 1969). Accordingly, one-third of the slices were exposed to 10 µM colchicine (Sigma) for 1 h followed by a medium change. After 24 h (5 days in vitro), all slices were rinsed briefly in phosphate-buffered saline (PBS; pH 7.3) and processed for immunohistochemistry.
Intact half-septum preparation
The brain from 8- to 17-day-old rats was removed and dropped into a pre-cooled high-sucrose modified ACSF solution (mACSF; Jones et al. 1999) with the following composition (mM): sucrose 252, KCl 3, MgSO4 2, NaHCO3 24, NaH2PO4 1.25, CaCl2 1.2 and glucose 10; pH to 7.4 with 95% O2–5% CO2. The cerebellum and frontal part of the brain were removed and a sagital cut was made through the interhemispheric sulcus to separate the two hemispheres (method adapted from Khalilov et al. 1997). The half-septum and hippocampus were carefully dissected out from each hemisphere by inserting a flat spatula in the lateral ventricle and sliding along both dorsal and ventral surfaces of hippocampus and septum. The half-septa were separated from the hippocampus by cutting the septo-hippocampal fibres with a scalpel. Each half-septum was left to rest in high-sucrose solution (room temperature) for 1–3 h prior to the start of experiments. For electrophysiology, single half-septa were transferred to a fully submerged recording chamber (same as for slices), weighted on a nylon mesh and perfused with oxygenated ACSF (3–4 ml min–1). A 30-min rest was allowed before recordings started. Placement of the half-septum was such that the MS/DB was at the surface of the preparation (see Fig. 3A) thereby allowing direct access for the recording electrodes. A 10 x objective was used to visualize the positioning of electrodes in this preparation.
|
Patch pipettes (3–6 M
) were pulled from borosilicate glass capillary tubing (Warner Instrument Corp., Hamden, CT, USA) and filled with internal solution containing (mM): potassium gluconate 144, MgCl2 3, EGTA 0.2, Hepes 10, ATP 2 and GTP 0.3; pH 7.2 (285–295 mosmol l–1). Whole-cell recordings in voltage-clamp and current-clamp modes were performed at room temperature using a whole-cell patch-clamp amplifier PC-505 A (Warner Instrument Corp.). The resting membrane potential was measured in current-clamp mode once a stable recording was obtained. Cells were only used for experiment when the resting membrane potential was more negative than –45 mV, spikes overshot 0 mV and the series resistance was less than 30 M. Cells recorded in the various experimental preparations used in this study appeared equally healthy based on these criteria. The presence of a hyperpolarization-activated cation current (Ih) or a depolarizing sag was monitored using hyperpolarizing steps from –60 mV to –95 mV. The firing frequency was estimated using a series of depolarizing steps. The membrane potential (Vm) output signal was filtered on-line (0.1–0.2 kHz) and acquisition speed was set between 2 and 10 kHz in the pClamp 8.0.1 software (Axon Instruments, Union City, CA, USA). Whole-cell recordings in the intact half-septum were performed as noted above for slices except that the whole-cell configuration was established blindly by descending the recording electrode into the tissue (Blanton et al. 1989; Margrie et al. 2002). For this technique to work well, high positive pressure had to be applied to the back of the pipette. Field potentials were recorded conventionally with glass micropipettes (2 M, filled with normal ACSF) using the 100 x
Vm output of an AM-system 3100 (Everett, WA, USA) or a Dagan BVC-700 A (Minneapolis, MN, USA) amplifier equipped with a low-cut filter set at 0.3 Hz. Extracellular recordings were low-pass filtered on-line at 1 kHz and traces shown were additionally filtered off-line for clarity.
Measurement of spontaneous synaptic activity and bursts
MS/DB cells were patched, briefly characterized and held at a hyperpolarized membrane potential (usually –80 or –70 mV) to monitor spontaneous synaptic activity. Slices were perfused with mACSF solution that was Mg2+-free and that contained the pro-convulsant agent 4-AP (50 µM) and the GABAA receptor antagonist bicuculline (10 µM). This combination of solutions was chosen because, neither bicuculline nor 4-AP alone (with or without Mg2+) triggered reproducible bursts in the preparations. Although most recorded cells showed increased spontaneous synaptic activity in this solution, we concentrated our analysis on spontaneous excitatory bursts. A burst was defined as a period of sustained depolarization originating with rapid onset from baseline (–80 mV), remaining above a preset threshold (three times the baseline noise) for 400 ms and having a peak amplitude of 15 mV. These criteria for burst analysis were used mainly to facilitate the comparison of spontaneous synaptic activity between the different conditions. Burst duration was measured as the time during which the membrane potential was above baseline and burst amplitude was measured at the peak of synaptic potentials when possible or just before the first action potential when intense firing prevented adequate measurement (Fig. 1A). We found that holding the membrane at a hyperpolarized potential facilitated measurement of burst amplitude. All bursts were measured from a holding potential of –80 mV for slices and –70 mV for intact half-septum preparation.
|
Cultured MS/DB slices were fixed for 1 h at room temperature with PBS containing 4% paraformaldehyde followed by several rinses in PBS. Slices were then treated for 1 h with PBS containing 0.3% Triton X-100. Blocking was performed with 5% bovine serum albumin (BSA; Jackson Immunoresearch, West Grove, PA, USA), 5% normal horse serum (Vector Laboratories, Burlingame, CA, USA), and 0.1% Triton X-100 in PBS for 1 h. The slices were incubated overnight at 4°C in the presence of primary antibodies diluted in blocking buffer (with BSA diluted to 0.5%). Final dilutions for antibodies were 1: 1000 for polyclonal rabbit anti-VGLUT2 (Synaptic Systems, Goettingen, Germany), 1: 2500 for polyclonal guinea-pig anti-VGLUT2, 1: 2500 for monoclonal mouse IgG2a anti- glutamic acid decarboxylase 67 (GAD67), 1: 500 for monoclonal mouse IgG1 anti-synaptophysin (Chemicon International; Temecula, CA, USA) and 1: 1000 for monoclonal mouse anti-choline acetyltransferase (ChAT) kindly provided by Dr L. Descarries (Université de Montréal, Quebec, Canada; Cossette et al. 1993; Chedotal et al. 1994). After extensive washing in PBS, the slices were incubated overnight at 4°C in the same antibody diluent buffer containing the appropriate species-specific fluorophore-conjugated secondary antibodies diluted 1: 400: Alexa Fluor 568 goat anti-rabbit or Alexa Fluor 488 goat anti-guinea-pig IgG (VGLUT2), Alexa Fluor 568 goat anti-mouse IgG2a (GAD67) (Invitrogen), and Cy3 donkey anti-mouse IgG (ChAT and synaptophysin) (Jackson Immunoresearch). Following the final washes in PBS, the slices were briefly rinsed in water then mounted on slides in Vectashield medium (Vector Laboratories) and subsequently examined with a Nikon fluorescence microscope and a Nikon PCM2000 confocal microscope system. Z-series of 16–51 optical sections were scanned and carried out sequentially (for dual labelling) with the 488 nm and 543 nm lines of the laser to reveal Alexa Fluor 488 and Alexa Fluor 568 or Cy3, respectively. Images were acquired either at 0.6 µm, 0.2 µm or 0.14 µm intervals with a 20 x, 60 x and 100 x objective lens, respectively. Images were adjusted for contrast only using the Simple PCI Program Rev 3.5 (Compix Inc., C-Imaging Systems, Cranberry Township, PA, USA).
Data analysis
All electrophysiological data were analysed using pCLAMP 9.0 sofware (Axon Instruments). To compare bursting activity in MS/DB cells between different conditions, we used 1-min recording episodes. All results are expressed as mean ± S.E.M. For statistical analyses we performed paired or unpaired Student's t test (two-tailed) or ANOVA using Prism (GraphPad Software, Inc., San Diego, CA, USA). P < 0.05 was considered to be statistically significant. Figures were compiled using pCLAMP or Origin (5.0, Microcal Software, Northampton, MA) for plotting electrophysiological data and analyses.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In the following experiments, we examined whether glutamatergic neurones within the MS/DB could be activated (Fig. 1) and then further investigated whether glutamatergic synaptic terminals originating from outside this nucleus contributed to the glutamatergic activity (Fig. 2). As an electrical stimulation of afferent fibres would activate both cut glutamatergic synaptic fibres originating from outside and those from glutamatergic neurones within the MS/DB, intrinsic glutamatergic neurones were activated by increasing the overall excitability of the slices. Moreover, we hypothesized that if MS/DB glutamatergic neurones were interconnected, increasing the overall excitability of MS/DB slices would favour the emergence of large and sustained synaptically mediated spontaneous depolarizations reflecting the synchronized discharge of several glutamatergic neurones. In contrast, synaptically mediated glutamate responses originating from cut excitatory afferents or isolated glutamatergic neurones would be small and transient (Barbarosie & Avoli, 1997; Avoli et al. 2002; Dzhala & Staley, 2003). To increase neuronal excitability of MS/DB slices, we used mACSF solution containing 4-AP (50 µM), the GABAA receptor antagonist bicuculline (10 µM), and no Mg2+. MS/DB neurones exhibited only sparse spontaneous synaptic activity in normal ACSF but perfusion of MS/DB slices with the mACSF elicited a gradual increase in spontaneous EPSPs over the first 20 min. Because of the slow development of the spontaneous activity, most experiments were then initiated using slices that had already been perfused in modified ACSF for more than 30 min. We found that such treatment induced large bursting activity in recorded MS/DB neurones. Figure 1A shows an example from a cell demonstrating large and long-lasting bursts occurring at regular intervals of 1–3 s. In 19/62 neurones, spontaneous bursts recorded from a holding potential of –80 mV consisted of sharp-onset prolonged depolarizations (400 ms, 15 mV; see Methods) that often triggered action potential firing (in 13/19 cells with bursts). These bursts had a mean duration of 838.3 ± 70.3 ms (range, 405–2263 ms), a mean amplitude of 22 ± 1.8 mV, and a mean number of bursts per min of 5.1 ± 1.05. Although bursts occurring at relatively regular intervals (such as in Fig. 1A) were observed in rare occasions, most bursting cells did not display any apparent pattern in acute slices. The spontaneous bursts of MS/DB neurones were not dependent on intrinsic membrane properties, rather they were synaptically mediated because: (1) their frequency and duration did not change at different levels of membrane potentials (Fig. 1A); (2) they were associated with the production of large inward currents (EPSCs) in voltage clamp (Fig. 1B); and (3) burst-associated EPSCs had a reversal potential near 0 mV (n = 6), a value typical of AMPA receptor-dependent EPSPs in the MS/DB (Schneggenburger et al. 1992; Armstrong & MacVicar, 2001). To verify that spontaneous bursts were mediated by presynaptic release of glutamate, we bath applied the selective AMPA/kainate receptor antagonist DNQX (20 µM). In 5 out of 5 cells, the mACSF-induced burst activity and burst-associated currents were completely abolished by DNQX (Fig. 1C and D). In one other cell (not shown), the burst activity persisted 15 min after DNQX perfusion (although the duration of bursts was dramatically reduced) but was stopped in a reversible manner after addition of the NMDA receptor antagonist D-aminophosphonovalerate (APV). Taken together, these results suggest that the spontaneous bursts are mediated by the release of glutamate and mostly involve postsynaptic AMPA receptors, and to a lesser extent NMDA receptors.
|
We investigated whether neurones from the lateral septum, a region suspected to contain glutamatergic projecting neurones (Kiss et al. 2002; Kocsis et al. 2003; Lin et al. 2003; Hajszan et al. 2004), might have contributed to the synaptically mediated glutamate responses that were observed in the MS/DB. We therefore challenged slices with mACSF that contained only the MS/DB (mini-slices; see Methods and Fig. 2A) without the lateral septum. We found that, in comparison to regular lateral septum-containing slices, a similar proportion of neurones from mini-slices responded to the mACSF with spontaneous burst (16/53 cells, 30.2%). Moreover, these bursts were also comparable in terms of duration (760 ± 68.3 ms), amplitude (21.1 ± 1.5 mV) and frequency (5.6 ± 1.2 bursts min–1) (P > 0.05; unpaired t test) and they were similarly blocked by DNQX (data not shown). This indicates that glutamatergic neurones within the MS/DB are sufficient to sustain this bursting and that the lateral septum region does not further contribute to the generation of spontaneous glutamatergic bursts in these experimental conditions.
We next examined whether the glutamatergic inputs that are known to originate from outside the septum (Jaskiw et al. 1991; Leranth & Kiss, 1996; Leranth et al. 1999; Kiss et al. 2000; Bokor et al. 2002) contributed to the spontaneous EPSPs observed in the acute slices challenged with the mACSF. Excitatory terminals remaining in the slice can potentially sustain repetitive synchronous discharge when activated with 4-AP (Strowbridge, 1999). We therefore eliminated glutamatergic synaptic afferents from outside by using an organotypic slice culture preparation of the MS/DB (Fig. 2B). In organotypic slices, all extrinsic input degenerates over the first few days in culture leaving only synaptic contacts from neurones within the slice (Buchs et al. 1993; Gahwiler et al. 1997). Similar to the neurones recorded in acute slices, those recorded from organotypic slices displayed healthy membrane characteristics in normal ACSF. Interestingly, the addition of bicuculline alone in MS/DB organotypic slices induced large spontaneous excitatory bursts within the first 5 min of perfusion (Fig. 2C) in 7 out of 8 recorded neurones. The bursts had an average duration of 5034 ± 1035 ms (range, 1904–8837 ms), an average amplitude of 25.8 ± 2.9 mV and the mean number of burst per min was 4.0 ± 0.9. These bursts were rapidly blocked (n = 6/6 neurones) by superfusion of the AMPA receptor antagonist DNQX (20 µM) (Fig. 2C). Taken together these results demonstrate that the excitatory bursts observed in disinhibited MS/DB slices are generated by local glutamatergic neurones residing within the MS/DB while neurones of the lateral septum or excitatory afferent fibres from outside the septum do not contribute to these responses.
Network activity of MS/DB glutamatergic neurones in the intact half-septum in vitro
Excitatory interconnections between glutamatergic neurones form the basis for the generation and synchronization of some network rhythms in regions such as the CA3 of the hippocampus (Traub et al. 1999). Interestingly, we have observed that a small proportion of neurones from mACSF-bathed acute MS/DB slices showed very slow and large repetitive activity (as in Fig. 1A), suggesting that local glutamatergic neurones could be organized into a network of synchronously discharging neurones. We hypothesized that this putative network activity occurred in only a few slices because the synaptic connections necessary for such activity were damaged by the slicing procedure and that a more intact preparation would provide a more prominent and functional network. We therefore examined whether glutamate-mediated slow repetitive activity would be more frequent and robust when recorded in an intact half-septum preparation in vitro (see Methods and Fig. 3A). Using the whole-cell blind patch-clamp technique to record the activity of single cells in mACSF-perfused half-septum, we observed large amplitude, regular and sustained bursting activity in a significantly larger proportion of recorded neurones (13 of 26 cells, 50%) in this preparation than in the acute slices. The duration of bursts was also significantly larger (2352 ± 383.4 ms; range, 423–8097 ms; P < 0.0001; one-way ANOVA; whole-slice, mini-slice, intact half-septum), and in contrast to acute slices, bursts occurred with a stable repetitive pattern (i.e. at regular intervals, similar to Fig. 3B and C) in most cells recorded from in the intact half-septa. Overall, the proportion of preparations that were recruited into burst activity in mACSF was larger for the half-septum experiments than for acute slices (Table 1). As these results suggested that the integrity of the glutamatergic network was better preserved in this type of preparation, we went on to perform extracellular recordings that might pick up the activity of simultaneously discharging neurones.
|
Glutamatergic neurones of the MS/DB activate cholinergic and non-cholinergic neurones
It remains to be shown that cholinergic, GABAergic and glutamatergic neurones of the MS/DB receive AMPA/kainate receptor-mediated synaptic input form local glutamatergic neurones of the MS/DB. Using electrophysiological characterization of MS/DB neurones, we first investigated whether the different neuronal types of the MS/DB received synaptic inputs from local glutamatergic neurones. We used previously reported electrophysiological characteristics to identify MS/DB neurones, namely the lack of expression of an Ih and a slow firing frequency in cholinergic neurones versus larger Ih and faster firing rates in GABAergic neurones (Griffith, 1988; Gorelova & Reiner, 1996; Serafin et al. 1996; Jones et al. 1999; Sotty et al. 2003). We also investigated whether cluster-firing neurones, previously suggested to be glutamatergic in nature (Sotty et al. 2003), received glutamatergic input. Our results show that slow-firing neurones lacking Ih (Fig. 4A; putative cholinergic), fast-firing neurones with an important Ih (Fig. 4B; putative GABAergic), and neurones that displayed cluster-firing (Fig. 4C; putative glutamatergic) all received glutamatergic-mediated EPSPs and bursts from the local glutamatergic network. Table 2 summarizes the electrophysiological properties of MS/DB cells showing glutamatergic input in mACSF. Of 29 fully characterized neurones that were that were recorded from acute MS/DB slices, most had the characteristics of fast-firing cells, eight were slow firing, and five cells were of the cluster-firing type.
|
|
|
|
|
Acetylcholine (ACh) originating from the septum plays an important role in learning and memory (Lee et al. 1994; Leung et al. 2003). It is well known that cholinergic muscarinic receptor activation in the MS/DB in vivo can positively affect memory and increase hippocampal theta (Monmaur & Breton, 1991; Lawson & Bland, 1993; Oddie et al. 1994; Givens & Olton, 1994, 1995; Markowska et al. 1995; Frick et al. 1996). Therefore, we investigated whether local glutamatergic neurones were involved in the actions of ACh in the MS/DB by examining if MS/DB glutamatergic neurones of the intact half-septum preparation could be activated by muscarinic receptor stimulation. Although occasional EPSPs were observed in a large proportion of the 50 neurones recorded in normal ACSF, a striking increase in spontaneous EPSPs was observed in 12 of these cells following perfusion of the preparation with carbachol (30 µM) (Fig. 7A, middle trace; 1543 ± 300% increase in the number of EPSPs per 30 s; ***P < 0.005, paired t test). Carbachol-induced EPSPs were completely abolished by the addition of the glutamatergic AMPA/kainate receptor antagonist DNQX (Fig. 7A, bottom trace; 6 out of 6 experiments; *P < 0.05, paired t test). Interestingly, we found that in a portion of the carbachol-responsive cells (n = 4/12 cells), carbachol elicited continuous or intermittent bursts of EPSPs at theta frequency (6–10 Hz range) appearing as significant peaks in the power spectra (Fig. 7A, middle trace and B); the average peak frequency for the four cells was 8.9 ± 1.6 Hz and the mean power peak amplitude was 0.35 ± 0.24 (range, 0.09–0.64). Carbachol-induced rhythmic and non-rhythmic EPSPs elicited in the presence of bicuculline (n = 2/2 cells; data not shown) were comparable to those in control suggesting that they were not reversed IPSPs. These results suggest that carbachol can induce the collective activity of MS/DB glutamatergic neurones and generate EPSPs at theta frequencies.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
One important conclusion from our study is that glutamatergic neurones within the MS/DB generated bursting activity on postsynaptic neurones that was blocked predominantly by the AMPA/kainate antagonist DNQX, and to a much lesser extent, AP-V. Such DNQX-sensitive responses were observed in acute regular slices containing the MS/DB, in trimmed MS/DB slices, in trimmed MS/DB organotypic slices and in the intact half-septum preparation. Similarly, the carbachol-induced rhythmic EPSPs in the MS/DB were also sensitive to this agent (see below). We suggest that MS/DB glutamatergic neurones activate predominantly AMPA, and to a lesser extent NMDA receptors, but probably not kainate receptors. Although DNQX is also known to block kainate receptors, previous studies have clearly shown that electrically activated synaptic responses in the MS/DB (Armstrong & Macvicar, 2001) or currents elicited by exogenous application of glutamate agonists on a variety of MS/DB neurones (Jasek & Griffith, 1998; Frye & Fincher, 2000; Hsiao & Frye, 2003) are generated solely by AMPA and NMDA receptors. The result of predominant activation of AMPA receptors by local MS/DB glutamatergic neurones is interesting in light of the evidence that cholinergic and non-cholinergic MS/DB neurones are known to express at least three types of postsynaptic glutamate receptors; that is, AMPA, NMDA and metabotropic type receptors (Schneggenburger et al. 1992, 1993a,b; Page & Everitt, 1995; Jasek & Griffith, 1998; Armstrong & MacVicar, 2001). This indicates that connections from local glutamatergic neurones activate mostly AMPA receptors (and to a lesser extent NMDA receptors), and may suggest that the other glutamate receptors are activated by glutamatergic populations located outside the MS/DB area. Recent evidence from Wu et al. (2003) does support this contention. They showed that nicotine agonists activated a glutamatergic population which induced a metabotropic glutamate receptor-mediated slow depolarization in MS/DB GABA-type neurones in regular acute MS/DB slices. However, the neurones activating the metabotropic-mediated responses were mostly localized outside the MS/DB because physically excluding the lateral septum from the MS/DB reduced the number of nicotinic responding GABAergic neurones from 90 to 40%. As other regions outside the MS/DB such as the supramammilary area (Leranth & Kiss, 1996; Kiss et al. 2000), the entorhinal cortex (Leranth et al. 1999), the frontal cortex (Jaskiw et al. 1991) and nucleus reuniens thalami (Bokor et al. 2002) may also provide important glutamatergic input, it will be important to determine what type of glutamate receptors are activated by these afferents to drive MS/DB neurones. Interestingly, a recent study using MS/DB slices has shown that activation of kainate receptors can synchronize MS/DB neurones at theta frequency (Garner et al. 2005). Therefore, although kainate receptors may not have a significant role postsynaptically in the MS/DB, their activation presynaptically could modulate neurotransmitter release and promote increased synchronization in a manner that is often observed in other brain regions such as the hippocampus (Huettner, 2003).
AMPA receptor-mediated network activity in the whole MS/DB preparation
A particularly striking result from the present study was that the glutamatergic neurones of the MS/DB are organized within a synaptically connected network of excitatory neurones. This conclusion stems from a number of observations. First, the AMPA-mediated bursts (or the voltage-clamped inward current) observed in a large portion of MS/DB neurones from acute slices were probably mediated by many synchronously discharging glutamatergic neurones because these events were sometimes in excess of 700 pA, an unlikely size if they were mediated by only one presynaptic neurone. Second, it was found that a greater number of neurones recorded in the disinhibited intact septum preparation displayed large repetitive AMPA-receptor-mediated bursts in comparison to those recorded in acute slices. This probably indicates that a greater number of neurones are recruited in the intact septum preparation which has greater network integrity compared to acute slices. Third, the bursting activity of neurones in the intact half-septum was also recorded extracellularly as long-duration field potentials, indicating that bursts were probably generated by synchronized discharges in many neurones. Accordingly, simultaneous recordings of neurones and extracellular fields from the intact septum preparation revealed that slow and large repetitive bursts were generated synchronously within the MS/DB, even when the extracellular and patch electrode were positioned several hundred micrometres away from each other. This suggested that the bursts could propagate throughout the glutamatergic network. It is interesting that cholinergic muscarinic receptors did not participate in the synchronous discharge as atropine did not affect these responses. This suggests that even though cholinergic neurones are probably activated by the 4-AP/bicuculline solution (slow-firing neurones thought to be cholinergic were often observed to fire bursts of action potentials in this solution), acetylcholine released from local axon collaterals may not contribute significantly to this type of rapid glutamatergic network activity in the MS/DB. However, it remains to be tested whether the synchronously activated MS/DB cholinergic neurones could contribute to the spontaneous burst/network activity through the activation of extrasynaptic nicotinic receptors in the MS/DB (Henderson et al. 2005).
It should be emphasized here that although the burst firing observed in this study was elicited under non-physiological conditions and is therefore unlikely to reflect any mechanism of MS/DB activity occurring in vivo, it is nonetheless indicative of the presence of a local excitatory network of glutamatergic cells in the MS/DB. Thus, the repetitive synchronous discharge activity observed in the disinhibited MS/DB is highly reminiscent of periodic burst discharges that can be recorded in vitro under similar experimental conditions from various brain structures containing an important number of glutamatergic neurones such as the cerebral cortex (Sanchez-Vives & McCormick, 2000), the brain stem (Koshiya & Smith, 1999) and the CA3 area of the hippocampus (Muller & Misgeld, 1991; Traub & Miles, 1991). Similarities between the bursting observed in the CA3 area and in the septum during disinhibition include large-amplitude, regular and sustained bursts with relatively long-lasting silent periods between them (Staley et al. 1998) and the primary requirement for AMPA receptors (Avoli et al. 1993; Traub et al. 1995). In the CA3, the repetitive bursting activity is due to propagating discharges through relatively extensive recurrent/reciprocal excitatory connections between CA3 pyramidal neurones (Li et al. 1994), driving the entire population into synchronized oscillatory activity (Traub & Wong, 1982; Miles & Wong, 1983, 1986; Traub & Miles, 1991). Of interest, our immunohistochemical data also suggested that glutamatergic neurones are part of an interconnected network as we observed that a significant number of VGLUT2-positive MS/DB neurones appeared to receive VGLUT2/synaptophysin-positive terminals that originated from local neurones in organotypic slices. Therefore, our immunohistochemical data together with the high occurrence of large repetitive glutamatergic bursts are strong evidence that glutamatergic neurones of the MS/DB are connected. Finally, it should be noted that the repetitive glutamatergic bursting activity appeared more prominent in organotypic than in acute slices and necessitated less disinhibition (only bicuculline was needed instead of combined 4-AP and bicuculline in the acute preparations). This may be due to an increase, or reorganization, of glutamatergic synaptic terminals in the MS/DB culture preparation that resulted in an enhancement of the excitation and of synchronous discharge.
Glutamatergic neurones in the MS/DB not only made connections together but also to cholinergic and GABAergic neurones. AMPA receptor-mediated EPSPs and bursts were frequently recorded in putative cholinergic and GABAergic neurones identified by their electrophysiological characteristics. Indeed, glutamatergic excitation was present in slow-firing neurones that lack an Ih, and in fast/burst-firing neurones that had a prominent Ih, which are considered to be key features distinguishing cholinergic and GABAergic neurones (Griffith & Matthews, 1986; Markram & Segal, 1990; Gorelova & Reiner, 1996; Sotty et al. 2003). Our immunohistochemistry data were also in agreement with this as VGLUT2-positive puncta were seen in close vicinity of the soma and dendrites of ChAT-positive and GAD67-positive neurones in the organotypic MS/DB mini-slices. These experiments thus further suggest that glutamatergic neurones within the MS/DB make synapses on cholinergic and GABAergic neurones. In a previous study Hajszan et al. (2004) found that VGLUT2-positive puncta can be localized close to neurones immunoreactive to parvalbumin, a calcium-binding protein found in a subset of GABAergic neurones, in the under- and over-cut septum in vivo (although the possibility was not excluded that some VGLUT2 terminals may have originated from the lateral septum (LS). Our results extend these observations by showing that VGLUT2-positive puncta from MS/DB glutamatergic neurones are likely to be present on GAD67-positive GABAergic neurones. Numerous VGLUT2-positive synapses were recently shown to make contact with cholinergic neurones in the MS/DB, but unlike in our study, the origin of the VGLUT2 puncta was not determined (Wu et al. 2004). These results suggest that local connections from glutamatergic neurones may be important to the output of cholinergic and GABAergic neurones (Fig. 9).
|
MS/DB glutamatergic neurones can generate EPSPs at theta frequency
The role of ACh in the MS/DB has been shown to be very important in hippocampal theta oscillation, and in learning and memory. The blockade of muscarinic receptors in the MS/DB in vivo can impair learning and disrupt hippocampal theta rhythm (Stewart & Fox, 1990; Givens & Olton, 1995), whereas injecting muscarinic agonists into the MS/DB results in continuous theta field activity (Lawson & Bland, 1993) and may improve memory (Markowska et al. 1995; Frick et al. 1996). Recent reports suggested that the role of ACh in memory may be a result of the activation of GABAergic neurones (Alreja et al. 2000) and/or cholinergic neurones (Henderson et al. 2005) in the MS/DB. Here we provide evidence that in addition, ACh can modulate MS/DB glutamatergic neurones because muscarinic agonist application into the whole septum elicited AMPA-receptor mediated EPSPs at theta frequency (6–10 Hz). Consistent with the idea that muscarinic stimulation activates the glutamatergic neurones in the MS/DB, we also show that carbachol can depolarize and increase the firing frequency of cluster-firing neurones in the intact half-septum. Taken together, these results suggest that muscarinic-induced hippocampal theta generation may be due to the combined activation of GABAergic, cholinergic and glutamatergic MS/DB neurones (Fig. 9). Although, MS/DB glutamatergic neurones appear to be activated by Ach at the relevant frequency, it remains to be determined if and how they are involved in theta and in learning and memory. A recent study has shown that blocking AMPA/kainate receptors in the septum of unanaesthetized alert walking rat in vivo (known has type 1 theta; Stewart & Fox, 1990) did not affect hippocampal theta (Leung & Shen, 2004). However, others have shown that hippocampal theta triggered by intraseptal ACh (using physostigmine) in anaesthetized (i.e. immobile type 2 theta) rats in vivo is strongly antagonized by intraseptal perfusion of AMPA receptor antagonists (Puma et al. 1996; Puma & Bizot, 1999). Therefore, it may be speculated that carbachol-activated MS/DB glutamatergic neurones contribute to atropine-sensitive theta (type 2 theta) recorded during immobility in anaesthetized (Stewart & Fox, 1990) or unanaesthetized rats prior to the initiation of movements (Oddie & Bland, 1998) but not to the atropine-insensitive theta observed in behaving walking rats (Lawson & Bland, 1993). It will be important to determine how MS/DB glutamatergic neurones are involved in hippocampal theta recorded during immobility in unanaesthetized animals. Although it remains unknown how MS/DB glutamate neurones contribute to theta rhythm, there is evidence that blocking AMPA/kainate receptors in the medial septum can abolish the expression of two different types of hippocampal-dependent memory formation (Izquierdo, 1994).
In conclusion, this study presents the first evidence that glutamatergic neurones in the MS/DB have a significant functional role. These neurones are part of a glutamatergic neuronal network that can be activated by acetylcholine and may potentially drive cholinergic and GABAergic neurones in a physiologically relevant manner. Therefore, MS/DB glutamatergic neurones could potentially be important in septal function and in hippocampal activity.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Alreja
M, Wu
M, Liu
W, Atkins
JB, Leranth
C
&
Shanabrough
M (2000). Muscarinic tone sustains impulse flow in the septohippocampal GABA but not cholinergic pathway: implications for learning and memory. J Neurosci
20, 8103–8110.
Amaral DG & Kurz J (1985). An analysis of the origins of the cholinergic and noncholinergic septal projections to the hippocampal formation of the rat. J Comp Neurol 240, 37–59.[CrossRef][Medline]
Armstrong
JN
&
MacVicar
BA (2001). Theta-frequency facilitation of AMPA receptor-mediated synaptic currents in the principal cells of the medial septum. J Neurophysiol
85, 1709–1718.
Avoli M, D'Antuono M, Louvel J, Kohling R, Biagini G, Pumain R, D'Arcangelo G & Tancredi V (2002). Network and pharmacological mechanisms leading to epileptiform synchronization in the limbic system in vitro. Prog Neurobiol 68, 167–207.[CrossRef][Medline]
Avoli
M, Psarropoulou
C, Tancredi
V
&
Fueta
Y (1993). On the synchronous activity induced by 4-aminopyridine in the CA3 subfield of juvenile rat hippocampus. J Neurophysiol
70, 1018–1029.
Baisden RH, Woodruff ML & Hoover DB (1984). Cholinergic and non-cholinergic septo-hippocampal projections: a double-label horseradish peroxidase-acetylcholinesterase study in the rabbit. Brain Res 290, 146–151.[CrossRef][Medline]
Barbarosie
M
&
Avoli
M (1997). CA3-driven hippocampal-entorhinal loop controls rather than sustains in vitro limbic seizures. J Neurosci
17, 9308–9314.
Bland BH, Oddie SD, Colom LV & Vertes RP (1994). Extrinsic modulation of medial septal cell discharges by the ascending brainstem hippocampal synchronizing pathway. Hippocampus 4, 649–660.[CrossRef][Medline]
Bland SK & Bland BH (1986). Medial septal modulation of hippocampal theta cell discharges. Brain Res 375, 102–116.[CrossRef][Medline]
Blanton MGLO, Turco JJ & Kriegstein AR (1989). Whole cell recording from neurons in slices of reptilian and mammalian cerebral cortex. J Neurosci Methods 30, 203–210.[CrossRef][Medline]
Bokor H, Csaki A, Kocsis K & Kiss J (2002). Cellular architecture of the nucleus reuniens thalami and its putative aspartatergic/glutamatergic projection to the hippocampus and medial septum in the rat. Eur J Neurosci 16, 1227–1239.[CrossRef][Medline]
Brashear HR, Zaborszky L & Heimer L (1986). Distribution of GABAergic and cholinergic neurons in the rat diagonal band. Neuroscience 17, 439–451.[CrossRef][Medline]
Buchs PA, Stoppini L & Muller D (1993). Structural modifications associated with synaptic development in area CA1 of rat hippocampal organotypic cultures. Brain Res Dev Brain Res 71, 81–91.[CrossRef][Medline]
Chandler JP & Crutcher KA (1983). The septohippocampal projection in the rat: an electron microscopic horseradish peroxidase study. Neuroscience 10, 685–696.[CrossRef][Medline]
Chedotal A, Cozzari C, Faure MP, Hartman BK & Hamel E (1994). Distinct choline acetyltransferase (ChAT) and vasoactive intestinal polypeptide (VIP) bipolar neurons project to local blood vessels in the rat cerebral cortex. Brain Res 646, 181–193.[CrossRef][Medline]
Cossette P, Umbriaco D, Zamar N, Hamel E & Descarries L (1993). Recovery of choline acetyltransferase activity without sprouting of the residual acetylcholine innervation in adult rat cerebral cortex after lesion of the nucleus basalis. Brain Res 630, 195–206.[CrossRef][Medline]
Danik M, Cassoly E, Manseau F, Sotty F, Mouginot D & Williams S (2005). Frequent coexpression of the vesicular glutamate transporter 1 and 2 genes, as well as coexpression with genes for choline acetyltransferase or glutamic acid decarboxylase in neurons of the rat brain. J Neurosci Res (in press).
Danik M, Puma C, Quirion R & Williams S (2003). Widely expressed transcripts for chemokine receptor CXCR1 in identified glutamatergic, gamma-aminobutyric acidergic, and cholinergic neurons and astrocytes of the rat brain: a single-cell reverse transcription-multiplex polymerase chain reaction study. J Neurosci Res 74, 286–295.[CrossRef][Medline]
Dzhala VI & Staley KJ (2003). Transition from interictal to ictal activity in limbic networks in vitro. J Neurosci 23, 7873–7880.
