To examine the mechanisms underlying milk-ejection bursts of oxytocin (OT) neurons during suckling, both in vivo and in vitro studies were performed on supraoptic OT neurons from lactating rats. The bursts were first recorded extracellularly in anesthetized rats. Burst-related electrical parameters were essentially the same as previous reports except for a trend toward transient decreases in basal firing rates immediately preceding the burst. From putative OT neurons in slices with extracellular recordings, bursts that closely mimicked the in vivo bursts were elicited by phenylephrine, an α1-adrenoceptor agonist, in a low-Ca2+ medium. Moreover, in whole cell patch-clamp recordings, the in vitro bursts were recorded from immunocytochemically identified OT neurons. After a transient decrease in the basal firing rate, the in vitro bursts started with a sudden increase in the firing rate, quickly reaching a peak level, then gradually decaying, and ended with a postburst inhibition. A brief depolarization of the membrane potential and an increase in membrane conductance appeared after the onset of the burst. Spikes during a burst were characterized by a significant increase in the duration and decrease in the amplitude around the peak rate firing. These bursts were significantly different from short-lasting burst firing of vasopressin neurons in membrane potential changes, time to reach peak firing rate, spike amplitude and duration during peak rate firing. Our extensive analysis of these results suggests that the in vitro burst is a useful model for further study of mechanisms underlying milk-ejection bursts of OT neurons in vivo.
During suckling stimulation, oxytocin (OT) neurons in magnocellular nuclei of the hypothalamus display brief, explosive discharges of action potentials recurring every 3–15 min (Belin and Moos 1986; Lincoln and Wakerley 1975). Mechanisms underlying the milk-ejection burst, nevertheless, have remained elusive, mainly due to the limitation of in vivo recording methodology. On the other hand, in vitro studies have elucidated some burst-related features of OT neurons. In organotypic slice cultures of hypothalamus from newborn rats, magnocellular OT neurons show spontaneous bursting behavior driven by excitatory synaptic inputs (Jourdain et al. 1996). However, the usefulness of data from the cultures of newborn rats as a model to study the bursts remains to be confirmed in lactating animals.
Generally, the neurochemical environment plays a decisive role in modulating neuronal firing patterns. Negoro and Wakerley (for review, see Wakerley et al. 1994) have successfully evoked milk-ejection burst-like firing activity in putative OT neurons in hypothalamic slices from lactating rats by using phenylephrine (α1-adrenoceptor agonist). Because α1-adrenoceptors are the main receptors mediating the facilitatory effect of noradrenalin on parturition (Douglas et al. 2001) and lactation (Parker and Crowley 1993), findings from in vitro bursts provide further evidence for a decisive role of local neurochemical environments in controlling the firing pattern of OT neurons. The precise features of these in vitro bursts, however, remain to be characterized.
In establishing an in vitro bursting model, attempts to simulate all neurochemical conditions in advance would be impractical. There is, however, a documented tendency for gradually increased excitability of OT neurons between two consecutive bursts during suckling (Brown and Moos 1997; Brown et al. 2000; Wang et al. 1996). In previous work (Li and Hatton 1996), we found that reducing extracellular Ca2+ concentration could increase the excitability of neurons in the supraoptic nucleus (SON). Lowering extracellular Ca2+ concentration induced intrinsic bursting in nonburster hippocampal CA 1 pyramidal neurons (Su et al. 2001) and reduced the activity-dependent depression of excitatory interneuronal inputs to motor neurons in the spinal cord (Parker 2000). Moreover, extracellular Ca2+ in the synaptic cleft and nearby areas is depleted after intense activation (Rusakov and Fine 2003); depletion of extracellular Ca2+ appears to be a common result of the intense activation of postsynaptic neurons. Thus a low-Ca2+ medium was used as a basal condition for evoking in vitro bursts from OT neurons with phenylephrine in the present work. This is further compared with the in vivo bursts evoked by suckling stimulation in lactating rats. Our results showed that the in vitro bursts evoked under the imposed experimental conditions simulated the in vivo bursts in many important respects. As is true in vivo, these in vitro bursts are distinguishable from short-lasting phasic bursts of vasopressin (VP) neurons by, e.g., their time to peak firing rate, action potential duration and amplitude, and the presence or absence of plateau potentials.
All procedures in animal experiments were in accordance with the guidelines on the use and care of laboratory animals set by National Institutes of Health and approved by the Institutional Animal Care and Use Committee of University of California, Riverside, CA.
In vivo study
Adult primiparous rats (Wistar strain), lactating for 8–13 days were used for experiments. The methods were the same as previously described (Wang et al. 1996). In brief, the mother was anesthetized with urethan (1.1 g/kg body wt, ip) after separation overnight from all but 1 of the litter of 10 pups. Cannulae were placed into the right atrium and into an inguinal mammary teat for measurement of intramammary pressure through an electromagnetic pressure transducer. The rat was fixed in the prone position on a stereotaxic frame. After exposure of the dorsal surface of the cortex, a bipolar electrode for stimulating the neurohypophysis and antidromically activating SON neurons was implanted according to the atlas of Paxinos and Watson (1986). Conventional extracellular recordings of single units from SON OT neurons started 3 h after the operation by using a two-channel amplifier.
In vitro study
Sprague Dawley (Holtzman strain) rats lactating for 8–13 days were used for the experiment. Rats were decapitated with a guillotine. Brains were quickly removed and put in an oxygenated, ice-cold artificial cerebrospinal fluid (ACSF) for 1 min. The ACSF contained (in mM) 126 NaCl, 2.5 KCl, 1.3 MgSO4, 2.4 CaCl2, 1.3 NaH2PO4, 26 NaHCO3, and 10 glucose, pH 7.4 adjusted with 3-(N-morpholino) propanesulfonic acid (MOPS, ∼2 mM) and maintained with 95% O2–5% CO2 gas mixture. Low-Ca2+ solutions were made by decreasing the CaCl2 from 2.4 to 0.7 mM and replacing it with equivalent millimolar MgCl2. Hypothalami were dissected from the brain and cut coronally on a vibratome into 300 μM thick slices. After preincubation at room temperature for 1 h, the slices were used for electrical recordings at 35°C. Extracellular recording was used for detecting bursts while keeping intracellular components undisturbed. Whole cell patch-clamp recordings were used for measurement of postsynaptic currents, membrane potential and action potentials. Patch pipettes were filled with a solution containing (in mM) 145 K-gluconate, 10 KCl, 1 MgCl2, 10 HEPES, 1 EGTA, 2 Mg-ATP, and 0.5 Na2-GTP, pH 7.3, adjusted with KOH. For immunocytochemical identification of OT neurons, 0.5% Lucifer yellow (LY) was added to the pipette solution. Patch electrodes were guided to SON cells under visual observation through an upright microscope (Leica DM LFSA) equipped with water-immersed objectives, IR/DIC, and filters for fluorescent microscopy. Whole cell recordings were obtained from the somata of SON magnocellular neurons during perfusion of ACSF through a gravity feed perifusion system at a rate of 1.2–1.5 ml/min. An Axoclamp 2B amplifier was used for collecting electrical signals that were filtered at 3 kHz and sampled at 5 kHz by Clampex 9 software through a 1200 AD/DA converter (Axon Instruments). Data were stored in a PC computer for off-line analysis.
Immunocytochemical identification of OT neurons
After recording in vitro bursts with electrodes containing LY in the pipette solution, neurons were fixed with 4% paraformaldehyde at 4°C overnight and treated with 0.3% Triton-X100 for 30 min. After incubation with mouse monoclonal OT-neurophysin antibody (PS 38, 1:400 dilution) for 4 h at room temperature, a donkey or goat antimouse antibody (Alexa Fluor 546 or Alexa Fluor 647 labeled, 1:1000) was applied for 1.5 h to label OT neurons. Sections were sealed on glass slides with Vectashield to avoid bleaching then examined with a laser scanning confocal microscope (Leica TCP SP2). Three-dimensional overlap of LY images with Alexa Fluor 546 or 647 staining was taken as an identification of OT neurons.
Data collection and analysis
To identify OT neurons in vivo, the same criteria were taken as previously reported (Wang et al. 1996). To define putative OT neurons in slices, the absence of phasic firing activity was required under either basal conditions or in response to application of any one of the following solutions: 1 nM OT (Yamashita et al. 1987), 1 μM glutamate or hyperosmotic ACSF (Hatton et al. 1978) applied with 1 μM phenylephrine in extracellular recordings. In patch-clamp recordings, only those magnocellular neurons showing nonphasic firing patterns (even after application of phenylephrine, and injection of 10- to 50-pA depolarizing current to bring cells into active firing) and a sustained outward rectification (SOR) (Hirasawa et al. 2003; Stern and Armstrong 1995) evoked by 11 steps of hyperpolarizing current from 40 mV, were considered to be OT neurons. A burst was defined as a short period (<7 s) of firing showing an explosive increase in the firing rate with a peak frequency 10 times more than the basal firing rate, which quickly reached a peak firing rate, slowly decayed in the firing frequency, and ended with a postburst inhibition. Burst amplitude refers to the total number of spikes from burst onset to the initial time of the postburst inhibition as seen in vivo. Burst onset refers to a time when the firing rate increased by more than twofold over the basal firing frequency. The end of a burst was attained when interspike intervals became more than two times the preceding interspike interval during the decay phase of a burst and longer than the average basal firing interspike interval. For comparisons between cells and episodes of altered firing rates, we adopted the index of dispersion (Brown and Moos 1997; Brown et al. 2000). ANOVA, paired Student's t-test, χ2 test, and Pearson's correlation were used for statistical analyses with significance level set at P < 0.05. Burst-related parameters (spikes and afterhyperpolarization, AHP) in patch-clamp recordings were analyzed with Clampfit 9 software (Axon Instruments) after filtering at 1 Hz. All measures were expressed as means ± SE, except when otherwise indicated. A liquid junction potential of -8 mV was left uncorrected.
Chemicals, drugs and antibodies used
Chemicals in the ACSF and in the immunocytochemistry were all from Fisher Scientific except otherwise indicated. Phenylephrine, prazosin, and LY were from Sigma (St. Louis, MO). OT (American Peptide, Sunnyvale, CA) was kept in -20°C freezer in a stock of 1 mM and diluted to the working solution before being used. Other drugs were freshly prepared and dissolved in the ACSF before experiments. Phenylephrine and prazosin solutions were kept from light exposure. A monoclonal antibody against OT-neurophysin (PS38) was provided by Dr. H. Gainer, National Institutes of Health. Alexa Fluor 647-labeled goat anti-mouse IgG, Alexa Fluor 546-labeled donkey anti-mouse IgG were from Molecular Probes (Eugene, OR). Vectashield was purchased from Vector Laboratories (Burlingame, CA). Goat and donkey sera were from ICN Biochemicals (Irvine, CA).
The electrical activity of OT neurons in the SON was first observed in lactating rats using paired-extracellular recordings. Then in vitro bursts were investigated in hypothalamic slices. Finally, whole cell patch-clamp recordings plus the immunocytochemical identification were used to further characterize the intrinsic electrical features of bursting OT neurons.
Milk-ejection bursts evoked by suckling
To establish the criteria for judging in vitro bursts, paired extracellular recordings of the firing activities of OT neurons were carried out in 19 lactating rats (Fig. 1A). Electrophysiological features of the bursts, such as synchronization, recruitment, and dynamic changes in amplitude and interburst intervals, are essentially the same as previously reported (Belin and Moos 1986; Lincoln and Wakerley 1975). A sample of those parameters collected between the second and third bursts is summarized in Table 1. To seek clues for explaining the hyperpolarization-evoked rebound depolarization in OT neurons in vitro (Stern and Armstrong 1995) and GABA-evoked burst in vivo (Moos 1995), we looked closely at basal firing shortly (20–60 s) before the burst. Despite the tendency to increase firing rate as the next burst neared as previously reported (Brown and Moos 1997; Brown et al. 2000; Wang et al. 1996), basal firing rates in the majority (129/197, 65.5%) of the bursts declined transiently within 5–35 s prior to the following burst. Comparing the firing rate 0–20 s prior to the burst with the average firing rate within 1 min preceding the burst revealed that the firing rate in 102 of 197 bursts (51.8%) from 62 neurons transiently decreased, whereas 56 of 197 (28.4%) increased, and in the other 39 bursts (19.8%), the change was <30%. As a whole, the firing rate (2.1 ± 0.10 Hz) at 20 s preceding the burst reduced significantly (P < 0.05, n = 60) compared with that at 0–60 s before the burst (2.4 ± 0.12 Hz). Correlation analysis revealed that the basal firing rate (between the 2nd and the 3rd bursts) was positively correlated with the peak firing rate (average of the 2nd and the 3rd bursts), whereas negatively correlated with the latency to the first burst (Table 1). Typical examples of the preburst basal firing activity are shown in Fig. 1B.
In vitro bursts of putative OT neurons examined in extracellular recordings
In 21 lactating rats, effects of phenylephrine on the firing of 60 putative OT neurons were studied in the low-Ca2+ ACSF with extracellular recordings. Phenylephrine (1 μM) significantly increased the basal firing rate (from 1.2 ± 0.4 to 3.1 ± 1.3 Hz; P < 0.05, n = 20) of putative OT neurons in normal-Ca2+ ACSF, but only two of them (10%) were evoked bursts. In low-Ca2+ ACSF, slow firing or irregularly firing neurons were brought into continuous or phasic firing. After 5 min in low-Ca2+ ACSF, the firing rate of putative OT neurons increased significantly (P < 0.05) from 1.1 ± 0.2 to 2.1 ± 0.15 Hz. The low-Ca2+ ACSF itself could make putative OT neurons present a “preburst firing” pattern (a cluster of 5–10 spikes lasting 0.2–0.5 s), which occasionally turned into a full burst (4/60, 6.7%). To increase the number of bursting OT neurons, low-Ca2+ ACSF was used as a basal condition in observing phenylephrine actions (1 μM, 2 min, 1–3 times in an interval of 15 min per neuron per slice contingent on the occurrence of a burst). In 22 of these 60 neurons (36.7%), including three pairs, one to four bursts were evoked. The ratio of neurons showing the bursts increased with increasing application times. The cumulative probabilities of the neurons showing bursts were 1.7, 6.7, and 36.7% for the first, second, and third application of phenylephrine, respectively (χ2 test, P < 0.05). As shown in Fig. 2, these in vitro bursts displayed the features of in vivo bursts: short-duration, high-frequency, recurrence, exponential decay in firing rate after peak firing, and postburst inhibition, although burst amplitudes were relatively small, postburst inhibition was short, and there was no synchronization between two nearby bursting neurons. The frequency distributions of spikes in the in vitro and in vivo bursts were well matched (Fig. 2A3). In low-Ca2+ ACSF, VP neurons were induced to fire phasically by application of phenylephrine (Fig. 2B). Corresponding to the changes in basal firing activity preceding the in vivo bursts, putative OT neurons also showed a tendency toward a transient reduction in basal firing rate preceding the in vitro bursts after their excitability was increased by low-Ca2+ ACSF and phenylephrine. Among these in vitro bursts, 42.1% decreased, 21.1% increased, whereas the other 36.8% showed a slight reduction (5–26%) in the firing rate 0–20 s preceding the in vitro bursts compared with that during the whole period of 0–60 s before the bursts. Moreover, basal firing rate also showed a tendency toward negative correlation with burst latency and positive correlation with burst amplitude and the peak firing rate similar to those seen in vivo. A detailed comparison of in vitro and in vivo bursts from lactating rats is presented in Table 1.
In vitro bursts of OT neurons in whole cell patch-clamp recordings
To define the in vitro burst and the chemical nature of recorded neurons, whole cell patch-clamp recordings were made from 50 SOR-positive SON neurons from 30 lactating rats. These putative OT neurons (n = 16) had resting membrane potential of -54.4 ± 0.9 mV, input resistance of 698.4 ± 52.4 MΩ, and electrode series resistance of 18.3 ± 2.6 MΩ within 3 min after the whole cell configuration. The membrane resistance reduced to 381.5 ± 75.5 MΩ (P < 0.05), but the electrode series resistance (14.3 ± 2.5 MΩ) did not change significantly after 20 min in the whole cell mode then remained relatively stable, a result that is similar to that reported by another laboratory (Muller et al. 1999).
Besides those observed with extracellular recordings, low-Ca2+ ACSF also caused a decrease in the spike amplitude and interspike intervals and an increase in spike rise and decay time constants (τ), spike duration, firing rate, and instantaneous firing rate. Meanwhile, the AHPs of spikes also changed dramatically. In spite of the increase in the firing rate, AHP duration increased significantly, accompanying an increase in both rise τ and decay τ. Table 2 summarizes the effects of low Ca2+ on OT neurons, examples of which are shown in Fig. 3A. In addition, low-Ca2+ ACSF also tended (0.05 < P < 0.1, n = 11) to increase the average frequency of EPSCs and the instantaneous EPSC frequency, but significantly (P < 0.05) decreased their amplitudes (from 13.6 ± 1.1 to 11.2 ± 0.6 pA). Low-Ca2+ ACSF had no significant effect on the average frequency of IPSCs (P > 0.05, n = 8) but it tended to reduce IPSC amplitude (0.05 < P < 0.1). Noteworthy were periodic clusterings of synaptic events that were intermingled with disbursed synaptic events, which were triggered or reinforced by the low-Ca2+ ACSF, including both EPSCs and IPSCs. Phenylephrine further reinforced the effects of low Ca2+ on the frequency of clustered synaptic events (Fig. 3B), particularly increasing the instantaneous frequency of EPSCs (from 253.2 ± 43.3 Hz in low-Ca2+ ACSF to 334.0 ± 59.4 Hz, n = 7, P < 0.05), which extended into postwashout periods.
Thirty-one in vitro bursts were observed in 14 of 36 (38.9%) OT neurons after application of phenylephrine in low-Ca2+ ACSF. Of these 14 cells, six were examined immunocytochemically and identified as OTergic (Fig. 4C). The in vitro bursts recorded in patch-clamp mode as shown in Fig. 4B are generally similar to those observed with extracellular recordings. Shortly before the bursts, the firing rate in majority (25 of 31, 80.6%) of the bursting neurons tended to decrease. The average firing rate 0–4 s before the burst (2.6 ± 0.42 Hz) was significantly (P < 0.05, n = 31) lower than that 0–60 s before the burst (3.5 ± 0.47 Hz) while membrane potential kept at a relative stable depolarizing level (4.8 ± 0.8 mV). It seems that neither an acute depolarization nor a sudden increase in firing rate shortly before the burst was necessary to trigger the onset of a burst. However, nearing the burst, spikes and their AHPs did show subtle changes. In comparison with 0–60 s before the burst, spikes during 0–4 s preceding the burst showed a dramatic increase in the rising slope (from 34.7 ± 1.7 to 44.5 ± 2.5 mV/ms, P < 0.05, n = 19), a significant decrease in the instantaneous firing rate (17.5 ± 2.1 to 7.0 ± 1.5 Hz, P < 0.01), a decrease in the rising slope of the AHPs (2.3 ± 0.3 to 1.6 ± 0.3 mV/ms, P < 0.05), and an increase in the area of the AHP (234.2 ± 22.8 to 309.5 ± 31.3 mV · ms, P < 0.05).
After the onset of the burst, a brief depolarization appeared, which reached a peak level around the peak rate firing, then repolarized gradually (Fig. 5A). Accompanying the depolarization during the bursts, membrane conductance (Fig. 5C) around the peak rate firing (±150 ms), increased significantly (P < 0.005, n = 8) from 2.79 ± 0.30 nS at 30 s before the burst to 3.76 ± 0.10 nS. The increased conductance quickly recovered to basal levels during the last one-third of the burst (2.73 ± 0.50 nS, n = 3), even though the membrane potential was not yet completely repolarized to the preburst level, i.e., still depolarized by 1.7 ± 0.28 mV. Accompanying the occurrence of the bursts, action potentials also presented several significant changes. As shown in Fig. 5, A and B, spikes that occurred during the burst were characterized by a significant decrease in amplitude, increase in duration at half-amplitude and in the rising and decay τ, and a decrease in the time course of the AHPs. Those changes were most significant at the peak rate firing portion of the bursts (Fig. 5B and Table 3).
To further explore the electrophysiological basis for the in vitro bursts, we compared data from bursting and nonbursting OT neurons (Fig. 6). Before applying the low-Ca2+ ACSF, there was no significant difference between the bursting and the nonbursting neurons in terms of their spike amplitudes, durations, rising and decay τ, firing rates, and instantaneous firing rates or membrane potentials. However, their AHPs did differ significantly. The AHPs of nonbursting neurons had a fast rising τ (1.2 ± 0.3 ms, n = 7 vs. 2.3 ± 0.6 ms in bursting cells, n = 8, P < 0.05) and a slow decay τ (273.1 ± 106.24 vs. 53.2 ± 28.6 ms, P < 0.05), which were accompanied by a tendency toward (0.05 < P < 0.1) bigger AHP amplitude (10.1 ± 1.0 vs. 8.0 ± 1.7 mV). On application of low-Ca2+ ACSF and phenylephrine, similar changes in electrical features were observed in bursting and nonbursting neurons, while still remaining significantly different (P < 0.05–0.001), quantitatively, in many aspects of their activities. In the nonbursting neurons, the increases in spike duration were smaller, enhanced spike frequency as well as instantaneous frequency were less dramatic, but their AHP amplitude was larger, rising τ was shorter while the decay τ was longer, in comparison to corresponding periods in the bursting neurons.
To establish an in vitro burst model of OT neurons, it is essential to exclude the “bursts” of VP neurons. Some bursts from VP neurons can be misconstrued as bursts from OT neurons in extracellular recordings if the neurons were not further identified. There were two types of burst firing from VP neurons that deserve to be clarified. The first type was shortlasting phasic firing based on a silent or irregular firing pattern. Compared with spikes in the bursts from OT neurons, spikes in short phasic bursts had short durations at half-amplitude, short rising τ (1.5 ± 0.1 ms), lower peak firing rates, variable times to peak rate firing, fast rising τ of the AHPs (1.9 ± 0.16 ms), and abrupt termination of firing (0.1 ± 0.05 s) from a relatively high firing rate. Moreover, these spikes were superimposed on clearly sustained plateau potentials. In addition, there were no significant changes in the spike amplitude and duration around and within the short phasic bursts. The second type was “burst firing” during continuous firing phase of VP neurons. In 4 of the 34 (11.8%) phasically firing SOR-negative neurons, five bursts were also evoked by phenylephrine. Their activity generally matched well with the bursts of OT neurons in extracellular recordings. However, there was no significant increase in spike duration at half-amplitude nor significant decrease in spike amplitude around peak firing rate within the bursts of phasic neurons. The details of these bursts from putative VP neurons are also presented in Table 3.
Using low-Ca2+ medium to simulate the depletion of extracellular Ca2+ after intense activation (Rusakov and Fine 2003) and to elevate the excitability of SON neurons (Li and Hatton 1996) simulating the general excitatory responses of OT neurons toward bursts during suckling (Brown and Moos 1997; Brown et al. 2000; Wang et al. 1996), and application of phenylephrine, mimicking adrenergic actions during nursing and parturition (review see Wakerley et al. 1994), we successfully elicited milk ejection burst-like electrical activity in immunocytochemically identified OT neurons from lactating rats. Although the requisites of synchronization, longer postburst inhibition, and higher burst amplitudes are not yet satisfied, present work still provides a suitable model to further clarify the inherent features of and gating mechanisms underlying the bursts.
General considerations of in vitro bursts
Bursting activity is a common feature of many neurons recorded in vitro (for review, see Wang and Rinzel 1997). Examples are the epileptiform discharges of hippocampal neurons (Su et al. 2001), rhythmic bursts of thalamocortical neurons (Beierlein et al. 2002), pulse generators of GnRH neurons in the hypothalamus (Suter et al. 2000), oscillatory bursts of subthalamic neurons (Beurrier et al. 1999), bursting neurons in the deeper layers of the superior colliculus (Saito and Isa 2003), respiratory pattern generators in the brain stem (Dunin-Barkowski et al. 2003), and burst networks in the spinal cord (Darbon et al. 2003). Burst firing patterns of these neurons can be elicited simply by raising neuronal excitabilities; the occurrence of those bursts is “predictable.” Their firing patterns and time courses are also different from the bursts of OT neurons, resembling more closely the oscillatory burst firing of VP neurons: i.e., based on a plateau potential, short interburst intervals, and spontaneous occurrence (for review, see Armstrong 1995). The in vitro bursts of OT neurons occurred suddenly without requiring immediate, detectable changes in membrane potential shortly before the bursts; their firing rates quickly reached the peak level accompanying a dramatic increase in the spike duration and decrease in the spike amplitude riding on an arc of brief depolarization, which was not a plateau potential; the interburst intervals were relatively long and firing activity transiently reduced preceding the bursts. Moreover, there was almost no spontaneous occurrence of the bursts in OT neurons in vitro under control conditions. Examples showing the bursting firing features of OT neurons are also the differential responses of VP and OT neurons to excitatory agents in the present work. It is noteworthy that during the bursts of VP neurons, the peak firing rate (46.5 Hz) was even higher than that of bursting OT neurons (40.4 Hz), but there was no significant increase in spike duration around the peak rate portion within the bursts, indicating that the increased spike duration of the bursting OT neurons is not a simple spike broadening (Andrew and Dudek 1985; Oliet and Bourque 1992) that commonly exists in burst firing patterns. Therefore it is clear that the in vitro bursts are a specific response of OT neurons to the challenge of an appropriate neurochemical environment.
In other experiments, we also elicited the bursts from putative OT neurons in slices from virgin female and male rats by applying phenylephrine (unpublished data). This again raises the question of whether general mechanisms of excitability are involved rather than those that are specific to the state of the animal at the cellular level. In fact, under some physiological conditions pulsatile secretion of OT is also found, such as sexual intercourse in human (Kroeger 1996; Murphy et al. 1990) and in nonlactating rats (Hughes et al. 1987), which is also related to the activation of OT neurons (Flanagan-Cato and McEwen 1995; Yanagimoto et al. 1997) and the mobilization of adrenergic systems (Kruger et al. 2003). Although no evidence is available to support the possibility of synchronized burst firing of OT neurons in virgin female and male rats during sexual intercourse, these in vitro bursts are closely associated with normal in vivo physiological processes and specific cellular environments that are decided by those physiological processes.
Noteworthy also are several differences between the bursts of OT neurons recorded in vivo and in vitro. Such differences are probably due to the isolation that exists in slices and the nature of perfusion systems. In the slices, OT neurons receive fewer active synaptic inputs; connections between OT neurons and modulatory structures were largely blocked. Also the local neurochemical environment around a superficial OT neuron in a thin slice could not be well maintained as that in vivo. Thus the bursts in vitro presented relatively small amplitude, short postburst inhibition, and no synchronization among bursts of different widely separated OT neurons. The shortening of postburst silent periods may not be related to the surgical procedure because the bursts of OT neurons in organotypic cultures also had short postburst silent periods (Jourdain et al. 1998). Taken together, this bursting model primarily meets the needs of investigation of the mechanisms underlying the milkejection burst.
Neural circuits and the in vitro bursts
In organotypic cultures of neonate rats, the milk-ejection burst-like activities occurred spontaneously (Jourdain et al. 1996), something that is driven by clustering excitatory inputs from glutamatergic neurons (Israel et al. 2003). From their results, it seems any excitatory input could trigger the bursts if it is strong enough to hold membrane potential in a depolarized state. Although similar clustering EPSCs were also observed in the present experiment, we cannot confirm a hypothesis that the bursts were directly triggered by external excitatory events in lactating animals. In adult hypothalamic slices, OT neurons do not readily display burst firing on application of excitatory neurotransmitters. Membrane potential oscillations were visibly uncorrelated with the bursts: in the present study, bursts occurred in a sudden and unheralded fashion. In addition, we did not find clustering EPSPs during the bursts of OT neurons. Our results are consistent with the report that there was little correlation between the electrical activity of perinuclear zone neurons and OT neurons during suckling (Dyball and Leng 1986). However, we cannot rule out the possibility that due to the high-input resistance, the effect of clustering excitatory synaptic inputs could be masked while subtly influencing the intrinsic bursting features of OT neurons. On the other hand, it is also possible that a cluster of inhibitory synaptic inputs causes a transient decrease in the excitability of OT neurons shortly before the burst, as earlier hypothesized (Hatton 1997), judging from the transient decrease in the firing rate preceding the in vivo and in vitro bursts relative to the general increase toward a burst (Brown and Moos 1997; Brown et al. 2000; Wang et al. 1996), occurrence of clusters of inhibitory synaptic inputs under present experimental conditions and GABA-triggered bursts (Moos 1995), hyperpolarizing current facilitating burst firing in the present conductance experiment and other report (Stern and Armstrong 1995).
Membrane electrical features and the in vitro bursts
In studying mechanisms underlying the bursts, an essential question is the transition between low-frequency basal firing and the high-frequency burst firing. Changes in the electrical features shortly before the bursts and the different responses between bursting and nonbursting OT neurons to phenylephrine stimulation do provide some answers for this question: e.g., a decrease in the rising slope of the AHPs and an increase in the rising slope of spikes (preceding a burst) or a short decay time course of the AHPs (bursting neurons). These changes would apparently enhance the potential for OT neurons to fire bursts. Because the increase in the rising slope of spikes mainly reflects the increase in Na+ currents (review see Hatton and Li 1998), low-Ca2+ medium increases the persistent Na+ current and reduced the screen effect of Ca2+ on voltage-dependent Na+ currents (Li and Hatton 1996), the activation of Na+ currents can be one of the driving forces for the bursts. The time course for the rising phase of the AHPs is mainly decided by Ca2+-activated K+ currents (for review, see Hatton and Li 1998) and phenylephrine decreases in K+ currents (for review, see Piascik and Perez 2001); it is possible that a dramatic decrease in K+ currents together with an increase in the Na+ currents initiated the bursts on the basis of a general excitation. Moreover, the shorter time course of the AHP decay phase may reflect the existence of a larger hyperpolarization-activated inward current in the bursting OT neurons (Ghamari-Langroudi and Bourque 2000) that provides an extra excitatory drive for the onset of the bursts. It is not clear about the relationship between the firing rate reduction and the changes in the spike features shortly before the occurrence of the bursts, but it is predictable that an intense modulation of synaptic activity and dynamic changes in these inherent membrane electrical features decide the bursts of OT neurons.
Application of the in vitro bursting model
From the present study, an in vitro burst-firing episode should be considered as a mirror of the milk-ejection burst, providing that the following criteria are met. The first criterion is the general similarity of an in vitro burst firing to the milk-ejection burst as revealed by extracellular recordings. The second criterion is an absence of the firing features typical of VP neurons: plateau potentials and phasic firing. The third is a combination of a decrease in the spike amplitude, a transient depolarization of membrane potential, and an increase in spike duration during the peak rate firing in the burst.
We thank Dr. Hideo Negoro for advice with this work, Dr. Harold Gainer for providing PS 38 antibody, and T. A. Ponzio, C. Turenius, and Q. Z. Yang for helpful comments on an earlier draft of the manuscript. G R A N T S This research was supported by National Institute of Neurological Disorders and Stroke Grant NS-09140 to G. I. Hatton and Heilongjiang Education Committee Grant 9541053 to Y.-F. Wang.
- Copyright © 2004 by the American Physiological Society