- Email: [email protected]
A novel mouse brain slice preparation of the hippocampo–accumbens pathway
Journal of Neuroscience Methods 137 (2004) 49–60
A novel mouse brain slice preparation of the hippocampo–accumbens pathway Robert T. Matthews a,∗ , Olusegun Coker a , Danny G. Winder b b
a Department of Anatomy & Physiology, Meharry Medical College, Nashville, TN 37208, USA Department of Molecular Physiology & Biophysics, Center for Molecular Neuroscience, Kennedy Center, Vanderbilt University Medical School, Nashville, TN, USA
Received 5 November 2003; received in revised form 20 January 2004; accepted 3 February 2004
Abstract The nucleus accumbens (NAc) is an important component of circuitry that underlies reward related behaviors and the rewarding properties of drugs of abuse. Glutamatergic afferents to the nucleus are critical for its normal function and for behaviors related to drug addiction. An angled, sagittal mouse brain slice preparation has been designed to facilitate concurrent stimulation of two major glutamatergic afferent pathways to the nucleus accumbens. Medium spiny neurons at the medial core/shell boundary of the accumbens were depolarized by stimulation of either hippocampal or limbic cortical afferents through activation of AMPA-type glutamate receptors. High frequency but not low frequency stimulation of hippocampal afferents depolarized medium spiny neurons to a membrane potential that resembled the up state observed upon high frequency stimulation in vivo. The magnitude of the membrane depolarization was positively correlated with the amplitude of the stimulus-evoked EPSP. Concurrent stimulation of hippocampal and limbic cortical afferents at theta frequency selectively induced a long-term depression (LTD) in the magnitude of stimulus-evoked EPSPs on the hippocampal afferent only. These data suggest that this brain slice preparation can be used to study mechanisms underlying synaptic plasticity at two of the critical glutamatergic afferent synapses in the nucleus accumbens as well as characterizing potential interactions between afferents. Additionally, LTD at hippocampo–accumbens synapses can be induced at a stimulus frequency known to support reinstatement of drug seeking behavior. © 2004 Elsevier B.V. All rights reserved. Keywords: Glutamatergic synapse; Nucleus accumbens; Fornix; Long-term depression
1. Introduction The nucleus accumbens (NAc) has been shown to be an important component of the limbic circuitry involved in long-lasting behaviors induced by chronic exposure to drugs of abuse such as cocaine (Everitt and Wolf, 2002; Hyman and Malenka, 2001). Regulation of excitatory inputs that converge on the NAc is increasingly thought to be a key mechanism for modulation of long-term behavioral responses to drugs of abuse. The NAc receives glutamatergic inputs from a variety of sources, including the limbic neocortex, the ventral subiculum and hippocampus (hippocampal formation), the amygdala and the dorsal medial thalamus (Pennartz et al., 1994).
∗ Corresponding author. Tel.: +1-615-327-6288; fax: +1-615-327-6713. E-mail address: [email protected] (R.T. Matthews).
0165-0270/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jneumeth.2004.02.001
Several pieces of evidence suggest that the hippocampal input plays a key role in behaviors dependent on the NAc. First, stimulation of the hippocampal formation at theta frequency in vivo has been shown to cause reinstatement of cocaine seeking behavior in rats that had previously self-administered cocaine (Vorel et al., 2001). Second, the hippocampal input appears to play a critical role in modulating resting membrane potential of medium spiny neurons, the output neurons of the NAc (O’Donnell and Grace, 1995). Medium spiny neurons switch between hyperpolarized “down” and depolarized “up” states in vivo, and lesions of the fornix eliminate the up state. Finally, in vivo experiments have demonstrated long-term potentiation (LTP) at hippocampal formation/accumbens synapses following high frequency stimulation (Drevets et al., 2001; Onn et al., 2000). To date, most attempts to study the hippocampo–NAc pathway have utilized in vivo approaches. Thus, the mechanisms of up/down state transitions, high frequency-dependent LTP, the effect of theta frequency stimulation on synaptic
50
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
efficacy and the possibility of long-term depression (LTD) at this synapse are largely unknown. The development of an in vitro model for the study of this pathway has proven difficult, likely due to the serpentine route taken by the hippocampal efferents. In the present study, we have developed and verified a reliable method for the study of this synapse in vitro, and have performed a preliminary characterization of this synapse. We find that, through the use of an angled parasagittal brain slice, we can reliably evoke glutamatergic synaptic responses in NAc neurons in response to stimulation in or near the fornix. These responses are severely attenuated in fornix lesioned animals, and do not exhibit paired-pulse interactions with synaptic responses elicited by stimulating afferents entering the rostral end of the NAc. We find that theta frequency stimulation of these synapses results predominantly in LTD of the hippocampal input, with no change in the limbic neocortical input, suggesting that distinct forms of plasticity may exist at these synapses.
2. Materials and methods 2.1. Slice preparation and current clamp recording C57bl/6J male mice (Jackson Laboratories), 6–10 weeks old, were used for preparation of brain slices. Animal experiments were carried out in accordance with the National Institute of Health Guide for the Care and Use of Animals (NIH publication no. 80–23) revised in 1996. Mice were housed under standard conditions, 2–4 mice per cage with unlimited access to food and water. Except for lesion experiments, mice were anesthetized with isoflurane, brains quickly removed and chilled in ice-cold buffered salt solution. The whole brain was trimmed to expose an angled parasagittal surface (Fig. 1A), glued to a glass slide and cut with a vibratome to yield a single 400 m thick brain slice that contained the central portion of the NAc and most of the ipsilateral fornix (see Fig. 1B). The angle of the cut that best preserved functional hippocampal and neocortical afferents was approximately 35◦ from vertical. To reliably produce this angle, the brain was placed in a commercially available stainless steel sagittal brain slicing jig (Ted Pella, Inc.) with the brain placed at an angle in the jig such that the olfactory tubercle was just level with the top of the jig on one side prior to blocking. This procedure provided enough stability for angled trimming of the whole brain with razor blades such that usable vibratome brain slices were obtained from greater than 90% of mice. Slices were immediately transferred to an interface-type recording chamber (Fine Science Tools) and permitted to recover for 1–2 h at 29 ◦ C before beginning recordings at the same temperature. Sharp electrode current clamp recordings were performed using 90–150 M pipettes filled with 3 M K+ acetate. Recording electrodes were pulled on a Flaming/Brown Micropipette Puller (Sutter Instruments) using thick-walled filament-containing borosilicate glass capillaries. Voltage
signals were amplified with an Axoclamp 2B amplifier and data was collected and analyzed with pClamp 8.1 software. Drugs were applied by adding them to the perfusion solution reservoir. Flow rate was 2–3 ml/min resulting in a 4–6 min delay for drugs to enter the slice chamber. Perfusion solution consisted of NaCl 124 mM, KCl 4.4 mM, CaCl2 2.5 mM, NaH2 PO4 1 mM, MgSO4 1.2 mM, NaHCO3 26 mM, glucose 10 mM. Perfusion solutions were continuously bubbled with 95% O2 /5% CO2 . Picrotoxin (25 M) was present, unless otherwise noted, in all experiments in order to block GABA-A dependent IPSPs that were found in preliminary experiments to mask EPSPs in some cases. 2.2. Synaptic stimulation Stimulating electrodes were made from either twisted bipolar nichrome wire (0.002 in./50 m diameter) with minimal tip separation or bipolar tungsten wire (0.005 in./127 m diameter) with tip separation of about 250 m. Monopolar square wave pulses were delivered to wire electrodes with a Grass S88 stimulator and a Grass constant voltage stimulus isolation unit. Stimulating electrodes were positioned on the surface of the slice just rostral to the column of the fornix where the precommissural fibers enter the lateral septal nucleus (tungsten wire) and in the white matter at the rostral/dorsal border of the NAc (nichrome wire, Fig. 1B). These electrode placements were presumed to be stimulating hippocampal formation afferents and limbic neocortical inputs, respectively. However, stimulation of local accumbens fibers by the neocortical electrode cannot be ruled out because of its position at the border of the nucleus. Medium spiny neurons, located either in the accumbens core just dorsal to the anterior commissure, or in the adjacent medial–dorsal core/shell transition, were impaled with recording electrodes and allowed to stabilize for 10–15 min. In the rostral/caudal plane, recorded neurons were located at the midpoint of the nucleus to slightly caudal to the midpoint (Fig. 1B). These neurons were located in a region of the nucleus previously shown to have a relatively high density of afferents originating from the ventral subiculum (Groenewegen et al., 1987). Pulses between 80 and 200 s duration and 2–130 V were applied to stimulating electrodes in order to measure input/output curves for each afferent. For paired pulse facilitation (PPF) experiments, pulse pairs with inter-pulse intervals of 10–200 ms and about 50% of maximal amplitude were given on each input to find an interval that produced a significant facilitation with near complete return of membrane potential to resting level before the second pulse. Six pulse pairs with inter-pair intervals of 20 s were averaged for data comparisons. For pulse train stimulation designed to measure membrane depolarization, only the fornix stimulating electrode was used and stimulus strength was set to produce an EPSP at 80–100% of maximal EPSP size for that neuron. The magnitude of membrane depolarization was measured repeatedly during the stimulus train at time points just prior to the next stimulus artifact.
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
51
Fig. 1. (A) Coronal diagram from a mouse brain atlas (Konsman, 2003) at an intermediate rostral/caudal level of the NAc showing the angle of the sagittal section (between the dark lines) used for electrophysiological recording. The ventromedial (right-facing) surface was placed facing up in the recording chamber for proper placement of stimulating electrodes. (B) Photograph of a typical mouse brain slice with all electrodes in place. A tungsten-stimulating electrode is placed in the precommissural fornix fibers where they enter the septal nuclei. A nichrome-stimulating electrode is placed at the rostral/dorsal border of the NAc where fibers enter from the limbic neocortex. The tip of the recording electrode is indicated by the ∗ D, dorsal; R, rostral; Hippo, hippocampus.
52
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
2.3. Fornix lesions
2.6. Data analysis
Unilateral fornix lesions were done in a subset of mice. Mice were anesthetized with trichloroethylene and mounted in a stereotaxic apparatus. After deflecting the skin over the skull, a hole was drilled over the fornix at 0.9 mm posterior to bregma. A 2 mm wide knife made from a razor blade was lowered to a depth that cut the fornix on one side of the brain without damaging the underlying thalamus. The medial most portion of the fornix (≈0.3 mm) was spared in order to prevent bleeding from the sagittal sinus (Fig. 5). Mice were permitted to recover for 3–6 days and then anesthetized with isoflurane for generating brain slices as described above. Sham lesioned animals received the same treatment except the knife blade was not lowered into the brain or cut through only the overlying neocortex. On experiment days, one sham and one lesioned animal were used in a single blind protocol. NAc neurons were impaled and tested for minimum criteria of healthy neurons before and after input/output data were collected from stimulation of afferents as described for paired pulse experiments. Any neurons that failed to meet criteria were discarded. Criteria included firing threshold between −50 and −45 mV, action potential peak above 0 mV, multiple spiking during membrane depolarization above threshold and evidence of inward rectification during hyperpolarizing current pulses.
Appropriate statistical analyses (indicated within figure legends) were performed using Prizm and InStat software (Graphpad). Multiple regression analyses were done with Microcal Origin 7.0.
2.4. Synaptic plasticity Baseline EPSP amplitudes, about 50% of maximum and >4 mV, were recorded following sequential stimulation of limbic neocortical/local fiber inputs and hippocampal formation (fornix) inputs. Pairs of stimuli, one to each input, were separated by 200 ms and followed by hyperpolarizing pulses used to measure membrane resistance. The inter-pulse interval was long enough to avoid gating of one afferent by the other (O’Donnell and Grace, 1995). Pulse pairs were given every 20 s. Baseline data were collected for 20 min followed by stimulation trains at various frequencies given simultaneously on both inputs while current was injected into the neuron to maintain membrane potential near firing threshold (∼−55 mV). Theta frequency stimulation consisted of 10 trains of 8 Hz stimulation, 6 s per train with 7 s inter-train intervals, at the same stimulus intensity used for data collection before and after theta stimulation, but of opposite polarity. After theta stimulation, holding current was turned off and every 20 s. stimulus pairs were restarted. Any neuron whose membrane resistance changed by more than 15% over the next 90 min of data collection was discarded from further analysis.
3. Results 3.1. Mouse medium spiny neuron electrophysiology Medium spiny neurons were examined with sharp electrode current clamp recordings. Basic membrane properties recorded in the presence of picrotoxin resembled those reported previously for rat medium spiny neurons (Chang and Kitai, 1986; O’Donnell and Grace, 1993; Pennartz and Kitai, 1991). Mean action potential amplitude was 61.0mV± 1.7 (S.E.M.; n = 22), spike width at 1/3 above base was 1.30 ms ± 0.08, input resistance was 158 M ± 18 and resting membrane potential was −79.0 mV ± 2.0. Neurons fired single spikes or trains of spikes in response to membrane depolarization above −45 mV with no bursting (Fig. 2A). Stimulation of precommissural fornix fibers or neocortical fibers resulted in synaptic potentials in the majority of medium spiny neurons tested. In the absence of picrotoxin, synaptic responses were sometimes a mixture of IPSPs and EPSPs. Because the objective of this study was to characterize glutamatergic inputs to medium spiny neurons, IPSPs were routinely blocked with 25 M picrotoxin (data not shown). After blockade of IPSPs, monosynaptic EPSPs could be elicited from fornix fiber stimulation with typical latencies of 4.8–5.9 ms and from neocortical fibers with latencies of 4.2–5.0 ms. Fig. 2C shows the graded responses of a medium spiny neuron to stepped fornix fiber stimulation that produced an action potential at the highest stimulation voltage. 3.2. EPSPs are AMPA receptor dependent The selective AMPA receptor antagonists, CNQX (10 M, n = 3) and DNQX (10 M, n = 2) blocked EPSPs generated by stimulation of either fornix or neocortical fibers. Blockade was 90–100% complete in all cases, suggesting that at resting membrane potential, most of the current was carried by AMPA-type glutamate receptors. However, a small percentage of the current underlying the EPSPs from these sources might be controlled by other receptor types. Fig. 3A and B shows the effect of CNQX on fornix and neocortical induced EPSPs while Fig. 3C shows the effect of DNQX on the fornix-induced EPSPs in a different neuron.
2.5. Pharmacology 3.3. Paired pulse facilitation All drugs were bath applied. Picrotoxin, CNQX, and DNQX were purchased from Sigma (St. Louis, MO). DMSO (0.02%, v/v) was the carrier for picrotoxin and CNQX.
To begin to test whether the presumed fornix fiber and neocortical fiber stimulating electrodes were stimulating the
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
53
Fig. 2. Examples of recordings from mouse medium spiny neurons. (A) Superimposed voltage responses to 50 pA stepped current injections. Note the ramped depolarizing response (arrow) leading up to the action potential and the time-independent type of inward rectification to hyperpolarizing pulses that are typical of medium spiny neurons in other species. (B) I/V plot from the neuron shown in (A). (C) Superimposed responses of a different neuron to repeated stimulation of fornix fibers in 4 V steps starting at 60 V (0.1 ms pulses every 10 s). Arrows indicate stimulus artifact.
same or different afferent fibers, a paired pulse experiment was designed to examine facilitation of EPSP amplitude within and between electrode sites. Four medium spiny neurons from three mice were tested for the presence of PPF with stimulation on each electrode alone and with stimulation of one electrode followed by the other electrode to create the pulse pair. Pairs of stimulation pulses on either electrode produced significant facilitation of the second EPSP in one neuron (P < 0.01), but no significant facilitation was observed when pulse pairs were presented on alternating electrodes (Fig. 4). In three neurons, paired pulse facilitation was seen on only one input but no significant facilitation between electrodes was observed. These data suggest that our electrode placements are stimulating two different inputs. 3.4. EPSPs following ipsilateral fimbria/fornix lesions Unilateral lesions (n = 5) or sham lesions (n = 5) were performed on mice 3–6 days prior to creating brain slices. Lesions were located at the rostral tip of the hippocampus, approximately 1.2 mm caudal to the fornix/septal area boundary where the stimulating electrode was positioned in
the brain slice (Fig. 5, inset). Because of the angled nature of the brain slice, the lesion was not visible in the slice used for recordings and therefore recordings from sham versus lesioned animals could be done blind. Two to five medium spiny neurons in each brain slice were tested for the presence of EPSPs following stimulation of neocortical and presumed hippocampal afferents (total of 34 neurons). In four of the five slices from lesioned animals (13 neurons), no EPSPs were detected when the fornix was stimulated, even when stimulus intensity was increased to 130 V and 300 s. One slice from a lesioned animal had three neurons with EPSPs of up to 5 mV at the highest stimulus intensity. The mean maximal EPSP amplitude from all neurons in slices from lesioned animals was 2.0 mV±0.7 (Fig. 5). In contrast, typical EPSPs of 10 mV in amplitude or larger (mean = 15.2 mV ± 1.9) were observed in the same neurons when neocortical fibers were stimulated (Fig. 5). Eighteen neurons in brain slices from sham lesioned animals had typical EPSPs following stimulation of either electrode (11.7 ± 1.3 on fornix electrode; 14.9 ± 1.9 on neocortical electrode, Fig. 5). Statistical comparison of the mean amplitudes of EPSPs from all neurons tested showed a significant reduction in the
54
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
Fig. 3. The effects of two selective AMPA type glutamate receptor antagonists, DNQX (n = 2 neurons) and CNQX (n = 3 neurons), on EPSP amplitudes induced by stimulation of either the fornix fibers or neocortical fibers. (A) Plot of the sizes of fornix-induced EPSPs over time in a medium spiny neuron, before and after bath application of CNQX. (B) Plot of neocortical-induced EPSPs in the same neuron. (C) Averaged traces of the fornix-induced EPSPs (n = 70 for each trace) from a different neuron before drug application (control), 15–25 min after adding DNQX and 10–20 min after drug washout. The input resistance did not change (data not shown).
Fig. 4. Data from a medium spiny neuron tested for paired pulse facilitation within inputs (A) and across inputs (B). (A) Pairs of stimulation pulses separated by 80 ms applied to the fornix electrode (100 s/63 V) produced significant facilitation of the second EPSP (paired Student’s t-test) as did pairs of stimuli applied to the local/neocortical electrode (50 s/60 V). (B) When pulse pairs were presented on alternating electrodes, no significant facilitation was observed. No picrotoxin was present in this experiment.
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
Fig. 5. Effects of a fornix lesion or sham lesion on the maximum EPSP size that could be recorded with stimulation of each input. Stimulus intensity was 100 s in duration and up to 120 V, or up to a voltage that induced an action potential. Only the fornix fiber dependent EPSP was significantly reduced by a prior lesion (unpaired Student’s t-test). Inset: Diagram from mouse brain atlas at the level just rostral to the hippocampus. The hatched area on the left side of the diagram indicates the location and extent of the fimbria/fornix lesion. Lesion placement was verified in three practice animals by coronal sectioning of the brain (data not shown). The lesion location was at least 1 mm caudal to the placement of stimulating electrodes for activation of fornix fibers. In addition to the fornix, the overlying neocortex and corpus callosum were also lesioned.
size of the EPSPs generated by the fornix electrode in lesioned animals as compared to sham animals (P < 0.001). These data strongly suggest that stimulation of tissue at the fornix/septal area boundary in angled sagittal brain slices selectively activates accumbens afferents originating from the hippocampal formation. 3.5. Control of up/down states by fornix stimulation Previous in vivo studies have demonstrated that the hippocampal input to the NAc plays a critical role in the initiation of up state transitions, and that single pulse stimulation of this input was sufficient to initiate an up state. Surprisingly, we find that single pulse stimulation of fornix fibers was not sufficient to induce medium spiny neurons to switch to an up state in our in vitro preparation, even when the EPSP was sufficient to induce an action potential (see Fig. 2C as an example). Since the up state may be a necessary component of the mechanisms underlying plasticity, especially calcium and/or NMDA receptor dependent forms of plasticity, we tested a variety of train frequencies in an attempt to induce an up state in medium spiny neurons in vitro. Eleven medium spiny neurons were recorded while stimulating fornix fibers at train frequencies of 5–100 Hz for 1 s or more (Fig. 6C and D). As with all neurons in this
55
study, resting membrane potential before the stimulus train was equivalent to the down state (<−75 mV). Each neuron received 1–4 stimulus trains of different frequencies, separated by at least 5 min. Picrotoxin was present in two of the eleven experiments, however data from these two experiments were not different from data without picrotoxin and were therefore included in the analysis. In order to compare membrane responses across different stimulus frequencies, maximum membrane depolarization during the first 1 s of the train was plotted against train frequency (Fig. 6A). The membrane depolarization was roughly correlated with train frequency (R2 = 0.57, F = 28.34, P > 0.001). A depolarization to the approximate level of the up state observed in vivo, −55 to −65 mV, was seen only at stimulus frequencies above 20 Hz. However, the size of the maximal EPSP varied greatly between neurons in this experiment (range of 2–30 mV). Therefore, a multiple regression analysis with amount of membrane depolarization as the dependent variable, and train frequency and EPSP size as independent variables was done (Fig. 6B). A better correlation between degree of membrane depolarization and both EPSP size and train frequency was observed (R2 = 0.71, F = 24.85, P > 0.001) as compared to train frequency alone. 3.6. Synaptic plasticity with theta frequency stimulation We next tested the hypothesis that synaptic plasticity can be induced at the hippocampal afferent/medium spiny neuron synapse in vitro. Fig. 7A shows one neuron’s averaged baseline EPSPs to stimulation of neocortical fibers followed by stimulation of fornix fibers, with enough time between stimulation of each electrode to allow the membrane potential to return to baseline. Using this protocol, we chose to test the effect of trains of theta frequency stimulation on EPSP amplitude because it is known that subpopulations of hippocampal formation neurons frequently fire bursts of action potentials in the theta frequency range, and because theta stimulation of the subiculum has been shown to reinstate cocaine seeking behavior in rats (Vorel et al., 2001). Ten trains, 7 s each, of 8 Hz stimulation (6 s inter-train interval) applied to both electrodes simultaneously resulted in a lasting depression of the fornix-induced EPSP in six of eight neurons (Fig. 7B) while no significant change was observed in the neocortical-induced EPSP (Fig. 7C). Two neurons showed a non-significant trend toward an LTP of the fornix-induced EPSP after this stimulus protocol. These data suggest that theta frequency stimulation can selectively induce long-term depression of fornix-dependent but not limbic cortical-dependent EPSPs recorded in medium spiny neurons of the mouse NAc.
4. Discussion One of the major goals of this study was to create a brain slice preparation that would keep intact a large portion of
56
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
Fig. 6. Correlations of fornix stimulation frequency and fornix-induced EPSP size on the degree of membrane depolarization of medium spiny accumbens neurons. (A) Plot of stimulation frequency vs. change in membrane potential (r 2 = 0.57; linear regression analysis of variance). Each open circle represents one neuron’s response after a single stimulus train. The dotted lines indicate the range of action potential thresholds for medium spiny neurons. The circles marked with ‘x’ are representative responses to theta frequency and high frequency stimulation that are shown in parts (C and D), respectively. (B) A 3D plot of EPSP size vs. stimulation frequency vs. change in membrane potential (R2 = 0.71; multiple regression analysis of variance). Each filled dot represents data plotted in (A) with the added dimension of EPSP size. (C and D) Voltage traces from the same neuron that show the effects of 2 s of theta (8 Hz) stimulation (C) or 1 s of 50 Hz stimulation (D) of the fornix on membrane potential. The thick bar represents the time course of the stimulus train. The arrow indicates an action potential that was generated by the first stimulus pulse of the train. Stimulus artifacts are truncated by filtering the signals with an eight pole Bessel filter set at low pass 500 Hz.
the NAc along with two of its major glutamatergic afferents, one from the hippocampal formation and one from the limbic neocortex. A second goal was to further characterize the physiology of the hippocampal afferent to the NAc, including postsynaptic receptor mechanisms and interactions between the hippocampal inputs and glutamatergic inputs from neocortical afferents. Specifically, we chose to look at the effects of afferent stimulation on medium spiny neurons because of previous work suggesting a critical role for these neurons in plasticity that may underlie reinstatement of drug seeking behavior (Everitt and Wolf, 2002; Nicola et al., 2000; Winder et al., 2002).
4.1. Brain slice preparation We have found that an angled parasagittal slice preparation of mouse forebrain can preserve a 400 m thick section of the accumbens along with enough of the afferents originating from the frontal cortex and the hippocampal formation to obtain robust synaptic potentials in medium spiny neurons from both sources. Parasagittal preparations from rat brain have been previously reported to also be appropriate for recording limbic cortical and fornix-induced EPSPs but we were not able to reliably obtain EPSPs from the fornix in mouse brain slices without changing the angle of the sagit-
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
57
Fig. 7. Selective LTD-like response at fornix synapses. (A) Trace averages of 100 responses following sequential local/neocortical fiber and fornix fiber stimulation from an example neuron. EPSPs are approximately 60% (0.09 ms/20 V) and 80% (0.09 ms/50 V) of maximal responses in this neuron, respectively. Note the short latency to initiation of the EPSPs (range for all neurons tested was 5–6 ms for the fornix and 3–5 ms for limbic cortical fiber simulation). Stimulus pairs were presented every 20 s. This stimulation protocol, along with a small hyperpolarizing pulse to monitor membrane resistance (not shown), was typically used to measure synaptic plasticity after tetanic stimulation of inputs. Open arrows indicate stimulus artifacts. (B) Plot of averaged data (±S.E.M.) showing the effect of theta stimulation on fornix-induced EPSPs in six of eight neurons that showed a sustained decrease in the size of the EPSP (group average EPSP size 45–50 min after theta stimulation was significantly smaller than the group average of 5 min just prior to stimulation; paired Student’s t-test, P > 0.01). Two neurons showed a non-significant trend toward LTP (data not shown). For graphing, data points were normalized to the mean of the EPSP amplitudes for the 5 min period immediately prior to theta stimulation. Arrow indicates point of interruption of data collection (approx. 2.5 min) for application of trains of theta stimulation (see Section 2 for details). (C) Averaged effect of theta stimulation on limbic cortical-induced EPSPs in five of the eight neurons tested for fornix input plasticity in (B) (other three neurons were tested only for fornix input plasticity). (D) Averaged effect of theta stimulation on input resistance in the six neurons shown in (B). In (B and C), superimposed averaged (n = 15) EPSP traces from one neuron just before and 45 min after theta stimulation are shown as insets.
tal cut (Pennartz and Kitai, 1991; Charara and Grace, 2003; Pennartz et al., 1992,1993). Synaptic potentials could be obtained from neurons throughout the nucleus, including from neurons located in the dorsal/medial core/shell boundary, the part of the nucleus that has been shown to have the densest plexus of fibers originating from the hippocampal formation (Groenewegen et al., 1987) and are most responsive to fornix stimulation in vivo (DeFrance et al., 1985). Stimulation of precommissural fornix fibers at the point where they exit the fornix adjacent to the septal nuclei resulted in IPSP/EPSP
complexes of variable amplitude. IPSPs were not observed in the presence of picrotoxin, suggesting GABA-A receptor dependent currents were responsible. The sources of IPSPs are unclear but could include afferents from neurons in the lateral septal nucleus or polysynaptic pathways involving GABAergic neurons within the NAc (Brog et al., 1993; Hussain et al., 1996). EPSPs generated from stimulation of precommissural fornix fibers were substantially blocked by AMPA receptor antagonists and by lesions of the ipsilateral fornix caudal to
58
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
the site of stimulation. In addition, PPF experiments showed that stimulation of precommissural fibers did not directly activate the same neocortical/local circuit fibers that were activated by stimulating the rostral border of the NAc. Together, these data strongly suggest that the glutamatergic fornix afferents that originate from the hippocampal formation can be activated independent of neocortical afferents in this mouse brain slice preparation. Anatomical data has shown that the likely sources of the glutamatergic fornix fibers to the accumbens are the pyramidal neurons of the hippocampus and subiculum (Groenewegen et al., 1987). An important reason for developing this model is to facilitate future experiments designed to elucidate mechanisms of synaptic physiology and plasticity at the hippocampal/ accumbens synapse by using transgenic mice, as has been done at the limbic cortical/accumbens synapse (d’Alcantara et al., 2001; Mazzucchelli et al., 2002). 4.2. Medium spiny neuron up/down states NAc medium spiny neurons have been shown to oscillate in a stepwise manner between a hyperpolarized down state and a depolarized up state of the membrane potential (O’Donnell and Grace, 1995; Yim and Mogenson, 1988). These state changes appear to be critical for relaying afferent signals through the nucleus as only in the up state are neurons likely to fire action potentials in response to small sub threshold excitatory synaptic potentials. It is reasonable to suspect that plasticity is more easily induced when neurons are in the up state. Therefore, it is important to be able to control membrane potential states in a brain slice preparation to be used for the study of synaptic plasticity. Our data agree with previous in vitro studies that show no such spontaneous jumps to the up state occur in medium spiny neurons (Chang and Kitai, 1986; O’Donnell and Grace, 1993; Pennartz and Kitai, 1991; Yuan et al., 1992). Our new finding is that high frequency stimulation of fornix afferents (>20 Hz) is capable of depolarizing these neurons to a membrane potential that resembles an up state and that an up state-like depolarization is more likely to occur when stimulation produces EPSPs large enough to reach spike threshold. The depolarization response observed here is comparable to the response to high frequency fornix stimulation seen in anesthetized rats (O’Donnell and Grace, 1995). However, the depolarization response appears to differ from spontaneous up states seen in vivo in several important ways. Unlike in vitro, the up state can be induced, with a short delay, by a single fornix stimulus in anesthetized rats (O’Donnell and Grace, 1995). The onset and offset time courses of the stimulation-induced depolarization in vitro are not as rapid as spontaneous up/down state transitions in vivo, particularly the transition back to hyperpolarization. Finally, the brain slice preparation is missing a number of other afferents to the accumbens that may be important for normal control of state transitions. In particular, the afferents from the ventral tegmental area have been
shown to regulate state transitions when stimulated (Goto and O’Donnell, 2001). Further work is needed to define all the variables involved in reproducing in vitro, the up/down state transitions recorded in vivo, but we suggest that this new brain slice preparation can be used to model transitions to an upstate-like depolarized membrane potential that is dependent upon hippocampal input and can be used to study plasticity at the hippocampal/accumbens synapse. 4.3. Synaptic plasticity at the hippocampal/accumbens synapse Theta frequency (8 Hz) stimulation of hippocampal and neocortical afferents induced a moderate degree of LTD at the hippocampal-medium spiny neuron synapse (six of eight neurons), but not at the neocortical synapse in the same neurons. Theta frequency firing patterns are commonly observed in output neurons of the hippocampal formation. Theta frequency membrane oscillations giving rise to single spikes or burst firing have been recorded from pyramidal neurons in the ventral subiculum (Mattia et al., 1997). This suggests that theta frequency-dependent LTD may have behavioral relevance and may have been involved in the relapse to cocaine seeking behavior induced by theta burst stimulation of the ventral subiculum (Vorel et al., 2001). Previous in vivo field recordings in the NAc have shown that high frequency stimulation of hippocampal afferents resulted in a moderate degree of LTP (15–25% increase in field EPSP for up to 60 min (Boeijinga et al., 1993; Mulder et al., 1997,1998)), but no in vivo data are available on the long-term effects of low frequency (theta) stimulation. If theta frequency-dependent LTD also occurs in vivo, the strength of momentary excitatory hippocampal signals to accumbens neurons may be increased during LTP or decreased during LTD by previous rhythmic, synchronized output of the hippocampal formation. The lack of effect of low frequency stimulation on plasticity at the neocortical synapse as reported here contrasts with the LTD observed in brain slices reported by others following 5 or 13 Hz stimulation of neocortical afferents. However, it is important to note that the present results were obtained with ten 6 s trains of theta stimulation. To obtain LTD with 5 Hz stimulation, afferents were stimulated continuously for 3 min while blocking potassium channels with cesium and clamping the neuron soma at −50 mV (Thomas et al., 2001). This treatment likely enhanced calcium entry during stimulation as compared to theta stimulation without blocking potassium channels. Indeed, these investigators report that chelation of calcium with BAPTA blocked LTD, suggesting that short trains of theta stimulation alone does not increase postsynaptic calcium concentrations to the level needed for LTD at this synapse. Others obtained 13 Hz-dependent LTD after 10 min of continuous stimulation, again suggesting that 6 s trains of stimulation delivered over approximately 2 min is insufficient to induce LTD at this synapse (Robbe et al., 2002). In addition, the 13 Hz
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
LTD was observed with recordings in the rostral portion of the nucleus, a region of the nucleus that has been suggested to be anatomically and functionally different from either the accumbens core or shell (Brog et al., 1993; Voorn et al., 1989). Our experiments were done specifically in medium spiny neurons located in the dorsal/medial accumbens at the core/shell boundary. This raises the possibility that regional differences within the NAc control the presence or absence of particular types of synaptic plasticity at selective inputs. Indeed, evidence has been found for regional differences in plasticity within the dorsal striatum, a nucleus with very similar anatomy and physiology to the NAc (Winder et al., 2002). This selective plasticity could be a mechanism by which hippocampal-dependent learning and memory has long-term effects on behaviors mediated by the NAc. Additional experiments are needed to determine whether the plasticity that we report here is a unique feature of the simultaneous afferent stimulation protocol. Our decision to test for plasticity with simultaneous stimulation of two afferents arose from findings in our lab and others (Pennartz and Kitai, 1991; Robbe et al., 2002; Schramm et al., 2002) that plasticity is not always a robust or consistent phenomena at medium spiny synapses when stimulating only one afferent pathway in brain slices from adult animals, possibly due to the preference of these neurons for a hyperpolarized “down” state in slice preparations. Also our data on up state transitions in the brain slice suggest that even strong stimulation at high frequency may not fully depolarize neurons to a membrane potential that would permit enough NMDA channel-dependent calcium entry to initiate some types of plasticity. Therefore, we chose to maximize postsynaptic depolarization with dual afferent stimulation plus somatic current injection to facilitate common forms of plasticity that require calcium influx. Future experiments should test theta frequency stimulation of each afferent alone. Afferent input gating may also have influenced the selective expression of LTD reported here. For medium spiny accumbens neurons, stimulation of one afferent can suppress or enhance the synaptic effect of other afferents in a time dependent manner (Goto and O’Donnell, 2002; Mulder et al., 1998; O’Donnell and Grace, 1995). These gating mechanisms may have contributed to the results presented here, possibly by reducing the size of all EPSPs starting with the second stimulus pulse and therefore reducing postsynaptic activation of AMPA receptors and limiting the magnitude or occurrence of plasticity. In summary, an angled, sagittal mouse brain slice preparation has been designed to facilitate stimulation of two major glutamatergic afferent pathways to the NAc. One of the pathways originates from hippocampal formation neurons as confirmed by elimination of its ability to induce EPSPs following lesions of the fimbria/fornix fibers at the rostral end of the hippocampus. Medium spiny neurons at the medial core/shell boundary of the accumbens can be depolarized by stimulation of either pathway through activation of AMPA-type glutamate receptors. Repetitive stimulation of
59
hippocampal afferents is capable of depolarizing medium spiny neurons to a membrane potential that resembles the up state observed in vivo, an effect that appears to be dependent on stimulus frequency and magnitude of the stimulus-evoked EPSP. Finally, theta frequency stimulation of both afferents can selectively induce a decrease in the magnitude of stimulus-induced EPSPs on one afferent only, suggesting that this brain slice preparation can be used to study mechanisms underlying synaptic plasticity within the NAc.
Acknowledgements This research was supported by NINDS (SNRP program; RTM), Meharry Medical College (RTM, OC), Vanderbilt Intramural Discovery Award (RTM, DGW), NIAAA and the Whitehall Foundation (DGW).
References Boeijinga PH, Mulder AB, Pennartz CM, Manshanden I, Lopes da Silva FH. Responses of the nucleus accumbens following fornix/fimbria stimulation in the rat. Identification and long-term potentiation of mono- and polysynaptic pathways. Neuroscience 1993;53:1049–58. Brog JS, Salyapongse A, Deutch AY, Zahm DS. The patterns of afferent innervation of the core and shell in the accumbens part of the rat ventral striatum: immunohistochemical detection of retrogradely transported fluoro-gold. J Comp Neurol 1993;338:255–78. Chang HT, Kitai ST. Intracellular recordings from rat nucleus accumbens neurons in vitro. Brain Res 1986;366:392–6. Charara A, Grace AA. Dopamine receptor subtypes selectively modulate excitatory afferents from the hippocampus and amygdala to rat nucleus accumbens neurons. Neuropsychopharmacology 2003;28:1412–21. d’Alcantara P, Ledent C, Swillens S, Schiffmann SN. Inactivation of adenosine A2A receptor impairs long term potentiation in the accumbens nucleus without altering basal synaptic transmission. Neuroscience 2001;107:455–64. DeFrance JF, Marchand JF, Sikes RW, Chronister RB, Hubbard JI. Characterization of fimbria input to nucleus accumbens. J Neurophysiol 1985;54:1553–67. Drevets WC, Gautier C, Price JC, Kupfer DJ, Kinahan PE, Grace AA, et al. Amphetamine-induced dopamine release in human ventral striatum correlates with euphoria. Biol Psychiatr 2001;49:81–96. Everitt BJ, Wolf ME. Psychomotor stimulant addiction: a neural systems perspective. J Neurosci 2002;22:3312–20. Goto Y, O’Donnell P. Network synchrony in the nucleus accumbens in vivo. J Neurosci 2001;21:4498–504. Goto Y, O’Donnell P. Timing-dependent limbic-motor synaptic integration in the nucleus accumbens. Proc Natl Acad Sci USA 2002;99:13189– 93. Groenewegen HJ, Vermeulen-Van der Zee E, te Kortschot A, Witter MP. Organization of the projections from the subiculum to the ventral striatum in the rat. A study using anterograde transport of Phaseolus vulgaris leucoagglutinin. Neuroscience 1987;23:103–20. Hussain Z, Johnson LR, Totterdell S. A light and electron microscopic study of NADPH-diaphorase-, calretinin- and parvalbumin-containing neurons in the rat nucleus accumbens. J Chem Neuroanat 1996;10:19– 39. Hyman SE, Malenka RC. Addiction and the brain: the neurobiology of compulsion and its persistence. Nat Rev Neurosci 2001;2:695–703. Konsman JP. In: Paxinos G, Franklin KBJ, editors. The mouse brain in stereotaxic coordinates. 2nd ed. (Deluxe). New York: Academic
60
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
Press, 2001. ISBN 0-12-547637-X. Psychoneuroendocrinology, 2003; 28:827–28. Mattia D, Kawasaki H, Avoli M. Repetitive firing and oscillatory activity of pyramidal-like bursting neurons in the rat subiculum. Exp Brain Res 1997;114:507–17. Mazzucchelli C, Vantaggiato C, Ciamei A, Fasano S, Pakhotin P, Krezel W, et al. Knockout of ERK1 MAP kinase enhances synaptic plasticity in the striatum and facilitates striatal-mediated learning and memory. Neuron 2002;34:807–20. Mulder AB, Arts MP, Lopes da Silva FH. Short- and long-term plasticity of the hippocampus to nucleus accumbens and prefrontal cortex pathways in the rat, in vivo. Eur J Neurosci 1997;9:1603–11. Mulder AB, Hodenpijl MG, Lopes da Silva FH. Electrophysiology of the hippocampal and amygdaloid projections to the nucleus accumbens of the rat: convergence, segregation, and interaction of inputs. J Neurosci 1998;18:5095–102. Nicola SM, Surmeier J, Malenka RC. Dopaminergic modulation of neuronal excitability in the striatum and nucleus accumbens. Annu Rev Neurosci 2000;23:185–215. O’Donnell P, Grace AA. Physiological and morphological properties of accumbens core and shell neurons recorded in vitro. Synapse 1993;13:135–60. O’Donnell P, Grace AA. Synaptic interactions among excitatory afferents to nucleus accumbens neurons: hippocampal gating of prefrontal cortical input. J Neurosci 1995;15:3622–39. Onn SP, West AR, Grace AA. Dopamine-mediated regulation of striatal neuronal and network interactions. Trends Neurosci 2000;23:S48–56. Pennartz CM, Ameerun RF, Groenewegen HJ, Lopes da Silva FH. Synaptic plasticity in an in vitro slice preparation of the rat nucleus accumbens. Eur J Neurosci 1993;5:107–17. Pennartz CM, Dolleman-Van der Weel MJ, Kitai ST, Lopes da Silva FH. Presynaptic dopamine D1 receptors attenuate excitatory and inhibitory limbic inputs to the shell region of the rat nucleus accumbens studied in vitro. J Neurophysiol 1992;67:1325–34.
Pennartz CM, Groenewegen HJ, Lopes da Silva FH. The nucleus accumbens as a complex of functionally distinct neuronal ensembles: an integration of behavioural. Prog Neurobiol 1994;42:719– 61. Pennartz CM, Kitai ST. Hippocampal inputs to identified neurons in an in vitro slice preparation of the rat nucleus accumbens: evidence for feed-forward inhibition. J Neurosci 1991;11:2838–47. Robbe D, Alonso G, Chaumont S, Bockaert J, Manzoni OJ. Role of p/q-Ca2+ channels in metabotropic glutamate receptor 2/3-dependent presynaptic long-term depression at nucleus accumbens synapses. J Neurosci 2002;22:4346–56. Robbe D, Kopf M, Remaury A, Bockaert J, Manzoni OJ. Endogenous cannabinoids mediate long-term synaptic depression in the nucleus accumbens. Proc Natl Acad Sci USA 2002;99:8384–8. Schramm NL, Egli RE, Winder DG. LTP in the mouse nucleus accumbens is developmentally regulated. Synapse 2002;45:213–9. Thomas MJ, Beurrier C, Bonci A, Malenka RC. Long-term depression in the nucleus accumbens: a neural correlate of behavioral sensitization to cocaine. Nat Neurosci 2001;4:1217–23. Voorn P, Gerfen CR, Groenewegen HJ. Compartmental organization of the ventral striatum of the rat: immunohistochemical distribution of enkephalin, substance P. dopamine, and calcium-binding protein. J Comp Neurol 1989;289:189–201. Vorel SR, Liu X, Hayes RJ, Spector JA, Gardner EL. Relapse to cocaine-seeking after hippocampal theta burst stimulation. Science 2001;292:1175–8. Winder DG, Egli RE, Schramm NL, Matthews RT. Synaptic plasticity in drug reward circuitry. Curr Mol Med 2002;2:667–76. Yim CY, Mogenson GJ. Neuromodulatory action of dopamine in the nucleus accumbens: an in vivo intracellular study. Neuroscience 1988;26:403–15. Yuan XR, Madamba S, Siggins GR. Opioid peptides reduce synaptic transmission in the nucleus accumbens. Neurosci Lett 1992;134: 223–8.
A novel mouse brain slice preparation of the hippocampo–accumbens pathway Robert T. Matthews a,∗ , Olusegun Coker a , Danny G. Winder b b
a Department of Anatomy & Physiology, Meharry Medical College, Nashville, TN 37208, USA Department of Molecular Physiology & Biophysics, Center for Molecular Neuroscience, Kennedy Center, Vanderbilt University Medical School, Nashville, TN, USA
Received 5 November 2003; received in revised form 20 January 2004; accepted 3 February 2004
Abstract The nucleus accumbens (NAc) is an important component of circuitry that underlies reward related behaviors and the rewarding properties of drugs of abuse. Glutamatergic afferents to the nucleus are critical for its normal function and for behaviors related to drug addiction. An angled, sagittal mouse brain slice preparation has been designed to facilitate concurrent stimulation of two major glutamatergic afferent pathways to the nucleus accumbens. Medium spiny neurons at the medial core/shell boundary of the accumbens were depolarized by stimulation of either hippocampal or limbic cortical afferents through activation of AMPA-type glutamate receptors. High frequency but not low frequency stimulation of hippocampal afferents depolarized medium spiny neurons to a membrane potential that resembled the up state observed upon high frequency stimulation in vivo. The magnitude of the membrane depolarization was positively correlated with the amplitude of the stimulus-evoked EPSP. Concurrent stimulation of hippocampal and limbic cortical afferents at theta frequency selectively induced a long-term depression (LTD) in the magnitude of stimulus-evoked EPSPs on the hippocampal afferent only. These data suggest that this brain slice preparation can be used to study mechanisms underlying synaptic plasticity at two of the critical glutamatergic afferent synapses in the nucleus accumbens as well as characterizing potential interactions between afferents. Additionally, LTD at hippocampo–accumbens synapses can be induced at a stimulus frequency known to support reinstatement of drug seeking behavior. © 2004 Elsevier B.V. All rights reserved. Keywords: Glutamatergic synapse; Nucleus accumbens; Fornix; Long-term depression
1. Introduction The nucleus accumbens (NAc) has been shown to be an important component of the limbic circuitry involved in long-lasting behaviors induced by chronic exposure to drugs of abuse such as cocaine (Everitt and Wolf, 2002; Hyman and Malenka, 2001). Regulation of excitatory inputs that converge on the NAc is increasingly thought to be a key mechanism for modulation of long-term behavioral responses to drugs of abuse. The NAc receives glutamatergic inputs from a variety of sources, including the limbic neocortex, the ventral subiculum and hippocampus (hippocampal formation), the amygdala and the dorsal medial thalamus (Pennartz et al., 1994).
∗ Corresponding author. Tel.: +1-615-327-6288; fax: +1-615-327-6713. E-mail address: [email protected] (R.T. Matthews).
0165-0270/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jneumeth.2004.02.001
Several pieces of evidence suggest that the hippocampal input plays a key role in behaviors dependent on the NAc. First, stimulation of the hippocampal formation at theta frequency in vivo has been shown to cause reinstatement of cocaine seeking behavior in rats that had previously self-administered cocaine (Vorel et al., 2001). Second, the hippocampal input appears to play a critical role in modulating resting membrane potential of medium spiny neurons, the output neurons of the NAc (O’Donnell and Grace, 1995). Medium spiny neurons switch between hyperpolarized “down” and depolarized “up” states in vivo, and lesions of the fornix eliminate the up state. Finally, in vivo experiments have demonstrated long-term potentiation (LTP) at hippocampal formation/accumbens synapses following high frequency stimulation (Drevets et al., 2001; Onn et al., 2000). To date, most attempts to study the hippocampo–NAc pathway have utilized in vivo approaches. Thus, the mechanisms of up/down state transitions, high frequency-dependent LTP, the effect of theta frequency stimulation on synaptic
50
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
efficacy and the possibility of long-term depression (LTD) at this synapse are largely unknown. The development of an in vitro model for the study of this pathway has proven difficult, likely due to the serpentine route taken by the hippocampal efferents. In the present study, we have developed and verified a reliable method for the study of this synapse in vitro, and have performed a preliminary characterization of this synapse. We find that, through the use of an angled parasagittal brain slice, we can reliably evoke glutamatergic synaptic responses in NAc neurons in response to stimulation in or near the fornix. These responses are severely attenuated in fornix lesioned animals, and do not exhibit paired-pulse interactions with synaptic responses elicited by stimulating afferents entering the rostral end of the NAc. We find that theta frequency stimulation of these synapses results predominantly in LTD of the hippocampal input, with no change in the limbic neocortical input, suggesting that distinct forms of plasticity may exist at these synapses.
2. Materials and methods 2.1. Slice preparation and current clamp recording C57bl/6J male mice (Jackson Laboratories), 6–10 weeks old, were used for preparation of brain slices. Animal experiments were carried out in accordance with the National Institute of Health Guide for the Care and Use of Animals (NIH publication no. 80–23) revised in 1996. Mice were housed under standard conditions, 2–4 mice per cage with unlimited access to food and water. Except for lesion experiments, mice were anesthetized with isoflurane, brains quickly removed and chilled in ice-cold buffered salt solution. The whole brain was trimmed to expose an angled parasagittal surface (Fig. 1A), glued to a glass slide and cut with a vibratome to yield a single 400 m thick brain slice that contained the central portion of the NAc and most of the ipsilateral fornix (see Fig. 1B). The angle of the cut that best preserved functional hippocampal and neocortical afferents was approximately 35◦ from vertical. To reliably produce this angle, the brain was placed in a commercially available stainless steel sagittal brain slicing jig (Ted Pella, Inc.) with the brain placed at an angle in the jig such that the olfactory tubercle was just level with the top of the jig on one side prior to blocking. This procedure provided enough stability for angled trimming of the whole brain with razor blades such that usable vibratome brain slices were obtained from greater than 90% of mice. Slices were immediately transferred to an interface-type recording chamber (Fine Science Tools) and permitted to recover for 1–2 h at 29 ◦ C before beginning recordings at the same temperature. Sharp electrode current clamp recordings were performed using 90–150 M pipettes filled with 3 M K+ acetate. Recording electrodes were pulled on a Flaming/Brown Micropipette Puller (Sutter Instruments) using thick-walled filament-containing borosilicate glass capillaries. Voltage
signals were amplified with an Axoclamp 2B amplifier and data was collected and analyzed with pClamp 8.1 software. Drugs were applied by adding them to the perfusion solution reservoir. Flow rate was 2–3 ml/min resulting in a 4–6 min delay for drugs to enter the slice chamber. Perfusion solution consisted of NaCl 124 mM, KCl 4.4 mM, CaCl2 2.5 mM, NaH2 PO4 1 mM, MgSO4 1.2 mM, NaHCO3 26 mM, glucose 10 mM. Perfusion solutions were continuously bubbled with 95% O2 /5% CO2 . Picrotoxin (25 M) was present, unless otherwise noted, in all experiments in order to block GABA-A dependent IPSPs that were found in preliminary experiments to mask EPSPs in some cases. 2.2. Synaptic stimulation Stimulating electrodes were made from either twisted bipolar nichrome wire (0.002 in./50 m diameter) with minimal tip separation or bipolar tungsten wire (0.005 in./127 m diameter) with tip separation of about 250 m. Monopolar square wave pulses were delivered to wire electrodes with a Grass S88 stimulator and a Grass constant voltage stimulus isolation unit. Stimulating electrodes were positioned on the surface of the slice just rostral to the column of the fornix where the precommissural fibers enter the lateral septal nucleus (tungsten wire) and in the white matter at the rostral/dorsal border of the NAc (nichrome wire, Fig. 1B). These electrode placements were presumed to be stimulating hippocampal formation afferents and limbic neocortical inputs, respectively. However, stimulation of local accumbens fibers by the neocortical electrode cannot be ruled out because of its position at the border of the nucleus. Medium spiny neurons, located either in the accumbens core just dorsal to the anterior commissure, or in the adjacent medial–dorsal core/shell transition, were impaled with recording electrodes and allowed to stabilize for 10–15 min. In the rostral/caudal plane, recorded neurons were located at the midpoint of the nucleus to slightly caudal to the midpoint (Fig. 1B). These neurons were located in a region of the nucleus previously shown to have a relatively high density of afferents originating from the ventral subiculum (Groenewegen et al., 1987). Pulses between 80 and 200 s duration and 2–130 V were applied to stimulating electrodes in order to measure input/output curves for each afferent. For paired pulse facilitation (PPF) experiments, pulse pairs with inter-pulse intervals of 10–200 ms and about 50% of maximal amplitude were given on each input to find an interval that produced a significant facilitation with near complete return of membrane potential to resting level before the second pulse. Six pulse pairs with inter-pair intervals of 20 s were averaged for data comparisons. For pulse train stimulation designed to measure membrane depolarization, only the fornix stimulating electrode was used and stimulus strength was set to produce an EPSP at 80–100% of maximal EPSP size for that neuron. The magnitude of membrane depolarization was measured repeatedly during the stimulus train at time points just prior to the next stimulus artifact.
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
51
Fig. 1. (A) Coronal diagram from a mouse brain atlas (Konsman, 2003) at an intermediate rostral/caudal level of the NAc showing the angle of the sagittal section (between the dark lines) used for electrophysiological recording. The ventromedial (right-facing) surface was placed facing up in the recording chamber for proper placement of stimulating electrodes. (B) Photograph of a typical mouse brain slice with all electrodes in place. A tungsten-stimulating electrode is placed in the precommissural fornix fibers where they enter the septal nuclei. A nichrome-stimulating electrode is placed at the rostral/dorsal border of the NAc where fibers enter from the limbic neocortex. The tip of the recording electrode is indicated by the ∗ D, dorsal; R, rostral; Hippo, hippocampus.
52
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
2.3. Fornix lesions
2.6. Data analysis
Unilateral fornix lesions were done in a subset of mice. Mice were anesthetized with trichloroethylene and mounted in a stereotaxic apparatus. After deflecting the skin over the skull, a hole was drilled over the fornix at 0.9 mm posterior to bregma. A 2 mm wide knife made from a razor blade was lowered to a depth that cut the fornix on one side of the brain without damaging the underlying thalamus. The medial most portion of the fornix (≈0.3 mm) was spared in order to prevent bleeding from the sagittal sinus (Fig. 5). Mice were permitted to recover for 3–6 days and then anesthetized with isoflurane for generating brain slices as described above. Sham lesioned animals received the same treatment except the knife blade was not lowered into the brain or cut through only the overlying neocortex. On experiment days, one sham and one lesioned animal were used in a single blind protocol. NAc neurons were impaled and tested for minimum criteria of healthy neurons before and after input/output data were collected from stimulation of afferents as described for paired pulse experiments. Any neurons that failed to meet criteria were discarded. Criteria included firing threshold between −50 and −45 mV, action potential peak above 0 mV, multiple spiking during membrane depolarization above threshold and evidence of inward rectification during hyperpolarizing current pulses.
Appropriate statistical analyses (indicated within figure legends) were performed using Prizm and InStat software (Graphpad). Multiple regression analyses were done with Microcal Origin 7.0.
2.4. Synaptic plasticity Baseline EPSP amplitudes, about 50% of maximum and >4 mV, were recorded following sequential stimulation of limbic neocortical/local fiber inputs and hippocampal formation (fornix) inputs. Pairs of stimuli, one to each input, were separated by 200 ms and followed by hyperpolarizing pulses used to measure membrane resistance. The inter-pulse interval was long enough to avoid gating of one afferent by the other (O’Donnell and Grace, 1995). Pulse pairs were given every 20 s. Baseline data were collected for 20 min followed by stimulation trains at various frequencies given simultaneously on both inputs while current was injected into the neuron to maintain membrane potential near firing threshold (∼−55 mV). Theta frequency stimulation consisted of 10 trains of 8 Hz stimulation, 6 s per train with 7 s inter-train intervals, at the same stimulus intensity used for data collection before and after theta stimulation, but of opposite polarity. After theta stimulation, holding current was turned off and every 20 s. stimulus pairs were restarted. Any neuron whose membrane resistance changed by more than 15% over the next 90 min of data collection was discarded from further analysis.
3. Results 3.1. Mouse medium spiny neuron electrophysiology Medium spiny neurons were examined with sharp electrode current clamp recordings. Basic membrane properties recorded in the presence of picrotoxin resembled those reported previously for rat medium spiny neurons (Chang and Kitai, 1986; O’Donnell and Grace, 1993; Pennartz and Kitai, 1991). Mean action potential amplitude was 61.0mV± 1.7 (S.E.M.; n = 22), spike width at 1/3 above base was 1.30 ms ± 0.08, input resistance was 158 M ± 18 and resting membrane potential was −79.0 mV ± 2.0. Neurons fired single spikes or trains of spikes in response to membrane depolarization above −45 mV with no bursting (Fig. 2A). Stimulation of precommissural fornix fibers or neocortical fibers resulted in synaptic potentials in the majority of medium spiny neurons tested. In the absence of picrotoxin, synaptic responses were sometimes a mixture of IPSPs and EPSPs. Because the objective of this study was to characterize glutamatergic inputs to medium spiny neurons, IPSPs were routinely blocked with 25 M picrotoxin (data not shown). After blockade of IPSPs, monosynaptic EPSPs could be elicited from fornix fiber stimulation with typical latencies of 4.8–5.9 ms and from neocortical fibers with latencies of 4.2–5.0 ms. Fig. 2C shows the graded responses of a medium spiny neuron to stepped fornix fiber stimulation that produced an action potential at the highest stimulation voltage. 3.2. EPSPs are AMPA receptor dependent The selective AMPA receptor antagonists, CNQX (10 M, n = 3) and DNQX (10 M, n = 2) blocked EPSPs generated by stimulation of either fornix or neocortical fibers. Blockade was 90–100% complete in all cases, suggesting that at resting membrane potential, most of the current was carried by AMPA-type glutamate receptors. However, a small percentage of the current underlying the EPSPs from these sources might be controlled by other receptor types. Fig. 3A and B shows the effect of CNQX on fornix and neocortical induced EPSPs while Fig. 3C shows the effect of DNQX on the fornix-induced EPSPs in a different neuron.
2.5. Pharmacology 3.3. Paired pulse facilitation All drugs were bath applied. Picrotoxin, CNQX, and DNQX were purchased from Sigma (St. Louis, MO). DMSO (0.02%, v/v) was the carrier for picrotoxin and CNQX.
To begin to test whether the presumed fornix fiber and neocortical fiber stimulating electrodes were stimulating the
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
53
Fig. 2. Examples of recordings from mouse medium spiny neurons. (A) Superimposed voltage responses to 50 pA stepped current injections. Note the ramped depolarizing response (arrow) leading up to the action potential and the time-independent type of inward rectification to hyperpolarizing pulses that are typical of medium spiny neurons in other species. (B) I/V plot from the neuron shown in (A). (C) Superimposed responses of a different neuron to repeated stimulation of fornix fibers in 4 V steps starting at 60 V (0.1 ms pulses every 10 s). Arrows indicate stimulus artifact.
same or different afferent fibers, a paired pulse experiment was designed to examine facilitation of EPSP amplitude within and between electrode sites. Four medium spiny neurons from three mice were tested for the presence of PPF with stimulation on each electrode alone and with stimulation of one electrode followed by the other electrode to create the pulse pair. Pairs of stimulation pulses on either electrode produced significant facilitation of the second EPSP in one neuron (P < 0.01), but no significant facilitation was observed when pulse pairs were presented on alternating electrodes (Fig. 4). In three neurons, paired pulse facilitation was seen on only one input but no significant facilitation between electrodes was observed. These data suggest that our electrode placements are stimulating two different inputs. 3.4. EPSPs following ipsilateral fimbria/fornix lesions Unilateral lesions (n = 5) or sham lesions (n = 5) were performed on mice 3–6 days prior to creating brain slices. Lesions were located at the rostral tip of the hippocampus, approximately 1.2 mm caudal to the fornix/septal area boundary where the stimulating electrode was positioned in
the brain slice (Fig. 5, inset). Because of the angled nature of the brain slice, the lesion was not visible in the slice used for recordings and therefore recordings from sham versus lesioned animals could be done blind. Two to five medium spiny neurons in each brain slice were tested for the presence of EPSPs following stimulation of neocortical and presumed hippocampal afferents (total of 34 neurons). In four of the five slices from lesioned animals (13 neurons), no EPSPs were detected when the fornix was stimulated, even when stimulus intensity was increased to 130 V and 300 s. One slice from a lesioned animal had three neurons with EPSPs of up to 5 mV at the highest stimulus intensity. The mean maximal EPSP amplitude from all neurons in slices from lesioned animals was 2.0 mV±0.7 (Fig. 5). In contrast, typical EPSPs of 10 mV in amplitude or larger (mean = 15.2 mV ± 1.9) were observed in the same neurons when neocortical fibers were stimulated (Fig. 5). Eighteen neurons in brain slices from sham lesioned animals had typical EPSPs following stimulation of either electrode (11.7 ± 1.3 on fornix electrode; 14.9 ± 1.9 on neocortical electrode, Fig. 5). Statistical comparison of the mean amplitudes of EPSPs from all neurons tested showed a significant reduction in the
54
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
Fig. 3. The effects of two selective AMPA type glutamate receptor antagonists, DNQX (n = 2 neurons) and CNQX (n = 3 neurons), on EPSP amplitudes induced by stimulation of either the fornix fibers or neocortical fibers. (A) Plot of the sizes of fornix-induced EPSPs over time in a medium spiny neuron, before and after bath application of CNQX. (B) Plot of neocortical-induced EPSPs in the same neuron. (C) Averaged traces of the fornix-induced EPSPs (n = 70 for each trace) from a different neuron before drug application (control), 15–25 min after adding DNQX and 10–20 min after drug washout. The input resistance did not change (data not shown).
Fig. 4. Data from a medium spiny neuron tested for paired pulse facilitation within inputs (A) and across inputs (B). (A) Pairs of stimulation pulses separated by 80 ms applied to the fornix electrode (100 s/63 V) produced significant facilitation of the second EPSP (paired Student’s t-test) as did pairs of stimuli applied to the local/neocortical electrode (50 s/60 V). (B) When pulse pairs were presented on alternating electrodes, no significant facilitation was observed. No picrotoxin was present in this experiment.
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
Fig. 5. Effects of a fornix lesion or sham lesion on the maximum EPSP size that could be recorded with stimulation of each input. Stimulus intensity was 100 s in duration and up to 120 V, or up to a voltage that induced an action potential. Only the fornix fiber dependent EPSP was significantly reduced by a prior lesion (unpaired Student’s t-test). Inset: Diagram from mouse brain atlas at the level just rostral to the hippocampus. The hatched area on the left side of the diagram indicates the location and extent of the fimbria/fornix lesion. Lesion placement was verified in three practice animals by coronal sectioning of the brain (data not shown). The lesion location was at least 1 mm caudal to the placement of stimulating electrodes for activation of fornix fibers. In addition to the fornix, the overlying neocortex and corpus callosum were also lesioned.
size of the EPSPs generated by the fornix electrode in lesioned animals as compared to sham animals (P < 0.001). These data strongly suggest that stimulation of tissue at the fornix/septal area boundary in angled sagittal brain slices selectively activates accumbens afferents originating from the hippocampal formation. 3.5. Control of up/down states by fornix stimulation Previous in vivo studies have demonstrated that the hippocampal input to the NAc plays a critical role in the initiation of up state transitions, and that single pulse stimulation of this input was sufficient to initiate an up state. Surprisingly, we find that single pulse stimulation of fornix fibers was not sufficient to induce medium spiny neurons to switch to an up state in our in vitro preparation, even when the EPSP was sufficient to induce an action potential (see Fig. 2C as an example). Since the up state may be a necessary component of the mechanisms underlying plasticity, especially calcium and/or NMDA receptor dependent forms of plasticity, we tested a variety of train frequencies in an attempt to induce an up state in medium spiny neurons in vitro. Eleven medium spiny neurons were recorded while stimulating fornix fibers at train frequencies of 5–100 Hz for 1 s or more (Fig. 6C and D). As with all neurons in this
55
study, resting membrane potential before the stimulus train was equivalent to the down state (<−75 mV). Each neuron received 1–4 stimulus trains of different frequencies, separated by at least 5 min. Picrotoxin was present in two of the eleven experiments, however data from these two experiments were not different from data without picrotoxin and were therefore included in the analysis. In order to compare membrane responses across different stimulus frequencies, maximum membrane depolarization during the first 1 s of the train was plotted against train frequency (Fig. 6A). The membrane depolarization was roughly correlated with train frequency (R2 = 0.57, F = 28.34, P > 0.001). A depolarization to the approximate level of the up state observed in vivo, −55 to −65 mV, was seen only at stimulus frequencies above 20 Hz. However, the size of the maximal EPSP varied greatly between neurons in this experiment (range of 2–30 mV). Therefore, a multiple regression analysis with amount of membrane depolarization as the dependent variable, and train frequency and EPSP size as independent variables was done (Fig. 6B). A better correlation between degree of membrane depolarization and both EPSP size and train frequency was observed (R2 = 0.71, F = 24.85, P > 0.001) as compared to train frequency alone. 3.6. Synaptic plasticity with theta frequency stimulation We next tested the hypothesis that synaptic plasticity can be induced at the hippocampal afferent/medium spiny neuron synapse in vitro. Fig. 7A shows one neuron’s averaged baseline EPSPs to stimulation of neocortical fibers followed by stimulation of fornix fibers, with enough time between stimulation of each electrode to allow the membrane potential to return to baseline. Using this protocol, we chose to test the effect of trains of theta frequency stimulation on EPSP amplitude because it is known that subpopulations of hippocampal formation neurons frequently fire bursts of action potentials in the theta frequency range, and because theta stimulation of the subiculum has been shown to reinstate cocaine seeking behavior in rats (Vorel et al., 2001). Ten trains, 7 s each, of 8 Hz stimulation (6 s inter-train interval) applied to both electrodes simultaneously resulted in a lasting depression of the fornix-induced EPSP in six of eight neurons (Fig. 7B) while no significant change was observed in the neocortical-induced EPSP (Fig. 7C). Two neurons showed a non-significant trend toward an LTP of the fornix-induced EPSP after this stimulus protocol. These data suggest that theta frequency stimulation can selectively induce long-term depression of fornix-dependent but not limbic cortical-dependent EPSPs recorded in medium spiny neurons of the mouse NAc.
4. Discussion One of the major goals of this study was to create a brain slice preparation that would keep intact a large portion of
56
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
Fig. 6. Correlations of fornix stimulation frequency and fornix-induced EPSP size on the degree of membrane depolarization of medium spiny accumbens neurons. (A) Plot of stimulation frequency vs. change in membrane potential (r 2 = 0.57; linear regression analysis of variance). Each open circle represents one neuron’s response after a single stimulus train. The dotted lines indicate the range of action potential thresholds for medium spiny neurons. The circles marked with ‘x’ are representative responses to theta frequency and high frequency stimulation that are shown in parts (C and D), respectively. (B) A 3D plot of EPSP size vs. stimulation frequency vs. change in membrane potential (R2 = 0.71; multiple regression analysis of variance). Each filled dot represents data plotted in (A) with the added dimension of EPSP size. (C and D) Voltage traces from the same neuron that show the effects of 2 s of theta (8 Hz) stimulation (C) or 1 s of 50 Hz stimulation (D) of the fornix on membrane potential. The thick bar represents the time course of the stimulus train. The arrow indicates an action potential that was generated by the first stimulus pulse of the train. Stimulus artifacts are truncated by filtering the signals with an eight pole Bessel filter set at low pass 500 Hz.
the NAc along with two of its major glutamatergic afferents, one from the hippocampal formation and one from the limbic neocortex. A second goal was to further characterize the physiology of the hippocampal afferent to the NAc, including postsynaptic receptor mechanisms and interactions between the hippocampal inputs and glutamatergic inputs from neocortical afferents. Specifically, we chose to look at the effects of afferent stimulation on medium spiny neurons because of previous work suggesting a critical role for these neurons in plasticity that may underlie reinstatement of drug seeking behavior (Everitt and Wolf, 2002; Nicola et al., 2000; Winder et al., 2002).
4.1. Brain slice preparation We have found that an angled parasagittal slice preparation of mouse forebrain can preserve a 400 m thick section of the accumbens along with enough of the afferents originating from the frontal cortex and the hippocampal formation to obtain robust synaptic potentials in medium spiny neurons from both sources. Parasagittal preparations from rat brain have been previously reported to also be appropriate for recording limbic cortical and fornix-induced EPSPs but we were not able to reliably obtain EPSPs from the fornix in mouse brain slices without changing the angle of the sagit-
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
57
Fig. 7. Selective LTD-like response at fornix synapses. (A) Trace averages of 100 responses following sequential local/neocortical fiber and fornix fiber stimulation from an example neuron. EPSPs are approximately 60% (0.09 ms/20 V) and 80% (0.09 ms/50 V) of maximal responses in this neuron, respectively. Note the short latency to initiation of the EPSPs (range for all neurons tested was 5–6 ms for the fornix and 3–5 ms for limbic cortical fiber simulation). Stimulus pairs were presented every 20 s. This stimulation protocol, along with a small hyperpolarizing pulse to monitor membrane resistance (not shown), was typically used to measure synaptic plasticity after tetanic stimulation of inputs. Open arrows indicate stimulus artifacts. (B) Plot of averaged data (±S.E.M.) showing the effect of theta stimulation on fornix-induced EPSPs in six of eight neurons that showed a sustained decrease in the size of the EPSP (group average EPSP size 45–50 min after theta stimulation was significantly smaller than the group average of 5 min just prior to stimulation; paired Student’s t-test, P > 0.01). Two neurons showed a non-significant trend toward LTP (data not shown). For graphing, data points were normalized to the mean of the EPSP amplitudes for the 5 min period immediately prior to theta stimulation. Arrow indicates point of interruption of data collection (approx. 2.5 min) for application of trains of theta stimulation (see Section 2 for details). (C) Averaged effect of theta stimulation on limbic cortical-induced EPSPs in five of the eight neurons tested for fornix input plasticity in (B) (other three neurons were tested only for fornix input plasticity). (D) Averaged effect of theta stimulation on input resistance in the six neurons shown in (B). In (B and C), superimposed averaged (n = 15) EPSP traces from one neuron just before and 45 min after theta stimulation are shown as insets.
tal cut (Pennartz and Kitai, 1991; Charara and Grace, 2003; Pennartz et al., 1992,1993). Synaptic potentials could be obtained from neurons throughout the nucleus, including from neurons located in the dorsal/medial core/shell boundary, the part of the nucleus that has been shown to have the densest plexus of fibers originating from the hippocampal formation (Groenewegen et al., 1987) and are most responsive to fornix stimulation in vivo (DeFrance et al., 1985). Stimulation of precommissural fornix fibers at the point where they exit the fornix adjacent to the septal nuclei resulted in IPSP/EPSP
complexes of variable amplitude. IPSPs were not observed in the presence of picrotoxin, suggesting GABA-A receptor dependent currents were responsible. The sources of IPSPs are unclear but could include afferents from neurons in the lateral septal nucleus or polysynaptic pathways involving GABAergic neurons within the NAc (Brog et al., 1993; Hussain et al., 1996). EPSPs generated from stimulation of precommissural fornix fibers were substantially blocked by AMPA receptor antagonists and by lesions of the ipsilateral fornix caudal to
58
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
the site of stimulation. In addition, PPF experiments showed that stimulation of precommissural fibers did not directly activate the same neocortical/local circuit fibers that were activated by stimulating the rostral border of the NAc. Together, these data strongly suggest that the glutamatergic fornix afferents that originate from the hippocampal formation can be activated independent of neocortical afferents in this mouse brain slice preparation. Anatomical data has shown that the likely sources of the glutamatergic fornix fibers to the accumbens are the pyramidal neurons of the hippocampus and subiculum (Groenewegen et al., 1987). An important reason for developing this model is to facilitate future experiments designed to elucidate mechanisms of synaptic physiology and plasticity at the hippocampal/ accumbens synapse by using transgenic mice, as has been done at the limbic cortical/accumbens synapse (d’Alcantara et al., 2001; Mazzucchelli et al., 2002). 4.2. Medium spiny neuron up/down states NAc medium spiny neurons have been shown to oscillate in a stepwise manner between a hyperpolarized down state and a depolarized up state of the membrane potential (O’Donnell and Grace, 1995; Yim and Mogenson, 1988). These state changes appear to be critical for relaying afferent signals through the nucleus as only in the up state are neurons likely to fire action potentials in response to small sub threshold excitatory synaptic potentials. It is reasonable to suspect that plasticity is more easily induced when neurons are in the up state. Therefore, it is important to be able to control membrane potential states in a brain slice preparation to be used for the study of synaptic plasticity. Our data agree with previous in vitro studies that show no such spontaneous jumps to the up state occur in medium spiny neurons (Chang and Kitai, 1986; O’Donnell and Grace, 1993; Pennartz and Kitai, 1991; Yuan et al., 1992). Our new finding is that high frequency stimulation of fornix afferents (>20 Hz) is capable of depolarizing these neurons to a membrane potential that resembles an up state and that an up state-like depolarization is more likely to occur when stimulation produces EPSPs large enough to reach spike threshold. The depolarization response observed here is comparable to the response to high frequency fornix stimulation seen in anesthetized rats (O’Donnell and Grace, 1995). However, the depolarization response appears to differ from spontaneous up states seen in vivo in several important ways. Unlike in vitro, the up state can be induced, with a short delay, by a single fornix stimulus in anesthetized rats (O’Donnell and Grace, 1995). The onset and offset time courses of the stimulation-induced depolarization in vitro are not as rapid as spontaneous up/down state transitions in vivo, particularly the transition back to hyperpolarization. Finally, the brain slice preparation is missing a number of other afferents to the accumbens that may be important for normal control of state transitions. In particular, the afferents from the ventral tegmental area have been
shown to regulate state transitions when stimulated (Goto and O’Donnell, 2001). Further work is needed to define all the variables involved in reproducing in vitro, the up/down state transitions recorded in vivo, but we suggest that this new brain slice preparation can be used to model transitions to an upstate-like depolarized membrane potential that is dependent upon hippocampal input and can be used to study plasticity at the hippocampal/accumbens synapse. 4.3. Synaptic plasticity at the hippocampal/accumbens synapse Theta frequency (8 Hz) stimulation of hippocampal and neocortical afferents induced a moderate degree of LTD at the hippocampal-medium spiny neuron synapse (six of eight neurons), but not at the neocortical synapse in the same neurons. Theta frequency firing patterns are commonly observed in output neurons of the hippocampal formation. Theta frequency membrane oscillations giving rise to single spikes or burst firing have been recorded from pyramidal neurons in the ventral subiculum (Mattia et al., 1997). This suggests that theta frequency-dependent LTD may have behavioral relevance and may have been involved in the relapse to cocaine seeking behavior induced by theta burst stimulation of the ventral subiculum (Vorel et al., 2001). Previous in vivo field recordings in the NAc have shown that high frequency stimulation of hippocampal afferents resulted in a moderate degree of LTP (15–25% increase in field EPSP for up to 60 min (Boeijinga et al., 1993; Mulder et al., 1997,1998)), but no in vivo data are available on the long-term effects of low frequency (theta) stimulation. If theta frequency-dependent LTD also occurs in vivo, the strength of momentary excitatory hippocampal signals to accumbens neurons may be increased during LTP or decreased during LTD by previous rhythmic, synchronized output of the hippocampal formation. The lack of effect of low frequency stimulation on plasticity at the neocortical synapse as reported here contrasts with the LTD observed in brain slices reported by others following 5 or 13 Hz stimulation of neocortical afferents. However, it is important to note that the present results were obtained with ten 6 s trains of theta stimulation. To obtain LTD with 5 Hz stimulation, afferents were stimulated continuously for 3 min while blocking potassium channels with cesium and clamping the neuron soma at −50 mV (Thomas et al., 2001). This treatment likely enhanced calcium entry during stimulation as compared to theta stimulation without blocking potassium channels. Indeed, these investigators report that chelation of calcium with BAPTA blocked LTD, suggesting that short trains of theta stimulation alone does not increase postsynaptic calcium concentrations to the level needed for LTD at this synapse. Others obtained 13 Hz-dependent LTD after 10 min of continuous stimulation, again suggesting that 6 s trains of stimulation delivered over approximately 2 min is insufficient to induce LTD at this synapse (Robbe et al., 2002). In addition, the 13 Hz
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
LTD was observed with recordings in the rostral portion of the nucleus, a region of the nucleus that has been suggested to be anatomically and functionally different from either the accumbens core or shell (Brog et al., 1993; Voorn et al., 1989). Our experiments were done specifically in medium spiny neurons located in the dorsal/medial accumbens at the core/shell boundary. This raises the possibility that regional differences within the NAc control the presence or absence of particular types of synaptic plasticity at selective inputs. Indeed, evidence has been found for regional differences in plasticity within the dorsal striatum, a nucleus with very similar anatomy and physiology to the NAc (Winder et al., 2002). This selective plasticity could be a mechanism by which hippocampal-dependent learning and memory has long-term effects on behaviors mediated by the NAc. Additional experiments are needed to determine whether the plasticity that we report here is a unique feature of the simultaneous afferent stimulation protocol. Our decision to test for plasticity with simultaneous stimulation of two afferents arose from findings in our lab and others (Pennartz and Kitai, 1991; Robbe et al., 2002; Schramm et al., 2002) that plasticity is not always a robust or consistent phenomena at medium spiny synapses when stimulating only one afferent pathway in brain slices from adult animals, possibly due to the preference of these neurons for a hyperpolarized “down” state in slice preparations. Also our data on up state transitions in the brain slice suggest that even strong stimulation at high frequency may not fully depolarize neurons to a membrane potential that would permit enough NMDA channel-dependent calcium entry to initiate some types of plasticity. Therefore, we chose to maximize postsynaptic depolarization with dual afferent stimulation plus somatic current injection to facilitate common forms of plasticity that require calcium influx. Future experiments should test theta frequency stimulation of each afferent alone. Afferent input gating may also have influenced the selective expression of LTD reported here. For medium spiny accumbens neurons, stimulation of one afferent can suppress or enhance the synaptic effect of other afferents in a time dependent manner (Goto and O’Donnell, 2002; Mulder et al., 1998; O’Donnell and Grace, 1995). These gating mechanisms may have contributed to the results presented here, possibly by reducing the size of all EPSPs starting with the second stimulus pulse and therefore reducing postsynaptic activation of AMPA receptors and limiting the magnitude or occurrence of plasticity. In summary, an angled, sagittal mouse brain slice preparation has been designed to facilitate stimulation of two major glutamatergic afferent pathways to the NAc. One of the pathways originates from hippocampal formation neurons as confirmed by elimination of its ability to induce EPSPs following lesions of the fimbria/fornix fibers at the rostral end of the hippocampus. Medium spiny neurons at the medial core/shell boundary of the accumbens can be depolarized by stimulation of either pathway through activation of AMPA-type glutamate receptors. Repetitive stimulation of
59
hippocampal afferents is capable of depolarizing medium spiny neurons to a membrane potential that resembles the up state observed in vivo, an effect that appears to be dependent on stimulus frequency and magnitude of the stimulus-evoked EPSP. Finally, theta frequency stimulation of both afferents can selectively induce a decrease in the magnitude of stimulus-induced EPSPs on one afferent only, suggesting that this brain slice preparation can be used to study mechanisms underlying synaptic plasticity within the NAc.
Acknowledgements This research was supported by NINDS (SNRP program; RTM), Meharry Medical College (RTM, OC), Vanderbilt Intramural Discovery Award (RTM, DGW), NIAAA and the Whitehall Foundation (DGW).
References Boeijinga PH, Mulder AB, Pennartz CM, Manshanden I, Lopes da Silva FH. Responses of the nucleus accumbens following fornix/fimbria stimulation in the rat. Identification and long-term potentiation of mono- and polysynaptic pathways. Neuroscience 1993;53:1049–58. Brog JS, Salyapongse A, Deutch AY, Zahm DS. The patterns of afferent innervation of the core and shell in the accumbens part of the rat ventral striatum: immunohistochemical detection of retrogradely transported fluoro-gold. J Comp Neurol 1993;338:255–78. Chang HT, Kitai ST. Intracellular recordings from rat nucleus accumbens neurons in vitro. Brain Res 1986;366:392–6. Charara A, Grace AA. Dopamine receptor subtypes selectively modulate excitatory afferents from the hippocampus and amygdala to rat nucleus accumbens neurons. Neuropsychopharmacology 2003;28:1412–21. d’Alcantara P, Ledent C, Swillens S, Schiffmann SN. Inactivation of adenosine A2A receptor impairs long term potentiation in the accumbens nucleus without altering basal synaptic transmission. Neuroscience 2001;107:455–64. DeFrance JF, Marchand JF, Sikes RW, Chronister RB, Hubbard JI. Characterization of fimbria input to nucleus accumbens. J Neurophysiol 1985;54:1553–67. Drevets WC, Gautier C, Price JC, Kupfer DJ, Kinahan PE, Grace AA, et al. Amphetamine-induced dopamine release in human ventral striatum correlates with euphoria. Biol Psychiatr 2001;49:81–96. Everitt BJ, Wolf ME. Psychomotor stimulant addiction: a neural systems perspective. J Neurosci 2002;22:3312–20. Goto Y, O’Donnell P. Network synchrony in the nucleus accumbens in vivo. J Neurosci 2001;21:4498–504. Goto Y, O’Donnell P. Timing-dependent limbic-motor synaptic integration in the nucleus accumbens. Proc Natl Acad Sci USA 2002;99:13189– 93. Groenewegen HJ, Vermeulen-Van der Zee E, te Kortschot A, Witter MP. Organization of the projections from the subiculum to the ventral striatum in the rat. A study using anterograde transport of Phaseolus vulgaris leucoagglutinin. Neuroscience 1987;23:103–20. Hussain Z, Johnson LR, Totterdell S. A light and electron microscopic study of NADPH-diaphorase-, calretinin- and parvalbumin-containing neurons in the rat nucleus accumbens. J Chem Neuroanat 1996;10:19– 39. Hyman SE, Malenka RC. Addiction and the brain: the neurobiology of compulsion and its persistence. Nat Rev Neurosci 2001;2:695–703. Konsman JP. In: Paxinos G, Franklin KBJ, editors. The mouse brain in stereotaxic coordinates. 2nd ed. (Deluxe). New York: Academic
60
R.T. Matthews et al. / Journal of Neuroscience Methods 137 (2004) 49–60
Press, 2001. ISBN 0-12-547637-X. Psychoneuroendocrinology, 2003; 28:827–28. Mattia D, Kawasaki H, Avoli M. Repetitive firing and oscillatory activity of pyramidal-like bursting neurons in the rat subiculum. Exp Brain Res 1997;114:507–17. Mazzucchelli C, Vantaggiato C, Ciamei A, Fasano S, Pakhotin P, Krezel W, et al. Knockout of ERK1 MAP kinase enhances synaptic plasticity in the striatum and facilitates striatal-mediated learning and memory. Neuron 2002;34:807–20. Mulder AB, Arts MP, Lopes da Silva FH. Short- and long-term plasticity of the hippocampus to nucleus accumbens and prefrontal cortex pathways in the rat, in vivo. Eur J Neurosci 1997;9:1603–11. Mulder AB, Hodenpijl MG, Lopes da Silva FH. Electrophysiology of the hippocampal and amygdaloid projections to the nucleus accumbens of the rat: convergence, segregation, and interaction of inputs. J Neurosci 1998;18:5095–102. Nicola SM, Surmeier J, Malenka RC. Dopaminergic modulation of neuronal excitability in the striatum and nucleus accumbens. Annu Rev Neurosci 2000;23:185–215. O’Donnell P, Grace AA. Physiological and morphological properties of accumbens core and shell neurons recorded in vitro. Synapse 1993;13:135–60. O’Donnell P, Grace AA. Synaptic interactions among excitatory afferents to nucleus accumbens neurons: hippocampal gating of prefrontal cortical input. J Neurosci 1995;15:3622–39. Onn SP, West AR, Grace AA. Dopamine-mediated regulation of striatal neuronal and network interactions. Trends Neurosci 2000;23:S48–56. Pennartz CM, Ameerun RF, Groenewegen HJ, Lopes da Silva FH. Synaptic plasticity in an in vitro slice preparation of the rat nucleus accumbens. Eur J Neurosci 1993;5:107–17. Pennartz CM, Dolleman-Van der Weel MJ, Kitai ST, Lopes da Silva FH. Presynaptic dopamine D1 receptors attenuate excitatory and inhibitory limbic inputs to the shell region of the rat nucleus accumbens studied in vitro. J Neurophysiol 1992;67:1325–34.
Pennartz CM, Groenewegen HJ, Lopes da Silva FH. The nucleus accumbens as a complex of functionally distinct neuronal ensembles: an integration of behavioural. Prog Neurobiol 1994;42:719– 61. Pennartz CM, Kitai ST. Hippocampal inputs to identified neurons in an in vitro slice preparation of the rat nucleus accumbens: evidence for feed-forward inhibition. J Neurosci 1991;11:2838–47. Robbe D, Alonso G, Chaumont S, Bockaert J, Manzoni OJ. Role of p/q-Ca2+ channels in metabotropic glutamate receptor 2/3-dependent presynaptic long-term depression at nucleus accumbens synapses. J Neurosci 2002;22:4346–56. Robbe D, Kopf M, Remaury A, Bockaert J, Manzoni OJ. Endogenous cannabinoids mediate long-term synaptic depression in the nucleus accumbens. Proc Natl Acad Sci USA 2002;99:8384–8. Schramm NL, Egli RE, Winder DG. LTP in the mouse nucleus accumbens is developmentally regulated. Synapse 2002;45:213–9. Thomas MJ, Beurrier C, Bonci A, Malenka RC. Long-term depression in the nucleus accumbens: a neural correlate of behavioral sensitization to cocaine. Nat Neurosci 2001;4:1217–23. Voorn P, Gerfen CR, Groenewegen HJ. Compartmental organization of the ventral striatum of the rat: immunohistochemical distribution of enkephalin, substance P. dopamine, and calcium-binding protein. J Comp Neurol 1989;289:189–201. Vorel SR, Liu X, Hayes RJ, Spector JA, Gardner EL. Relapse to cocaine-seeking after hippocampal theta burst stimulation. Science 2001;292:1175–8. Winder DG, Egli RE, Schramm NL, Matthews RT. Synaptic plasticity in drug reward circuitry. Curr Mol Med 2002;2:667–76. Yim CY, Mogenson GJ. Neuromodulatory action of dopamine in the nucleus accumbens: an in vivo intracellular study. Neuroscience 1988;26:403–15. Yuan XR, Madamba S, Siggins GR. Opioid peptides reduce synaptic transmission in the nucleus accumbens. Neurosci Lett 1992;134: 223–8.