| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
NEUROSCIENCE |
1 Department of Anatomy and Neurobiology, University of Tennessee Health Science Center, Memphis, TN 38163, USA
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
(Received 27 October 2006;
accepted after revision 26 February 2007;
first published online 1 March 2007)
Corresponding author W. E. Armstrong: Department of Anatomy and Neurobiology, University of Tennessee Health Science Center, Memphis, TN 38163, USA. Email: warmstrong{at}utmem.edu
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In lactating rats, OT neurons show highly stereotyped, synchronized milk ejection bursts of 50–80 Hz for 1–4 s (Poulain & Wakerley, 1982). This short burst is superimposed on a slower background of spike discharge ranging from < 1 to 
10 Hz, and ranging in pattern from irregular to regular (a.k.a. fast continuous). Recent studies suggest that the probability of milk ejection bursts is related to background discharge, such that a background firing rate of 1–3 Hz appears critical, and the degree of firing rate variability is positively correlated with burst probability (Brown & Moos, 1997; Brown et al. 2000; Moos et al. 2004). In response to other stimuli, in either male or female rats, OT neurons typically increase their firing rate and adopt a continuous activity pattern, not unlike that which can be observed between milk ejection bursts. Although OT release during milk injection is dependent on glutamate receptor activation (Parker & Crowley, 1993), and GABAergic activity strongly influences OT neuron discharge patterns (Moos, 1995; Leng et al. 2001), the precise role of synaptic activity in spike distribution is unknown.
Vasopressin neurons can also fire in three modes, including a slow irregular (< 1 Hz), fast continuous (> 1 Hz) and a phasic bursting firing (Poulain & Wakerley, 1982; Poulain et al. 1988). The latter two modes, especially phasic bursting, are adopted during VP release in response to stimuli such as dehydration and hypovolaemia. In contrast to OT neurons, this bursting is not synchronized among cells and is composed of alternating periods of activity (5–15 Hz) and silence, each lasting 20–40 s. Phasic bursting has been studied in vitro and is associated with intrinsic membrane properties such as the calcium-dependent depolarizing spike after-potential and its related plateau potential (Bourque, 1986; Renaud & Bourque, 1991). Nevertheless, in vivo, phasic bursting activity of VP neurons is dependent upon excitatory synaptic transmission (Nissen et al. 1995; Brown et al. 2004), and recent studies suggest that spike patterning within bursts of phasic neurons differs between in vivo and in vitro preparations (Sabatier et al. 2004).
Many in vivo patterns of activity of OT and VP neurons can be observed in hypothalamic slices or the hypothalamo-neurohypophysial explant in vitro, including phasic, continuous and slow irregular activity (Renaud & Bourque, 1991; Armstrong, 1995). Recent reports even demonstrate milk ejection-like activity in pharmacologically treated slices from lactating rats and male rats (Wang & Hatton, 2004, 2005) and in organotypic cultures (Jourdain et al. 1998; Israel et al. 2003). In the present study, we investigated how spontaneous excitatory and inhibitory synaptic activity contribute to the distribution of interspike intervals of identified OT and VP neurons in slices from adult virgin female rats, whether both cell types were equally influenced by this activity, and whether excitatory and inhibitory activity were equally influential. Some of these data have previously been published in abstract form (Li & Armstrong, 2002, 2003).
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Female, virgin Sprague–Dawley rats (150–250 g), taken randomly throughout the oestrus cycle, were deeply anaesthetized with sodium pentobarbitone (50 mg kg–1, I.P.) and perfused through the left ventricle with 60 ml of ice-cold, low-Na+ (replacing 125 mM NaCl with 210 mM sucrose) artificial cerebrospinal fluid (ACSF), which had been fully oxygenated (95% O2–5% CO2). The rat was decapitated and the brain removed to a slush of the low-Na+ ACSF. Coronal slices (250 µm) were cut through the hypothalamus with a vibrating-blade microtome (VT 1000S, Leica) with the tissue submerged in this slush, which was also fully oxygenated (95% O2–5% CO2), then placed on a nylon net in a retaining chamber containing oxygenated, normal ACSF. The ACSF composition was as follows (mM): KCl, 2.5; MgSO4, 1.0; NaH2PO4, 1.25; NaHCO3, 26; D-glucose, 20; ascorbic acid, 0.45; CaCl2, 2.0; and NaCl, 125 (osmolality, 290–310 mosmol kg–1; pH 7.3–7.4). After incubation at 32–34°C for 45–60 min, the slices were stored at room temperature (22–24°C) in oxygenated ACSF until used. Recordings were made at 32–34°C, with a perfusion rate of 1.5–2.0 ml min–1.
Drug application
Picrotoxin (100 µM), a blocker of Cl– channels as found in GABAA receptors, or the more selective GABAA receptor antagonist gabazine (10 µM) was selected to block inhibitory transmission. The glutamate receptor antagonist 6,7-dinitro-quinoxaline-2,3(1H,4H)-dione (DNQX, 10 µM) was used to blocked AMPA/kainate receptors, and kynurenic acid (2 mM), a broad-spectrum glutamate receptor blocker, was used to block both N-methyl-D-aspartate (NMDA) and non-NMDA receptor activity. Tetrodotoxin (TTX; 0.5 µM) was used to block voltage-gated sodium channels and thus the action potentials. All drugs were purchased from Sigma/RBI, and applied to the ACSF perifusate.
Whole-cell electrode recording
Recording pipettes (5–7 M
) were pulled from thin-walled borosilicate capillary tubing (o.d. = 1.5 mm, i.d. = 1.17 mm, Warner Instrument Corp.) using a model P-80/PC Flaming-Brown horizontal micropipette puller (Sutter Instrument Co.). To reduce capacitative artifacts, beeswax was applied to the pipette shank. The patch solution consisted of (mM): potassium gluconate, 140; KCl, 10; Hepes, 10; MgATP, 4; Na-GTP, 0.3; phosphocreatine, 3.5; EGTA, 0.2 (osmolality, 285 mosmol kg–1; pH 7.2–7.3). The measured liquid junction potential using this solution was +10 mV, and the data were corrected for this offset. A second patch solution consisted of (mM): CsCl, 120; Hepes, 30; EGTA, 11; MgCl2, 2; CaCl2, 1.0; and MgATP, 4. Caesium hydroxide was used to adjust pH to 7.3. The final osmolality was around 295 mosmol kg–1. The liquid junction potential with this high-Cl– pipette solution was relatively small (3 mV), and thus data were not adjusted. Biocytin (2 mg ml–1) was added to the intracellular solution in order to label neurons for immunochemical identification.
Whole-cell recordings were made using either the Multiclamp 700A or Axopatch 200B amplifier and the Digidata interface (1320A or 1322A; Axon Instruments). Under visual guidance, cells lying at 20–80 µm from the slice surface were patched. A small amount of positive pressure was applied through the pipette as it advanced toward the neuron. Upon contacting the cell membrane, negative pressure was applied until a seal (> 1 G) was achieved. Residual electrode capacitance was then negated. Series resistance ranged from 10 to 20 M once whole-cell mode was achieved and was monitored periodically but not compensated. The synaptic activity recorded in voltage-clamp mode was digitized at 20 kHz and filtered at 2 kHz. The voltage output for current-clamp mode was digitized at 20 kHz and filtered at 10 kHz.
Data were transferred to the host computer (Dell Precision 330) and stored for further analysis on an Apple iMac computer. For each neuron, synaptic activity was first examined in voltage-clamp mode for 1 min periods to record inhibitory postsynaptic (IPSCs) and excitatory postsynaptic currents (EPSCs). These were distinguished from –50 to –60 mV, where IPSCs were outward and EPSCs were inward currents, and verified following application of drugs to selectively block one type or the other. Under current clamp, we then collected spike trains for 5 min periods. In order to normalize baseline conditions across neurons and preparations, spike trains were recorded near spike threshold. If the cell was spontaneously active at rest, negative DC current was used to hyperpolarize it closer to spike threshold. If the cell was silent, positive DC current was applied to depolarize the neuron near threshold so that firing was just elicited. Thus firing patterns herein are based on the minimal activity we could achieve (with at least a 1 Hz mean rate). As will be shown in the Results, the amount of DC injection did not vary significantly pre- and postdrug application. Following these recordings, I–V curves were obtained from current-clamp traces to estimate input resistance (Rin) using hyperpolarizing steps (–10 pA, 400 ms). Magnocellular SON neurons were confirmed at recording by the presence of transient outward rectification (Renaud & Bourque, 1991; Armstrong & Stern, 1997) using depolarizing current steps (+10 pA, 400 ms), and post hoc with immunochemistry (see below). We thus measured synaptic activity, spike trains and Rin before and after drug applications.
Data analysis
Spike times were retrieved in AxoGraph software (Axon Instruments) by setting a spike amplitude threshold of 0 mV in order to acquire a serial stream of interspike intervals (ISIs). The ISI data were then analysed in Igor 4.0 (Wavemetrics, Lake Oswego, OR, USA) using a customized subroutine. For irregularly/continuously firing neurons, only stationary spike trains were analysed. Spike trains with long-term tendencies to change firing rate as detected by the elevated tail of the serial correlogram were discarded, or in some cases trimmed to remove areas where rate was changing, so long as a contiguous, stationary area accounted for the majority of the recording time. This was important because long-term trends in firing rate contributed spurious variability. In addition to the mean firing rate, we used two scalable indices of variability: the coefficient of variation (CV) of the ISI and the Fano factor of the firing rate. The CV is the ratio of standard deviation to the mean ISI, whereas the Fano factor is the ratio of the variance of the number of spikes generated within a 1 s time window to the mean number of spikes in the same bin period (Koch, 1999). Each index is useful for determining changes in spike train variability independent of changes in mean firing rate, but their time window is different, since CV is based on the absolute ISIs and the Fano factor is integrated over 1 s. We also plotted the ISI histogram, instantaneous firing rate over time, joint interval histogram and serial correlogram for each period of activity.
Neurons were classified according to their pattern of activity with a membrane potential near spike threshold. For phasic neurons, bursts were identified by an Igor subroutine according to the following criteria: there was at least one burst and one silent period in the 5 min recording; the minimum burst duration and silence duration (when more than one burst was present) were both 3 s; and the minimum mean intraburst firing rate was 
2 spikes s–1 (Poulain et al. 1988; Sabatier et al. 2004). The remaining cells were classified as belonging to a continuously/irregularly firing continuum (Poulain et al. 1988). The joint interval histogram plots each ISI versus the ISI immediately following (ISI+1), and is useful to appreciate the variability of pairs of ISIs, with more irregularly firing neurons displaying a more dispersed distribution.
The serial correlogram (Perkel et al. 1967) is the plot of the serial correlation coefficient of interval lengths. The serial correlation coefficient (
) of order j is the ratio of the corresponding autocovariance (Cj) to the interval variance (
2):
|
| (1) |
|
| (2) |
j:
|
| (3) |
The serial correlogram is useful to appreciate the order dependence of ISIs and to determine long-term stationarity. If ISIs are independently distributed, approaches 0. A large serial correlation coefficient, j
=
Cj/2, (positive or negative) for any order j suggests that a particular spike time is dependent on previous ISIs (e.g. a positive correlation at the first order means that adjacent ISIs covary in length). A prolonged, elevated tail of the serial correlogram suggests that long-term changes in firing patterns, such as a slowly decelerating or accelerating firing rate over the 5 min recording period. For irregularly/continuously firing neurons, we used this plot to trim data so that long-term trends in firing rate were removed and so that relatively stationary distributions were examined. Otherwise, slow changes in firing rate over time produce spuriously large CVs.
Excitatory postsynaptic currents and IPSCs were captured with AxoGraph or MiniAnalysis (Synaptosoft, Decatur, GA, USA). The amplitude and frequency of synaptic events were calculated both before and after drug treatment. For experiments to study amplified miniature IPSCs (mIPSCs) with a high-Cl– pipette solution, 10 µM DNQX and 40 µM 2-amino-5-phosphonopentanoic acid (AP5) were added to block glutamatergic transmission, and 0.5 µM TTX was added to block spikes. When holding at –60 mV in voltage clamp, mIPSCs were large inward currents. The total amount of mIPSCs includes both monoquantal events whose rise and decay could be fully observed and multiple events whose rise or decay could not be fully observed but had discernible peaks.
Since ISI distributions ranged from normal to Poisson, a non-parametric paired test (Wilcoxon's signed rank test) was used to compare data before and after drug application. The non-parametric Mann–Whitney U test was used to determine differences between groups of neurons, for example, OT versus VP neurons.
Immunohistochemistry
For identification of recorded neurons, the slice was fixed in sodium phosphate-buffered (0.15 M; pH 7.4) 4% paraformaldehyde and 0.2% picric acid at 4°C, and stored for several days before histology. Accumulated slices were rinsed thoroughly in sodium phosphate-buffered saline (PBS) containing 0.5% Triton X-100 (TX) made from 0.01 M sodium phosphate and 0.015 M NaCl, and incubated in primary antibody for 1 or 2 days at 4°C. Vasopressin neurons were identified by a polyclonal antibody raised in rabbit against VP-neurophysin (VP-NP, provided by Alan Robinson, retired), diluted 1:20 000. Oxytocin neurons were labelled by a mouse monoclonal antibody specific for OT-neurophysin PS36 or PS38 (OT-NP, provided by Harold Gainer, NIH), diluted 1:500. Primary antibodies were diluted in PBS-TX, 0.04% sodium azide, and could be reused for several incubations. After incubation in primary antibody, the slices were rinsed in three changes of PBS-TX, then placed into a cocktail of secondary antibodies (1:200; Jackson Laboratories, West Grove, PA, USA) containing fluorescein isothiocyanate (FITC)-conjugated goat antirabbit immunoglobulin (GAR-FL), Texas Red isothiocyanate-conjugated goat antimouse immunoglobulin (GAM-TR) and, to label the biocytin-filled neuron, avidin-AMCA (7-amino-4-methylcoumarin-3-acetic acid; 1:400; Vector Laboratories, Burlington, CA, USA), for 1 or 2 days at 4°C. The slices were then rinsed with PBS-TX and mounted with a solution of glycerol and PBS (1:1) for observation under a fluorescence microscope (Nikon Optiphot). Photographs were captured with a digital camera (Sensyscam, Photometrics), and IPLabs (Scanalytics, Fairfax, VA, USA) software. Neurons were identified as either OT or VP neurons only if positive staining was accompanied by a negative reaction for the other peptide. Only a small minority of neurons reacted with both antibodies, in agreement with a previous study showing that a few SON neurons colocalize OT and VP, as well as their respective mRNAs, in normal animals (Mezey & Kiss, 1991).
The work is approved by the Institutional Animal Care and Use Committee of University of Tennessee Health Science Center (protocol no. 283), effective date 1/05/07.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
We were unable to consistently apply the electrophysiological criteria we developed previously, and which were very useful for sharp electrode recordings (Stern & Armstrong, 1995; Teruyama & Armstrong, 2002). These criteria include the presence of sustained outward rectification (SOR) and a rebound depolarization (RD) present in OT but not VP neurons. At this time, we are not certain why these characteristics are not consistently exhibited with whole-cell recording. A recent study using perforated patch recordings (avoiding dialysis) in SON neurons found expression of the SOR in a majority of OT but not VP neurons, using voltage-clamp protocols (Hirasawa et al. 2003). Interestingly, these authors also noted that most OT neurons also expressed a strong inward rectification at potentials more hyperpolarized (< –70 mV) to the range (–70 to –40 mV) where SOR was obvious.
We recorded 202 SON neurons in slices from 104 virgin rats where immunoidentification was attempted. Of these, 89 were identified as VP neurons, 83 as OT neurons (Fig. 1) and 30 were unidentified, yielding a success rate of 85%. Another 11 neurons were studied in a pilot study wherein immunoidentification was not attempted.
|
Of 157 SON neurons (135 irregularly/continuously firing; 22 phasic) examined, 103 (66%) were spontaneously active at rest and exhibited a range of firing rates and patterns resembling those observed in vivo: continuous firing, irregular firing and phasic firing (Poulain & Wakerley, 1982; Poulain et al. 1988). In order to standardize testing, positive DC current was applied to silent neurons to bring them near spike threshold, and for some spontaneously firing neurons we applied negative DC current, if necessary, to bring the membrane potential close to threshold. For continuously firing neurons, the amount of current injection needed to keep neurons at threshold was not significantly different before and after any drug treatment, in either OT or VP neurons (Table 1). Phasic neurons were not considered in this particular analysis owing to their lower numbers. Thus firing patterns herein are based on the minimal activity we could achieve near threshold. Figure 2 shows the three main kinds of firing patterns exhibited by SON neurons: fast continuous, with a Gaussian distribution of ISIs; slow irregular, with a Poisson distribution; and finally the phasic bursting characteristic of VP neurons (Poulain et al. 1988).
|
|
Synaptic activity is largely from miniature PSCs and largely GABAergic
Holding at –50 to –60 mV in voltage-clamp mode with the potassium gluconate pipette solution, both inward EPSCs and outward IPSCs were visible in all neurons. In each cell tested, picrotoxin (100 µM) blocked the overwhelming majority of IPSCs, and DNQX (10 µM) virtually all EPSCs (Fig. 3).
|
We further examined amplified IPSCs with a high-CsCl pipette solution on immunoidentified neurons because IPSCs exerted a more influential effect on firing patterns than that of EPSCs in both OT and VP neurons (see ‘Effects of blockade of IPSCs on continuously firing neurons’ section). Inhibitory PSCs were isolated by applying 40 µM AP5 and 10 µM DNQX, and the recorded and biocytin-filled neurons were identified as OT or VP. Tetrodotoxin did not change the frequency of the IPSCs of these pooled SON neurons (n
= 26; IPSC frequency before TTX, 4.19 ± 3.29 Hz; TTX, 4.35 ± 3.69 Hz; P
= 0.587). Moreover, the IPSC frequency of OT or VP neurons was not changed by TTX (Fig. 4). These data also showed that fewer IPSCs (3 Hz) were detected when using the potassium gluconate pipette solution compared with the high-CsCl pipette solution (6 Hz). Thus, it is possible that small IPSCs were masked by noise and could not be detected with potassium gluconate solution. Another factor contributing to this large difference in IPSC frequency detected by the two pipette solutions might be the ratio of OT and VP neurons recorded in the initial study of unidentified neurons, because the high-Cl– data revealed that OT neurons have far more IPSCs (9.1 ± 6.7 Hz) than do VP neurons (1.1 ± 0.7 Hz; P < 0.0002).
|
To determine the temporal distribution of events, we examined intersynaptic intervals of IPSCs in the presence of DNQX and AP5 in 12 of the 18 OT neurons, chosen because there were sufficient numbers of IPSCs (> 600 events min–1 recording time) to attempt to fit the distribution. In 11 of these 12 neurons, the tail of the intersynaptic interval histogram was well fitted to a single exponential and the serial correlogram was flat (Fig. 5). This is consistent with an essentially random distribution of events, which is also consistent with the relative lack of effect of TTX on synaptic frequency and amplitude. Thus, even though IPSCs may cluster (Fig. 4), these bouts are for the most part randomly distributed. In the one neuron the distribution of which could not be well fitted, there were multiple peaks in the intersynaptic interval distribution and a small oscillation in the serial correlogram, suggesting some order dependence of synaptic events.
|
Effects of blockade of IPSCs on continuously firing VP and OT neurons
We asked whether IPSCs and EPSCs contributed differentially to spike distribution, and whether both OT and VP neurons were affected. We noted with immuno-identification that irregularly/continuously firing neurons could be either OT or VP neurons. However, some neurons changed their firing pattern to phasic bursting during drug treatment; these and other phasic neurons will be dealt with below, and for the following analysis we consider only those neurons that fired irregularly/continuously in both control and drug periods. For OT neurons, the CV and Fano factor were significantly reduced by the blockade of IPSCs, but the mean firing rate and input resistance were unaffected (Fig. 6C). Virtually identical results were obtained for VP neurons, i.e. the CV and Fano factor were significantly reduced after the application of picrotoxin, with no effects on mean firing rate or input resistance (Fig. 6D). After applying picrotoxin, the initial peak of the averaged serial correlogram increased after picrotoxin for both OT (control, 0.180 ± 0.157; picrotoxin, 0.310 ± 0.177; P = 0.0258) and VP neurons (control, 0.287 ± 0.181; picrotoxin, 0.392 ± 0.160; P = 0.0457). This elevation persisted for 5–10 spikes in both cell types, suggesting that adjacent spikes were more positively correlated after the treatment with picrotoxin (Fig. 6E and F).
|
|
|
In a separate group of neurons, DNQX (10 µM) was applied to selectively block fast EPSCs. In contrast to the effects of blocking inhibitory transmission, we found no significant differences in firing variability (Table 2) or firing rate (not shown) for either type of neuron. However, the input resistance was significantly reduced after DNQX for both OT (1100 ± 271 versus 973 ± 302 M; n
= 10; P
= 0.022) and VP neurons (741 ± 252 versus 621 ± 276 M; n
= 10; P
= 0.022). It is difficult to explain this reduction in input resistance, since a conductance block should have increased, not decreased input resistance. Furthermore, the blockade of IPSCs failed to significantly increase input resistance, and there were more IPSCs than EPSCs. Since DNQX was dissolved in 0.1% DMSO, we also tested for vehicle effects on input resistance in 14 neurons, but the results were not significant (P
= 0.08).
|
; n
= 8; P
= 0.21) or VP neurons (624 ± 234 versus 626 ± 261 M; n
= 8; P > 0.99). No changes were found in the CV, Fano factor (Table 2, DNQX) or firing rate (not shown) for either cell type. The initial peak of the averaged serial correlogram likewise was not altered by kynurenic acid in either OT or VP neurons (not shown). These results confirm that the reduction of the input resistance observed in both cell types with DNQX was unrelated to the blockade of AMPA receptors, or EPSCs, and also that this reduction did not influence firing variability. Thus, in these coronal slices, blockade of EPSCs had little effect on the irregular/continuous firing pattern of either cell type, probably because of their relative infrequency compared with IPSCs. Phasic firing
As shown in Figs 2 and 9, phasic bursting neurons exhibited long periods of activity separated by comparable silent periods. We analysed 22 cells that showed phasic firing in the control state to give a baseline of the properties of phasic firing (Table 3) using the criteria listed in the Methods. The data are comparable to a recent in vitro study using extracellular recordings in slices (Sabatier et al. 2004), with the exception that intraburst firing rates reported here are relatively low, perhaps because a lower extracellular potassium concentration (2.5 versus 6.2 mM) was used in the present study.
|
|
Since picrotoxin but neither DNQX nor kynurenic affected the firing parameters of irregular/continuous firing, for statistical purposes we pooled the five phasic firing neurons from combined synaptic blockers (picrotoxin and DNQX) with an additional three neurons exposed only to picrotoxin, all eight of which maintained the phasic pattern in both periods (Fig. 9). We detected no significant differences in any of the firing parameters: intraburst firing rate, intraburst CV, intraburst Fano factor, proportion of time active, burst frequency, burst length, silence frequency and silence length (Fig. 9). Thus, in some neurons at least, phasic patterning was not influenced strongly by inhibitory synaptic activity.
Could the difference between the responsiveness to picrotoxin of continuously firing and phasic VP neurons be related to different amounts of inhibitory synaptic input during the control period? Indeed, the 21 VP neurons that fired continuously throughout the experiment, and were exposed to picrotoxin, had almost three times as many IPSCs (mean 
) as the eight phasic neurons tested with picrotoxin or the cocktail of DNQX and picrotoxin (mean
P
= 0.034), but did not differ in EPSC frequency (P
= 0.86). These results indicated that in the control state, the firing pattern adopted by VP neurons might relate to the frequency of spontaneous inhibitory inputs. We tested this further by comparing the frequency of IPSCs in the control state of VP continuously firing neurons (n
= 21) and all phasic bursting neurons (n
= 22). Although the VP continuously firing neurons tended to have more IPSCs (mean frequency of IPSCs, 2.40 ± 1.98 Hz) than those neurons with phasic firing in control states (mean frequency of IPSCs, 1.55 ± 1.72 Hz), the difference was not significant (P
= 0.09).
We then checked whether the expression of phasic activity was related to drug treatment on a broader scale. A contingency table analysis of the irregular/continuous versus phasic patterns before and after drug treatment revealed no tendency for VP neurons to fire phasically in control versus synaptic blockade (contingency table correlation test: P = 0.217). Thus, synaptic activity, or its relative absence, was not strongly related to the expression of phasic bursting.
Differences between OT and VP neurons
Although the CVs of both OT and VP irregularly/continuously firing neurons were responsive to picrotoxin, we noted differences in the control firing patterns of these two cell types (Fig. 10). The two firing variability parameters, CV (P < 0.0001) and Fano factor (P < 0.0001), were significantly different between OT (n = 33) and VP neurons (n = 41) in the control state, with OT neurons the more variable of the two. In contrast, the minimum mean firing rate of VP neurons near threshold was significantly larger than that of OT neurons (P < 0.0001). The input resistance was larger in OT neurons (Fig. 10D; P = 0.002). Adjacent spikes in VP neurons were more positively correlated than those of OT neurons, as detected by the higher initial peak of averaged serial correlation (P = 0.002; Fig. 10C). This effect appears to persist for 5–10 spikes.
|
As mentioned previously (Fig. 4), OT neurons also exhibited more IPSCs and mIPSCs than did VP neurons when using symmetrical Cl– concentrations to enhance Cl–-mediated events. A subset of these mIPSCs was further analysed for differences in amplitude and decay kinetics. For this analysis, putatively monoquantal mIPSCs were averaged and fitted with one or two exponentials. In most cases, the fit required two exponentials, and we weighted the two for an averaged time constant (
) based on the relative amplitude coefficients. No differences were found in the averaged mIPSC decay (OT, 9.9 ± 2 ms; VP, 8.9 ± 1.6 ms; P
= 0.229) or in the averaged peak mIPSC (OT, –130 ± 5 pA; VP, –138.8 + 48.9 pA; P
= 0.578) between the two cell types.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
GABAergic transmission in SON
While a minority of neurons probably get local input from the perinuclear zone in this preparation (Wuarin, 1997), previous studies have shown little effect of TTX on synaptic activity, either excitatory or inhibitory, indicating a paucity of spontaneously active SON afferents in the coronal slice (e.g. Kabashima et al. 1997; Brussaard et al. 1999). In agreement, we found that the frequency and amplitude of IPSCs were unaffected by TTX in both OT and VP neurons, and that the distribution of IPSCs was largely random. In addition, we found for the first time that: (1) spontaneous IPSCs predominated over EPSCs in vitro; and (2) OT neurons exhibited far greater numbers of IPSCs when compared with VP neurons. Morphologically, GABAergic innervation of OT and VP neurons appears roughly similar (Theodosis et al. 1986). Thus, if GABAergic synaptic input is comparable, our results suggest that the probability of spontaneous GABA release is much smaller onto VP than OT neurons, at least in virgin female rats. Whether this is related to intraterminal mechanisms governing this activity or some differential, tonic activation of presynaptic receptor is presently under investigation.
Previous studies have shown that the precise role of GABAergic activity in controlling OT neurons, at least during lactation, is complex. The GABAA receptor agonist muscimol and the antagonist bicuculline can inhibit the milk ejection reflex after intracerebroventricular application in vivo, and the background firing of both OT and VP neurons was found to be inhibited by both compounds (Voisin et al. 1995). In contrast, intravenous injection of bicuculline methochloride increased OT release during late pregnancy, and increased firing rate when applied to brain slices from late pregnant rats (Brussaard et al. 1997). Unfortunately, in the in vivo experiments, there is no way to determine the site of action of these drugs. Furthermore, both studies used bicuculline salts, which block the small conductance K+ channel (SK)-mediated, Ca2+-dependent after-hyperpolarization in central neurons (Debarbieux et al. 1998); these channels are prominent on MNCs and strongly gate firing rate (Renaud & Bourque, 1991).
Certainly, the degree of spontaneous synaptic activity would be expected to be larger in vivo (Bourque & Renaud, 1991). In slices, connections are obviously removed and dendrites truncated. The localized application of GABAA antagonists (picrotoxin or gabazine) near recorded SON neurons moderately increased the background firing rate yet interrupted the milk ejection reflex, whereas the application of GABA and the GABA agonist isoguvacine decreased the basal electrical activity but facilitated the reflex (Moos, 1995). During lactation, irregular firing precedes burst firing during the milk ejection reflex, and the ability of GABA to facilitate milk ejection bursts is correlated with its promotion of irregular firing (Brown et al. 2000). Since we have found that spontaneous IPSCs impart irregularity to the spike train near spike threshold, we suggest that one mechanism for the switch to irregular firing during background firing periods could be an increase in GABAergic activity. While the specific sources to OT or VP neurons are incompletely described, GABAergic inputs to the SON are thought to derive from multiple forebrain and hypothalamic sources (Roland & Sawchenko, 1993). The perinuclear zone of the SON possesses a substantial complement of GABA- and glutamic acid decarboxylase (GAD)-positive cell bodies that could support a local inhibitory projection system (Theodosis et al. 1986; Jhamandas et al. 1989; Roland & Sawchenko, 1993; Wuarin, 1997). This zone of GABAergic neurons is thought to mediate the inhibition of VP neurons following transient hypertension, possibly via a projection from the diagonal band of Broca (Jhamandas et al. 1989).
It has been proposed that GABAA-receptor signalling can powerfully influence the firing of OT neurons in a reproductive-state-specific manner (Brussaard et al. 1999). These changes include transient increases in mIPSC frequency during early lactation that might reflect synaptic plasticity (El Majdoubi et al. 1997), and pregnancy-related changes in the neurosteroid modulation of postsynaptic GABAA-mediated synaptic transients, each of which could dynamically alter inhibition to influence OT release. If the present results have bearing on this plasticity, we might expect that GABAergic synaptic activity could exert even greater influence over spike train variability during different reproductive states.
In suprachiasmatic nucleus (SCN) neurons, blockade of IPSPs by bicuculline or gabazine also converted irregular firing to a more regular firing pattern, but in the case of bicuculline, also depolarized the neurons and increased firing rate (Kononenko & Dudek, 2004). However, the depolarization and perhaps the strong increase in firing rate were probably due to the non-specific effects of bicuculline salts, known to block Ca2+-dependent after-hyperpolarizations, or even to their ability to inhibit tonic currents. Like SCN neurons, the regularization of spike trains in OT neurons was still found after selective IPSC blockade with gabazine.
Glutamatergic transmission in the SON
Although we found that EPSCs contributed little to spike distribution, this is most likely to result from their low frequency in this preparation and should not be interpreted as evidence for a diminished role for glutamate in producing patterns of activity in either cell type. Indeed, glutamate clearly plays an important role in the generation of OT release during lactation (Parker & Crowley, 1993) and, in certain in vitro models, bursting activity similar to that observed in vivo either can be induced pharmacologically in hypothalamic slices with noradrenaline (Wang & Hatton, 2004, 2005) or is present spontaneously in organotypic cultures (Jourdain et al. 1996; Israel et al. 2003) and, in both cases, is dependent on glutamatergic transmission. Likewise, phasic activity in VP neurons in vivo is suppressed or even blocked in the presence of glutamate receptor antagonists such as ketamine (Nissen et al. 1994) and CNQX (Nissen et al. 1995; Brown et al. 2004), even in the face of osmotic simulation.
Could the difference between IPSC and EPSC frequency have a morphological basis? GABAergic synapses are thought to account for 45–50% of all terminals in the SON (Decavel & Van den Pol, 1990), almost twice that of glutamatergic synapses (El Majdoubi et al. 1997). However, even this difference is less than the striking difference we found between IPSC and EPSC frequency. Perhaps factors controlling spontaneous release differ between the two terminal types, factors that relate to terminal release machinery or to tonic activity from a host of neuroactive substances known to act presynaptically in the SON, such as glutamate, GABA, OT and VP, among others.
Firing patterns in vitro
With the exception of milk ejection bursts, the firing patterns of MNCs evident in vivo (Poulain et al. 1988) are routinely present in vitro. In vivo, the slow irregular firing pattern characterizes both cell types when not activated, whereas continuous firing is more associated with OT neurons and phasic bursting with VP neurons (Poulain et al. 1988). However, continuously firing VP neurons are not uncommon in vivo (Renaud et al. 1987), and individual VP neurons can even exhibit this and the phasic pattern during the same recording period, depending on stimulation level (Poulain & Wakerley, 1982). Although we found that a majority of VP neurons exhibit phasic bursting activity at some point during long recordings, many exhibited only slow irregular/continuous activity. Thus, as we previously found in male rats (Armstrong et al. 1994), continuous activity is not a signature of OT neurons in vitro. Nevertheless, we found distinct differences in this pattern near threshold in the two cell types, in that OT neurons were slower and more irregular than (non-phasic) VP neurons. This irregularity was indicated not only by the higher CV of OT neurons, but also by their lower initial peak of the averaged serial correlation coefficients compared with VP neurons. We suggest that both of these factors are related to the increased amount of inhibitory input impinging on OT neurons. In addition, the fewer IPSCs on VP neurons, as judged by the response to gabazine, do not contribute as much to variability along the irregularly/continuously firing continuum (only the serial correlation coefficient was altered). This leads us to conclude that picrotoxin, which more consistently affects variability in VP neurons, may do so through the additional blockade of a tonic current. However, since input resistance was not significantly increased by picrotoxin in these neurons, these effects are either subtle or perhaps occur on dendrites, electronically remote from the soma. Interestingly, although each cell starts on average from a different CV, firing rate and tonic GABAergic input, picrotoxin has about the same magnitude of effect on CV in the two cell types. Although in vivo VP neuronal spike trains are reportedly more variable than those from OT neurons (Bhumbra & Dyball, 2004), this is explained by the authors by their inclusion of the interburst intervals of phasic activity of VP neurons for the calculation of CV, which weights long intervals at the expense of shorter ones.
Frequency and variability in the background spike train has physiological meaning for OT neurons during milk ejection. Injection of OT into the third ventricle increased background discharge of OT neurons and strongly facilitated the milk ejection reflex (Freund-Mercier & Richard, 1984). However, systemic hypertonic saline (Negoro et al. 1987) produced a strong and sustained increase in background activity, which delayed the occurrence of the following burst. Brown & Moos (1997) found that an intermediate range of background firing, from 1 to 3 Hz, was optimal for promoting the milk ejection burst. Subsequent analyses revealed that the variability of the background firing was a critical factor in predicting bursts (Brown et al. 2000; Moos et al. 2004). Within 1–2 min prior to a milk ejection burst, background firing rate increased and became irregular. Further increasing the firing rate, e.g. with hypertonic saline, reduced variability and inhibited milk ejection bursts, much as Negoro et al. (1987) found. Our data suggest that inhibitory activity, perhaps both tonic and synaptic, can alter variability and could be a factor that governs bursting. Indeed, GABA itself can reduce firing rate but increase the probability of bursting. This may be even more likely with expression of the sustained outward rectifier and rebound depolarization of OT neurons, which are sensitive to small hyperpolarizations from near spike threshold (Stern & Armstrong, 1995).
Phasic bursting neurons, however, were inconsistently affected by GABAergic and glutamatergic synaptic blockade. Consistent with previous studies in vitro (Andrew & Dudek, 1983; Hatton, 1982), this suggests that phasic firing per se in VP neurons does not require synaptic input in vitro. However, we caution overinterpretation of this result; our neurons were often held near spike threshold with current injection, allowing the activation of voltage-dependent conductances, such as the spike depolarizing after-potential (Andrew & Dudek, 1983; Bourque, 1986), that are critical to the expression of phasic activity. Since most VP neurons are slow firing and non-phasic in vivo under resting conditions (Poulain & Wakerley, 1982), activation through membrane depolarization is achieved synaptically or through direct osmotic depolarization. Indeed, recent evidence suggests a profound dependence of phasic activity on glutamatergic synaptic transmission in vivo (Nissen et al. 1995), even in the face of strong osmotic stimulation (Brown et al. 2004). Thus, osmotic depolarization achieved from the intrinsic osmoreceptivity of VP neurons in vivo is insufficient to sustain phasic activity in the absence of additional afferent excitatory drive. The present results that show little influence of synaptic activity on phasic activity also should be viewed in the context of two additional caveats: firstly, that phasic activity is labile over a recording period regardless of the presence of spontaneous synaptic activity and, secondly, that intracellular current injection is a steady force, unlikely to be replicated by neuronal factors in vivo. Our results do suggest, however, that ISIs during phasic bursts in vitro are temporally dispersed, largely from intrinsic rather than synaptic factors and, further, that these properties lead to a more regular firing pattern within bursts. This is consistent with the results of Sabatier et al. (2004), who found that trains within bursts in vivo have a more random distribution of spike times compared with the more regular firing neurons recorded in vitro.
Phasic (synaptic) inhibition and tonic inhibition
We confirmed the presence of a phasic (mediated by GABAergic synaptic transmission) and tonic inhibition in OT neurons. Tonic inhibition has been reported to affect firing properties in other neurons. In cortical interneurons, which express tonic GABAergic currents, the application of 100 µM picrotoxin increased the firing rate produced by injecting any particular current in current-clamp mode, probably reflecting an increased input resistance (Semyanov et al. 2003). A similar result was reported in cerebellar granule cells using a higher dose of gabazine (10 µM), which blocked both IPSCs and GABAergic tonic currents (Brickley et al. 2001). In the meantime, the amplitude and duration of sEPSPs in cerebellar granule cells were enhanced by 10 µM gabazine, which suggested that tonic activation of GABAergic receptors altered the response of granule cells to excitatory inputs (Brickley et al. 2001). In vivo whole-cell patch clamp recording showed that topically applied gabazine (0.5 mM) also produced a significant leftward shift in the spike frequency versus injected current curve in cerebellar granule cells, consistent with the in vitro studies (Chadderton et al. 2004).
According to Park et al. (2006), the blockade of GABAergic tonic currents induced similar effects in MNCs to those in other neurons previously tested, i.e. increased input resistance leading to a shift in the current–frequency curve. These tonic currents were not mediated by glycine receptors. For silent SON neurons, the application of picrotoxin or coapplication of gabazine and bicuculline, but not gabazine alone, depolarized the neuron to increase action potential firing (Park et al. 2006). Tonic inhibition was observed in both OT and VP neurons, whereas in the present study, tonic inhibition was not assessed in VP neurons owing to the highly unstable holding baseline. In addition, we did not find a consistent increase in input resistance to picrotoxin, though the tendency was in that direction, but we did find effects of picrotoxin in VP neurons that were not repeated with gabazine. Together, these studies suggest that tonic inhibition is present in MNCs in addition to phasic inhibition, and that it can modulate neuronal electrical properties in MNCs, as in other neurons. We avoided the use of bicuculline in this study because its salts are known to block Ca2+-dependent K+ currents mediated by SK channels, as are found in the SON and which mediate spike after-hyperpolarizations (Debarbieux et al. 1998).
Conclusion
In summary, we demonstrated that inhibitory activities could affect the firing patterns of both VP and OT neurons in vitro, and that OT neurons were different from VP neurons in several respects, but especially in that they expressed a greater spontaneous GABAergic synaptic input that in turn contributed a more consistent influence on firing variability. To achieve better insight into the differential contribution of phasic (synaptic) inhibition and tonic inhibition to the firing patterns of these two cell types, the restoration of phasic inhibitory conductance and/or tonic inhibitory conductance with dynamic clamp (Mitchell & Silver, 2003) would be useful. Finally, it will be important to determine to what degree the synaptic plasticity associated with reproductive state influences firing properties.
|
Footnotes |
|---|
|
References |
|---|
