Noradrenergic projections to the hypothalamus play a critical role in the afferent control of oxytocin and vasopressin release. Recent evidence for intrahypothalamic glutamatergic circuits prompted us to test the hypothesis that the excitatory effect of noradrenergic inputs on oxytocin and vasopressin release is mediated in part by local glutamatergic interneurons. The voltage response to norepinephrine (30–300 μm) was tested with whole-cell recordings in putative magnocellular neurons of the paraventricular nucleus (PVN) in hypothalamic slices (400 μm). Norepinephrine elicited an α1 receptor-mediated direct depolarization in 23% of the magnocellular neurons tested; however, the most prominent response, seen in 42% of the magnocellular neurons, was an α1receptor-mediated increase in the frequency of EPSPs. The norepinephrine-induced increase in EPSPs was blocked by tetrodotoxin and by ionotropic glutamate receptor antagonists, suggesting that norepinephrine excited presynaptic glutamate neurons to cause an increase in spike-mediated transmitter release. The increase in EPSPs also was observed in a surgically isolated PVN preparation (64% of cells) and with microdrop applications of norepinephrine (1 mm, 33% of cells) and glutamate (0.5–1 mm, 28%) in the PVN, indicating that the norepinephrine-sensitive presynaptic glutamate neurons are located within the PVN. Biocytin injection and subsequent immunohistochemical labeling revealed that both oxytocin and vasopressin neurons responded to norepinephrine. Our data indicate that magnocellular neurons of the PVN receive excitatory inputs from intranuclear glutamatergic neurons that express α1-adrenoreceptors. These glutamatergic interneurons may serve as an excitatory relay in the afferent noradrenergic control of oxytocin and vasopressin release under certain physiological conditions.
Magnocellular neurons of the hypothalamic paraventricular nucleus (PVN) and supraoptic nucleus (SON) are stimulated under certain physiological conditions to generate bursts of action potentials. Although intrinsic membrane ionic conductances capable of sustaining bursting activity have been characterized in magnocellular neurons (Bourque and Renaud, 1991;Legendre and Poulain, 1992), the synaptic mechanisms responsible for triggering bursts and for coordinating the bursting activity in response to specific sensory stimuli are less well understood.
Several lines of evidence suggest that noradrenergic inputs play a critical stimulatory role in the release of oxytocin and vasopressin under conditions of increased hormone demand. Oxytocinergic and vasopressinergic magnocellular neurons of the PVN and SON are contacted directly by noradrenergic synapses (Ginsberg et al., 1994; Michaloudi et al., 1997), and stimulation of the brainstem A1/A2 noradrenergic cell groups activates putative magnocellular neurons and causes oxytocin and vasopressin release (Day et al., 1984; Tanaka et al., 1985; Day and Sibbald, 1988; Kim et al., 1989). This effect is blocked by 6-hydroxydopamine lesion of the hypothalamic noradrenergic projections (Day et al., 1984). Intracerebroventricular or local hypothalamic injection of norepinephrine or α1-receptor agonists results in the release of oxytocin (Bridges et al., 1976;Tribollet et al., 1978) and vasopressin (Benetos et al., 1986;Willoughby et al., 1987). Similarly, α1-receptor agonists applied to hypothalamic explants causes depolarization and spike generation in SON magnocellular neurons and leads to an increase in oxytocin and vasopressin release (Armstrong et al., 1986; Randle et al., 1986a,b).
These noradrenergic afferents may be part of the ascending sensory pathways responsible for selectively activating the oxytocin neurons during parturition and reflex milk ejection and the vasopressin neurons during hemorrhage. During parturition, the norepinephrine concentration in the SON rises prior to and in parallel with increases in blood levels of oxytocin (Herbison et al., 1997). Blockade of noradrenergic inputs to the hypothalamus with 6-hydroxydopamine lesion or with α1-adrenoreceptor antagonists inhibits the reflex release of oxytocin associated with milk ejection (Tribollet et al., 1978;Clarke et al., 1979; Crowley et al., 1987). Similarly, reflex vasopressin release in response to unloading of arterial baroreceptors is accompanied by a rise in the concentration of norepinephrine in the PVN (Van Huysse and Bealer, 1991), which is caused by the activation of A1 noradrenergic projections (Day and Renaud, 1984).
Several studies suggest a possible role of glutamate in the triggering mechanism for oxytocin release. Glutamate levels in the SON have been found to rise abruptly just before parturition and to decline before delivery is terminated (Herbison et al., 1997), and intracerebroventricular injection of a glutamate receptor antagonist blocks the suckling-induced release of oxytocin (Parker and Crowley, 1993a). We recently found evidence for excitatory synaptic inputs to magnocellular neurons of the PVN and SON from intrahypothalamic glutamate neurons (Boudaba et al., 1997). The current study was conducted to determine whether the excitatory effect on oxytocin and vasopressin release of noradrenergic inputs is mediated by local glutamatergic circuits.
A preliminary account of this data has been published previously in abstract form (Daftary et al., 1996).
MATERIALS AND METHODS
Slice preparation. Male Sprague Dawley rats (50–150 gm; Charles River, Wilmington, MA) were deeply anesthetized with pentobarbitol sodium (50 mg/kg body weight) and decapitated. The brain was quickly and gently removed from the cranial cavity and immersed in cooled (1–2°C), oxygenated (100% O2) artificial CSF (aCSF). The composition of the aCSF was (in mm): 140 NaCl, 3 KCl, 1.3 MgSO4, 1.4 NaH2PO4, 2.4 CaCl2, 11 glucose, and 5 HEPES; pH was adjusted to 7.2–7.4 with NaOH. The hypothalamus was blocked with a razor, and 400 μm hypothalamic slices were sectioned in the coronal plane using a vibrating microtome (World Precision Instruments, Sarasota, FL). Two slices containing the PVN were identified, and a single slice was transferred immediately to a ramp-style, interface recording chamber where it was perfused with humidified, oxygenated aCSF maintained at 32–34°C and allowed to equilibrate for at least 1 hr before recordings were started. The second slice was stored submerged in a holding chamber in oxygenated aCSF at room temperature until it was used. In some experiments, the PVN was surgically isolated from the rest of the slice under a dissecting microscope using a scalpel.
Electrophysiological methods. Sharp microelectrodes were made from microfilament glass capillaries [1.0 mm outer diameter (o.d.), 0.6 mm inner diameter (i.d); World Precision Instruments], and patch pipettes were pulled from borosilicate glass (1.65 mm o.d., 1.2 mm i.d.; KG-33; Garner Glass, Claremont, CA) using a Flaming-Brown P-97 micropipette puller (Sutter Instruments, Novato, CA). Sharp electrodes were filled with 2 m potassium acetate. Patch pipettes were filled with a solution containing (in mm): 120 potassium gluconate, 10 HEPES, 1 NaCl, 1 CaCl2, 1 MgCl2, 2 Mg-ATP, 0.3 Na-GTP, and 10 EGTA; pH was adjusted to 7.2–7.4 with KOH. The osmolarity of the patch solution was made hyperosmotic (290–310 MΩ/l) with 20 mm d-sorbitol to reduce series resistance.
The slice was transilluminated in the recording chamber, and the recording electrode was positioned in the magnocellular PVN under visual guidance using a dissecting microscope. The electrode was lowered through the slice by 2- to 4-μm-steps with a piezoelectric microdrive (Nanostepper, Adams & List, Westbury, NY). Recordings were performed in current-clamp mode using an Axoclamp 2A amplifier (Axon Instruments, Foster City, CA) and were monitored continuously on a digital storage oscilloscope (Hitachi, Tokyo, Japan). Data were converted to digital video format (Neurocorder, Neurodata Instruments, New York, NY) and stored on videotape for off-line analysis. Episodes of spontaneous EPSPs (30–60 sec) 1 min before and 8–9 min into norepinephrine application were amplified 10×, filtered at 1 kHz with a filter/amplifier (Cygnus Technology, Delaware Water Gap, PA), and digitized at 2–4 kHz with a TL-1 interface and the pClamp 6.1 suite of software (Axon Instruments). EPSP responses to 100 μmnorepinephrine were analyzed for changes in frequency and amplitude with the Datapac program (Run Technologies, Laguna Hills, CA). Cumulative probability distributions of EPSP amplitude and instantaneous frequency were generated using an in-house program and compared with the Kolmogorov–Smirnov test. Population means were compared with the Student’s paired t test and the Wilcoxon signed rank test for frequency and amplitude values, respectively. Probability values <0.05 were considered significant. Means are expressed as ± SE.
Drug application. Norepinephrine, adrenoreceptor antagonists, tetrodotoxin (TTX), glutamate and glutamate receptor antagonists were dissolved in aCSF and either bath-applied, or, where indicated, norepinephrine and glutamate microdrops were applied by pressure on the surface of the slice using a picospritzer (General Valve, Fairfield, NJ). Norepinephrine (Sigma, St. Louis, MO) was bath-applied for 5–15 min at concentrations ranging from 30 μm to 300 μm. Norepinephrine (1 mm) and glutamate (0.5–1 mm) microdrops were applied under visual control at one or more sites in the PVN using a patch pipette with a broken tip positioned with a micromanipulator (Newport Corporation, Irvine, CA). Janus green (0.1%) was added to the norepinephrine and glutamate microdrop solutions to monitor visually the spread of the drops. Adrenoreceptor antagonists included the α1-adrenoreceptor antagonist prazosin hydrochloride (10 μm) and the β-adrenoreceptor antagonist propranolol hydrochloride (10 μm) (Research Biochemicals International, Natick, MA); glutamate receptor antagonists included the NMDA receptor antagonist d,l-2-amino-5-phosphonovalerate (AP5) (100 μm) and the non-NMDA receptor antagonist 5,6-dinitroquinoxaline-2, 3-dione (DNQX; 50 μm) (Tocris Cookson, Ballwin, MO). Adrenoreceptor and glutamate receptor antagonists were bath-applied for 15 min before the reapplication of norepinephrine. Tetrodotoxin (1.5–3 μm) (Sigma) was used to block voltage-gated sodium channels and spike-mediated transmitter release. Stock solutions (10 mm) of prazosin hydrochloride and propranolol hydrochloride were prepared in aCSF and stored in the dark at −20°C until use.
Biocytin histology and peptide immunohistochemistry.Biocytin was added to microelectrodes (1%) and to patch pipettes (0.3–0.5%) as an intracellular marker. The biocytin leaked into the recorded cells during patch recordings, or in the case of microelectrode recordings was iontophoresed intracellularly at the end of experiments by passing negative current pulses (−250 pA, 250 msec, 2 Hz) for 5–10 min.
After experiments, slices were removed from the recording chamber and fixed overnight in 4% paraformaldehyde in 0.1 m PBS at 4°C. They were then sectioned on a freezing microtome at 20–25 μm, and the biocytin-injected cells were labeled by incubating the sections for 4 hr in streptavidin-conjugated 7-amino-4-methyl-coumarin-3-acetic acid (AMCA; Molecular Probes, Eugene, OR). The AMCA was diluted 1:300 in 0.1 m PBS containing 0.5% Triton X-100. Sections were scanned under a fluorescence microscope using a UV/420K filter combination to detect the presence of biocytin-filled, AMCA-labeled neurons.
Sections containing the AMCA-labeled cells were placed in 2% normal sheep serum in 0.1 m PBS for 15 min. To determine whether the stained cells were oxytocin or vasopressin magnocellular neurons, we used a mixture of a rabbit polyclonal antibody to oxytocin (VA-10) and a mouse monoclonal antibody to vasopressin-associated neurophysin (PS41) on the same section. Both antibodies were kindly provided by Dr. H. Gainer (National Institutes of Health, Bethesda, MD) (Ben-Barak et al., 1985; Altstein et al., 1988). The polyclonal oxytocin antibody (1:2000) and the monoclonal vasopressin-associated neurophysin antibody (1:200) were applied together for 36 hr at 4°C in 0.1 mPBS + 1% normal sheep serum and 0.2% sodium azide. After treatment with the primary antibodies, sections were rinsed with 0.1m PBS, incubated for 1 hr in a mixture of anti-rabbit IgG conjugated to fluorescein isothiocyanate (FITC, 1:100; Vector Labs, Burlingame, CA) and anti-mouse IgG conjugated to rhodamine (1:100; Jackson ImmunoResearch Labs, West Grove, PA), and rinsed again in 0.1m PBS. They then were mounted, coverslipped, and examined under 450–490 nm excitation/515 nm barrier filters to detect the FITC-labeled oxytocin neurons and 515–560 nm excitation/580 nm barrier filters to see the rhodamine-labeled vasopressin neurons. Recorded cells were positively identified as either oxytocinergic or vasopressinergic only if they labeled positive for one of the two antibodies and negative for the other. No cells were found to be positively labeled for both antibodies, confirming the specificity of the antibodies. We have tested the polyclonal oxytocin antibody (VA-10) for specificity using preabsorption controls (Boudaba et al., 1996).
Putative magnocellular neurons of the PVN were distinguished from putative parvocellular neurons during recordings based on specific electrophysiological properties (Hoffman et al., 1991; Tasker and Dudek, 1991). In particular, magnocellular neurons have a prominent transient outward rectification, a relatively short membrane time constant, and linear current–voltage relations.
A total of 136 putative magnocellular neurons were recorded in this study: 122 cells were tested for their response to norepinephrine and 14 cells for their response to glutamate microdrops. Thirteen of the cells were recorded with sharp electrodes to control for washout of the norepinephrine signal; these cells showed the same responses to norepinephrine as the rest of the cells that were recorded with patch electrodes. The neurons recorded with sharp electrodes had a mean membrane potential of −54 ± 2 mV (SE), input resistance of 215 ± 24 MΩ, and action potential amplitude of 65 ± 2 mV (threshold–peak). The cells recorded with patch electrodes had a mean membrane potential of −66 ± 1 mV (corrected for a −11 mV junction potential), input resistance of 871 ± 65 MΩ, and action potential amplitude of 66 ± 1 mV (n = 58). The patch solution was weakly hyperosmotic, which resulted in a hyperpolarized resting potential. Application of norepinephrine caused a direct depolarization or an increase in EPSPs, or both, in the majority (60%) of magnocellular neurons recorded.
Direct effect of norepinephrine
Norepinephrine was considered to have a presumptive direct, postsynaptic effect if it resulted in a sustained change in membrane potential of at least 3 mV that was reversed with washout. Bath application of norepinephrine at concentrations of 30–300 μm elicited a reversible depolarization (7.27 ± 0.6 mV) in 18 of 90 (20%) putative magnocellular neurons tested (Fig.1 A). The depolarization occurred within ∼3–7 min of norepinephrine introduction into the recording chamber and was accompanied by a decrease in input resistance (15 ± 2%) in 4 of 16 cells tested. The response was not blocked by TTX (1.5–3 μm, n = 3), suggesting that it was not mediated by an increase in spike-mediated transmitter release. It was blocked or reduced by the α1-adrenoreceptor antagonist prazosin (10 μm, n = 4) but was not affected by the β-adrenoreceptor antagonist propranolol (10 μm,n = 3) (data not shown). The direct response to norepinephrine was qualitatively similar to that described in magnocellular neurons of the supraoptic nucleus (Randle et al., 1986a) and was not investigated further in this study.
Norepinephrine activation of presynaptic glutamate neurons
Bath application of norepinephrine (30–300 μm) caused a large increase in EPSPs in 38 of 90 (42%) putative magnocellular neurons tested in the whole slice (Fig.1 B). An increase in EPSPs was indicated by a qualitatively detectable rise in the frequency (i.e., of at least 10–20%) of positive-going synaptic potentials recorded at resting membrane potential. The increase in EPSPs occurred ∼7–8 min after introduction of the norepinephrine into the recording chamber, was maintained throughout the application, and reversed within 10–33 min of washout of the norepinephrine. Neurons responding to 100 μm norepinephrine were selected for EPSP frequency and amplitude analysis. Norepinephrine caused a significant increase in both the frequency (p < 0.001;n = 22; Student’s paired t test) and the amplitude (p < 0.05; n = 10; Wilcoxon signed rank test) of spontaneous EPSPs collected in 30–60 sec episodes. This was seen in individual cells as a significant shift in the cumulative EPSP frequency and amplitude distributions (p < 0.01; n = 5; Kolmogorov–Smirnov test) (Fig. 1 C). The mean percentage increases in EPSP frequency (n = 22) and amplitude (n = 10) were 145 and 52%, respectively (Fig.1 D). The marked rise in the frequency of EPSPs suggests that norepinephrine increases the probability of transmitter release from presynaptic excitatory neurons. The moderate increase in EPSP amplitude suggests that it may also modulate the postsynaptic responsiveness of the magnocellular neurons, although this effect is less robust. No apparent desensitization of the synaptic response was observed during the norepinephrine application or with a second application of norepinephrine (n = 3). Of the 38 cells that showed an increase in EPSPs, eight cells also depolarized in response to norepinephrine.
The norepinephrine-induced increase in EPSPs was blocked by the α1-adrenoreceptor antagonist prazosin hydrochloride (10 μm) in six of six cells tested (Fig.2). The β-receptor antagonist propranolol hydrochloride (10 μm) failed to block the EPSPs elicited by norepinephrine (n = 5). Thus, norepinephrine appears to activate presynaptic excitatory neurons by acting at α1-receptors.
To test whether norepinephrine caused an increase in EPSPs by acting at presynaptic neurons, we applied TTX (1.5–3 μm) to block spike-mediated transmitter release. Bath application of TTX blocked completely the norepinephrine-evoked increase in EPSPs in eight of eight cells tested (Fig. 3). This indicated that the increased EPSPs were caused by an increase in spike-evoked transmitter release and suggested that norepinephrine was acting at receptors at the somatic/dendritic region of presynaptic excitatory neurons. In the cells that responded with an increase in EPSPs, norepinephrine had no apparent effect on the depolarizing voltage responses to positive current pulses (n = 4), suggesting that norepinephrine was not acting postsynaptically to amplify spontaneous EPSPs by enhancing or attenuating voltage-gated currents.
That norepinephrine application led to an increase in the frequency of fast EPSPs suggested that it was acting to stimulate presynaptic glutamate neurons, causing an increase in glutamate release and activation of postsynaptic ionotropic glutamate receptors. We tested this hypothesis by bath applying the NMDA and non-NMDA receptor antagonists AP5 and DNQX, respectively, to block ionotropic glutamate receptors. In cells that had shown a norepinephrine-evoked increase in EPSPs, AP5 and DNQX blocked completely the synaptic response to norepinephrine in six of six cells tested (Fig.4). Thus, the presynaptic neurons that responded to norepinephrine with a spike-mediated increase in transmitter release onto PVN magnocellular neurons were glutamatergic, and the synaptic responses were caused by glutamate activation of ionotropic glutamate receptors.
Intranuclear localization of the presynaptic glutamate neurons
The TTX experiments suggested that the presynaptic glutamate neurons were present and intact in our slices, because the norepinephrine was probably acting at the somatic/dendritic regions of the presynaptic cells to cause action potential generation. This, along with the relatively high percentage of magnocellular neurons (42%) in our slices that responded to norepinephrine with an increase in EPSPs, indicated that the presynaptic glutamate neurons were in close proximity to the magnocellular neurons, possibly inside the PVN.
We conducted three experiments to test this hypothesis. The first experiment tested the effects of norepinephrine on the incidence of EPSPs in PVN magnocellular neurons in a slice preparation in which the PVN was surgically isolated by cutting away and removing the rest of the slice. In this preparation, bath application of norepinephrine caused an increase in the frequency of EPSPs in 7 of 11 (64%) putative magnocellular neurons tested (Fig. 5). Next, glutamate (0.5–1 mm) was applied as microdrops directly into the PVN to stimulate neurons focally within the PVN without activating axons of passage (Christian and Dudek, 1988). Glutamate microstimulation led to an increase in EPSPs in 4 of 14 (28%) putative magnocellular neurons tested (data not shown), suggesting the presence of intranuclear excitatory circuits. Finally, to determine whether these intranuclear excitatory circuits were the same circuits as those activated by bath application of norepinephrine, we applied microdrops of norepinephrine (1 mm) directly into the PVN. Norepinephrine microdrops also elicited an increase in EPSPs in 7 of 21 (33%) putative magnocellular neurons tested (Fig.6). Four of the seven (57%) neurons also exhibited a depolarization in response to norepinephrine. The results of these experiments all point to the presence of glutamate neurons within or very closely apposed to the PVN that are excited by norepinephrine and that send intranuclear projections to magnocellular neurons. Stimulation of these local glutamate interneurons via α1-receptor activation results in an increase in excitatory synaptic input to magnocellular neurons.
Immunohistochemical identification of magnocellular neurons
A total of 20 putative magnocellular neurons were recovered after biocytin labeling and oxytocin/vasopressin immunohistochemical processing with antibodies to oxytocin and vasopressin neurophysin. Only those cells that were immunopositive for one and negative for the other of the two antibodies were counted. Of the eight cells that responded to norepinephrine with a depolarization, five were immunopositive for oxytocin and immunonegative for vasopressin, and three were immunopositive for vasopressin and immunonegative for oxytocin. Of the 12 cells that responded to norepinephrine with an increase in EPSPs, three were immunopositive for vasopressin and immunonegative for oxytocin (Fig. 7 A), and nine were immunopositive for oxytocin and immunonegative for vasopressin (Fig.7 B). These results suggest that both oxytocin and vasopressin magnocellular neurons express functional α1-adrenoreceptors and receive intranuclear excitatory synaptic inputs from glutamate neurons located within the PVN.
Despite the apparent prominent role of norepinephrine in the control of oxytocin and vasopressin neuronal activity and hormone release, little is known about the mechanisms or the locus of the excitatory actions of norepinephrine in the PVN. This is the first study to characterize the physiological actions of norepinephrine in identified oxytocin and vasopressin neurons. Electrophysiological distinction between magnocellular and parvocellular neurons combined with intracellular dye injection and post hocimmunohistochemical double-labeling with selective antibodies for oxytocin and vasopressin-associated neurophysin provided a reliable means of identifying PVN magnocellular neurons.
Norepinephrine excited PVN magnocellular neurons either directly by membrane depolarization or indirectly via an increase in EPSPs. The depolarizing effect of norepinephrine was mediated by α1-receptor activation and was similar qualitatively to that described in magnocellular neurons of the SON (Randle et al., 1986a), and it was therefore not characterized in further detail in this study. The norepinephrine-induced increase in synaptic activity in magnocellular neurons, on the other hand, has not been reported previously and was the primary focus of this study.
Several observations indicate that the increase in EPSPs was caused mainly by a presynaptic action of norepinephrine. The norepinephrine-induced synaptic response was TTX-sensitive and was characterized by a robust increase in the frequency of EPSPs, suggestive of an enhanced probability of transmitter release. The majority of the cells that responded to norepinephrine with an increase in EPSPs (79%) did not show any change in resting membrane potential or input resistance, indicating that norepinephrine was not acting on postsynaptic conductances active at resting potential. Similarly, norepinephrine had no effect on depolarizing voltage responses to positive current pulses in these cells, suggesting that it was not acting on postsynaptic voltage-gated conductances. However, the norepinephrine-induced increase in EPSP amplitudes suggests that norepinephrine may also have a postsynaptic modulatory effect on ionotropic glutamate receptor-mediated currents in these cells, but these actions of norepinephrine appear less robust than the presynaptic actions.
The norepinephrine-induced increase in EPSPs was most likely due to the activation of receptors located in the somatic/dendritic region of presynaptic neurons. The response was blocked completely by TTX and was therefore dependent on spike generation, presumably at the initial segment of the axon. That the response was not mediated by modulation of spike-dependent conductances at presynaptic terminals or preterminal axons (Lena et al., 1993) is suggested by our experiments involving glutamate microdrops, whose excitatory actions should be restricted to the somatic/dendritic membrane (Christian and Dudek, 1988; Schrader and Tasker, 1997). Glutamate microdrops elicited EPSPs in a proportion of PVN cells comparable to the proportion of cells that responded to norepinephrine microdrops (28 vs 33%, respectively). If the glutamate and norepinephrine microdrops acted on the same presynaptic cells, which seems a likely possibility, then the norepinephrine, like the glutamate, was probably acting at presynaptic somatic/dendritic receptors.
Our data suggest that norepinephrine activated presynaptic glutamate interneurons located within, or in very close proximity to, the PVN. The fast kinetics of the EPSPs and their sensitivity to ionotropic glutamate receptor antagonists indicate that they were mediated by glutamate release. The glutamatergic synaptic inputs originated in neurons located inside the PVN or very close to the periphery of the nucleus because both norepinephrine microdrops applied within the PVN and bath application of norepinephrine in a surgically isolated PVN elicited EPSPs. The percentage of cells responding to norepinephrine with augmented EPSPs increased in the isolated PVN preparation, from ∼40 to >60%, which could be explained by excitatory actions of norepinephrine on perinuclear inhibitory neurons (Boudaba et al., 1996) that innervate the presynaptic glutamate neurons in the PVN. Severing these projections with surgical isolation of the PVN could lead to disinhibition of the PVN glutamate interneurons.
Twenty of the norepinephrine-responsive magnocellular neurons were identified immunohistochemically as oxytocinergic or vasopressinergic (i.e., labeled positive for one peptide and negative for the other). Approximately 63% of the neurons that responded to norepinephrine with a direct depolarization (five of eight) and 75% of the neurons that responded with an increase in EPSPs (9 of 12) were oxytocinergic, whereas the remaining cells, 37% and 25%, respectively, were vasopressinergic. It is not possible from these data to ascertain whether this represents a preferential responsiveness of oxytocin neurons to norepinephrine inputs or a sampling bias in our recordings. However, it is clear that both oxytocin and vasopressin neurons of the PVN express functional α1-adrenoreceptors and receive local excitatory synaptic inputs from α1-adrenoreceptor-expressing glutamate neurons in the PVN. A schematic diagram of a proposed model circuit is presented in Figure 8.
The intranuclear origin of the glutamatergic inputs to magnocellular neurons raises the question of the identity of the presynaptic cells. There are several possibilities, including a separate population of glutamatergic interneurons, a subtype of PVN parvocellular neuron that co-expresses glutamate, and other magnocellular neurons that co-express glutamate along with vasopressin or oxytocin. In the course of this study, we recorded the responses of both putative magnocellular and putative parvocellular neurons to norepinephrine, and interestingly, very few putative parvocellular neurons (∼2%) were depolarized by norepinephrine (Daftary et al., 1996) [note that all nonmagnocellular neurons located in the PVN were classified as parvocellular neurons;Hoffman et al. (1991)]. In contrast, 23% of the putative magnocellular neurons showed a relatively robust depolarization (7.27 ± 0.6 mV) in response to norepinephrine, as described above. This depolarization was strong enough in some of the neurons to generate a train of action potentials (Fig. 1 A), which if the cells co-released glutamate would induce a robust increase in EPSPs postsynaptically, similar to the synaptic response to norepinephrine reported here. Although more experiments are necessary to determine the identity of the presynaptic neurons, our current working hypothesis is that the glutamatergic inputs to oxytocin and vasopressin neurons arise either from other magnocellular neurons or from a separate, sparsely distributed population of glutamatergic interneurons in the PVN.
Indirect noradrenergic activation of oxytocin neurons via glutamate interneurons could account for seemingly paradoxical findings from both anatomic and physiological studies. Although there is compelling evidence that noradrenergic afferents and norepinephrine play a critical role in the control of oxytocin and vasopressin release (Tribollet et al., 1978; Clarke et al., 1979; Randle et al., 1986b;Crowley and Armstrong, 1992), several immunohistochemical studies have suggested that noradrenergic inputs are less densely concentrated in intranuclear regions occupied by oxytocin neurons than those in which vasopressin neurons predominate (McNeill and Sladek, 1980; Swanson et al., 1981; Hornby and Piekut, 1987; Cunningham and Sawchenko, 1988;Ginsberg et al., 1994; although see Michaloudi et al., 1997). Our current findings provide a physiological correlate to recent observations of noradrenergic synapses directly on both oxytocin and vasopressin neuronal somata (Michaloudi et al., 1997), but they also suggest that many of the oxytocin and vasopressin neurons in the PVN receive an indirect input from noradrenergic afferents by way of glutamatergic relay cells. This is consistent with the facilitatory interaction between adrenoreceptor and glutamate-receptor mechanisms on oxytocin release described in the lactating rat (Parker and Crowley, 1993b). These data also are in line with the observation that norepinephrine levels in the SON and PVN show a prolonged increase before and during the parturition-associated release of oxytocin, whereas glutamate levels rise sharply just before oxytocin release and subside rapidly (Herbison et al., 1997). Taken together, these findings suggest that noradrenergic afferents may be activated by sensory inputs in a tonic or slow phasic manner and that after a latency caused by an as-yet-unknown gating mechanism at the level of the PVN (and SON?), glutamate interneurons are stimulated to fire abruptly to trigger oxytocin neuron activation and bolus release of oxytocin. Parker and Crowley (1993b) found that the increase in oxytocin release caused by norepinephrine application in the SON in vivo was blocked by an AMPA receptor antagonist, but also that the glutamate-induced increase in oxytocin release was attenuated by an adrenoreceptor antagonist. This suggests that the situation is probably more complicated than the simple monosynaptic and disynaptic PVN circuits presented in Figure 8.
Although the role of the intranuclear glutamate circuits in the control of oxytocin and vasopressin release is not yet known, their presence provides a significant potential mechanism for the generation and coordination of the patterned electrical activity seen in oxytocin and vasopressin neurons. A recent report showed that local glutamatergic circuits in hypothalamic slice cultures are capable of driving bursting activity in individual oxytocin neurons (Jourdain et al., 1998). In addition to serving as a pattern generator, a small group of glutamate interneurons could provide synchronizing inputs to oxytocin neurons in the PVN, and if these glutamatergic projections extended to the other magnocellular nuclei, they could drive the synchronous bursting activity characteristic of these neurons during the milk ejection reflex and parturition. Future studies will need to determine whether similar intranuclear glutamatergic circuits are present in lactating female rats and whether glutamatergic projections serve to interconnect the magnocellular nuclei.
This work was funded by National Institute of Neurological Disorders and Stroke Grant NS31187. S.S.D. was partially supported by a predoctoral fellowship from the Louisiana American Heart Association. We thank Dr. H. Gainer for supplying oxytocin and vasopressin antibodies, Dr. A. Fancsik for his help with the computer analyses, and Drs. Shi Di and Andrei Belousov for their critical reading of this manuscript.
S.S.D. and C.B. contributed equally to this study
Correspondence should be addressed to Jeffrey Tasker, Department of Cell and Molecular Biology, Tulane University, New Orleans, LA 70118.