Report

Maternal Oxytocin Triggers a Transient Inhibitory Switch in GABA Signaling in the Fetal Brain During Delivery

See allHide authors and affiliations

Science  15 Dec 2006:
Vol. 314, Issue 5806, pp. 1788-1792
DOI: 10.1126/science.1133212

Abstract

We report a signaling mechanism in rats between mother and fetus aimed at preparing fetal neurons for delivery. In immature neurons, γ-aminobutyric acid (GABA) is the primary excitatory neurotransmitter. We found that, shortly before delivery, there is a transient reduction in the intracellular chloride concentration and an excitatory-to-inhibitory switch of GABA actions. These events were triggered by oxytocin, an essential maternal hormone for labor. In vivo administration of an oxytocin receptor antagonist before delivery prevented the switch of GABA actions in fetal neurons and aggravated the severity of anoxic episodes. Thus, maternal oxytocin inhibits fetal neurons and increases their resistance to insults during delivery.

Delivery is a stressful event associated with high risks to the fetal brain (1); however, whether the fetal brain prepares for delivery remains largely unknown. We addressed this issue by studying γ-aminobutyric acid (GABA)–mediated (GABAergic) signaling in the rat hippocampus. GABA is the principal inhibitory neurotransmitter in the adult brain. However, during fetal and postnatal periods GABA has a depolarizing action (24) and provides the major, and often the only, excitatory synaptic input to immature neurons (5). Excitatory action of GABA is due to elevated intracellular chloride and depolarized value of the GABA type A (GABAA) reversal potential (EGABA) (610). Excitatory GABA is pivotal in control of neuronal firing, generation of the primitive patterns of activity, intracellular calcium signaling, and neuronal development (2, 3). GABA signaling is therefore expected to be an essential element in physiological adaptation to stress during delivery.

To characterize the properties of GABA signaling during the perinatal period, we used cell-attached recordings from fetal and neonatal rat hippocampal slices [from embryonic day (E) 18 to postnatal day (P) 5; term is E21] (11). During fetal (E18 and E19) and postnatal periods (P1 to P5), activation of GABAA receptors increased the firing of action potentials in the majority of CA3 pyramidal cells [Fig. 1A; also (6, 7, 12)]. However, during a brief period extending from E20 to day of birth (P0), the proportion of cells excited by GABA sharply decreased. The loss of the excitatory effect of GABA peaked at E21 [∼1 to 2 hours before delivery (Fig. 1A)].

Fig. 1.

Transient perinatal loss of the GABAA-mediated excitation. (A) Responses of CA3 pyramidal cells recorded in cell-attached mode to the GABAA agonist isoguvacine. Below, summary plot of the proportion of cells excited by isoguvacine during the perinatal period. There is a transient loss of the excitatory effect of isoguvacine near term. Red corresponds to the fetuses whose mothers received SSR126768A. [E21 corresponds to the early phase of delivery (1 to 2 hours before birth); P0 is the day of birth; pooled data from 146 neurons.] (B) Cell-attached recordings of single GABAA channels with 1 μM of GABA in patch pipette (top trace); the channels were not observed in the presence of the GABAA antagonist picrotoxin (100 μM; bottom trace). (C) I-V relationships of the currents through GABAA channels in two cells at E21 and E18; their reversal potential corresponds to DFGABA. (D) Summary plot of the age dependence of DFGABA inferred from single GABAA channels recordings [mean ± SEM; 209 CA3 pyramidal cells (◯) and 17 neocortical pyramidal cells (▢); 6 to 24 patches for each point]. Red indicates pretreatment with SSR126768A (n values are 25 hippocampal and 9 neocortical patches). (E) Age dependence of the resting membrane potential (Em) of CA3 pyramidal cells inferred from the reversal of single NMDA channels recorded in cell-attached mode (n = 84 cells; 4 to 12 patches for each point). (F) Age dependence of the GABAA reversal potential (EGABA = Em + DFGABA). There is a transient hyperpolarizing shift of EGABA near birth.

The action of GABA depends on the direction of the transmembrane current elicited by GABA, its driving force (DFGABA) being the difference between EGABA and resting membrane potential (Em). Positive values of DFGABA determine excitatory actions of GABA in immature neurons (2, 3, 6, 7). To estimate DFGABA, we used cell-attached recordings of single GABAA channels (10) (Fig. 1, B to D), which affect neither EGABA nor Em. With 1 μM GABA in the pipette solution, GABAA channels were observed in ∼90% of patches (n = 553) but not in the presence of the GABAA receptor antagonist picrotoxin [100 μM; n = 25 (Fig. 1B)]. We found that DFGABA is strongly depolarizing in fetal and postnatal periods (Fig. 1D). However, DFGABA negatively shifted during a brief near-term period (from E20 to P0), switching to a hyperpolarizing value of –8.4 ± 2.5 mV at term [mean ± SEM, n = 34 (Fig. 1D)]. Thus, the disappearance of GABA-mediated excitation at term (Fig. 1A) coincides with a switch in polarity of GABA signals.

The perinatal changes in DFGABA could be due to a negative shift of EGABA or to a depolarizing shift of Em. By using cell-attached recordings of single N-methyl-d-aspartate (NMDA) channels as voltage sensors (12, 13), we found that during early fetal and postnatal periods Em is –78.7 ± 1.9 mV (n = 42; pooled data for E18 and E19 and P1 to P5) [also (13, 14)], but near term there is a small hyperpolarizing shift to –85.4 ± 1.5 mV [n = 14 at birth (Fig. 1E)]. Knowing DFGABA and Em, we calculated EGABA (EGABA = DFGABA + Em) and found that EGABA switches from –40 mV at E18 to –92 mV at E21 and then returns to depolarizing values shortly after birth (Fig. 1F). This corresponds to a decrease in [Cl]i (intracellular chloride concentration) from 18 mM to 4 mM (Fig. 1F).

Because GABA-induced depolarization raises [Ca2+]i (12, 1518), we used a multibeam two-photon microscope to monitor [Ca2+]i changes in hundreds of CA3 neurons in slices loaded with the calcium indicator fura 2-AM (Fig. 2). The GABAA receptor agonist isoguvacine (10 μM) elicited a robust increase of [Ca2+]i in the majority of neurons at E18 and at P2 to P4 [average of 56 ± 3%, n values of 7 movies and 931 cells (Fig. 2, B to D)]. In contrast, at E21 to P0, [Ca2+]i increased in only 31 ± 6% (n values of 10 movies and 1546 cells, P < 0.01) cells. Isoguvacine also decreased the frequency of spontaneous calcium events in many neurons near term [9 ± 2% (Fig. 2)]. Thus, during delivery, the ability of GABA to increase [Ca2+]i is significantly reduced.

Fig. 2.

Perinatal effects of GABAA receptors activation on [Ca2+]i in CA3 neurons. (A) Two-photon [Ca2+]i fluorescence image of the CA3 region from a P0 hippocampal slice loaded with fura 2-AM; scale bar indicates 100 μm. (Right) Automatically detected contours of the cells. (B) Three types of [Ca2+]i responses (ΔF/F) to bath application of the GABAA agonist isoguvacine (10 μM, 1 min): excited (red), inhibited (blue), or not affected (black) by isoguvacine (Isog). (C) Contour maps representing the distribution of different cellular responses to isoguvacine at E18, P0, and P5: excited (red), inhibited (blue), and not affected (open contours). Bottom histograms show the percentage of cells that are detected as being active on each movie frame corresponding to the above contour plots (140 ms per frame). Isoguvacine application (10 μM) produces a significant increase in the fraction of active neurons at E18 and P5, whereas it slightly decreased the activity at birth. (D) Histograms of the averaged fraction of imaged neurons excited (top) or inhibited or nonaffected (bottom) by isoguvacine at different ages (for P0, n values of 10 movies and 1546 cells; for E18, n values of 2 movies and 375 cells; and for P5, n values of 3 movies and 328 cells; *P < 0.01). Error bars indicate SEM.

What underlies the near-term switch in GABA actions? Because this phenomenon is not observed in fetal neurons grown in culture (8, 9), we hypothesized that it is related to parturition and in particular to maternal hormones released during delivery. Parturition is initiated by a massive release of oxytocin (19). In addition to its pivotal role in parturition, there are also indications that oxytocin exerts multiple effects in the adult central nervous system (1925). Maternal oxytocin easily crosses the placenta to reach the fetus (26), suggesting that oxytocin may be responsible for the hyperpolarizing switch in action of GABA in fetal neurons during delivery. With selective antibodies, we found a high density of oxytocin receptor immunoreactivity in the hippocampus and neocortex during the perinatal period (Fig. 3A and figs. S1 and S2). During cell-attached recordings from slices at E18 and P2, applications of oxytocin (1 μM) induced a negative shift in DFGABA (Fig.3,C and D) and suppressed GABA-mediated excitation (Fig. 3B). The effects of oxytocin were completely prevented by bath application of the selective oxytocin receptor antagonist atosiban (AT, 1 to 5 μM) (Fig. 3D). At term, application of oxytocin did not cause a significant effect on the hyperpolarizing DFGABA (Fig. 3D), suggesting that the effects of exogenous oxytocin are occluded by the endogenous hormone. In keeping with this hypothesis, we found that atosiban switched DFGABA from hyperpolarizing to depolarizing at term but not in fetal and postnatal neurons (Fig. 3D). Moreover, intracardial perfusion of E21 fetuses with artificial cerebrospinal fluid (ACSF, to wash out endogenous hormone) produced a similar shift of DFGABA from hyperpolarizing to depolarizing (fig. S3). Addition of 1 μM oxytocin at the end of the perfusion restored hyperpolarizing values of DFGABA (fig. S3). Therefore, endogenous oxytocin is present in slices at term and provides tonic activation of oxytocin receptors, leading to the shift of DFGABA. We also found that the negative shift in DFGABA from 38.4 ± 6.2 mV (n =8) to –7.2 ± 2.2 mV (n = 18) occurred in E18 fetuses obtained from mothers treated with oxytocin (50 μg/kg) during 2 hours before cesarean delivery.

Fig. 3.

Oxytocin causes a switch in GABAA signaling from depolarizing to hyperpolarizing. (A) Oxytocin-receptor immunostaining of a P0 rat hippocampus; (inset) CA3 pyramidal cells layer. Bars, 200 μm and (inset) 20 μm. (B) Histograms of the proportion of cells excited by brief application of isoguvacine (Isog) in control and in the presence of oxytocin (1 μM) at E18 and P2 (n = 57 cells). (C) I-V relationships of the currents through single GABAA channels recorded from two CA3 pyramidal cells at E18 in control and after addition of oxytocin (1 μM). (D) Histograms of DFGABA measured at E18, E21, and P2 in control conditions and in the presence of oxytocin (1 μM), antagonist of oxytocin receptors atosiban (1 to 5 μM), and oxytocin plus atosiban (n = 187 cells). (E) Two-photon imaging of [Cl]i in P4 hippocampal slice loaded with Cl indicator MQAE. Scale bar, 100 μm. (Right) [Cl]i fluorescence change produced by oxytocin application. The intensity of the fluorescence signal increases, and [Cl]i decreases, in the illustrated region. (F) Automatically detected contours of the cells imaged in (E), indicating the distribution of cells in which oxytocin application produced a significant decrease in [Cl]i (blue filled contours). On the right, chloride fluorescence changes (ΔF/F) in four representative neurons. Time resolution is 100 ms per frame.

To directly measure the effects of oxytocin on [Cl]i, we used two-photon chloride imaging in slices loaded with a chloride-sensitive dye, MQAE (Fig. 3, E and F) (27). In E18 and P4 slices, oxytocin (1 μM) produced a significant increase in the baseline fluorescence signal in half of the imaged cells, indicating a strong reduction of [Cl]i [average of 43 ± 4%, n values of 10 movies and 1792 cells (Fig. 3F)], and the effect was prevented by atosiban (5 μM; reduction of [Cl]i in 5 ± 3% cells, n values of 2 movies and 435 cells). Furthermore, incubation of P4 slices with oxytocin reversed the effect of isoguvacine on spontaneous calcium events [59 ± 17% imaged cells not activated by isoguvacine, n values of 3 movies and 228 cells (fig. S4)]. Thus, the effects of oxytocin fully matched the changes in GABA signaling occurring at term. This result indicates that oxytocin is sufficient to trigger changes of [Cl]i.

To determine whether endogenous oxytocin is necessary for the near-term switch in the action of GABA, we treated pregnant rats orally with the selective oxytocin receptor antagonist SSR126768A (28) (1 mg/kg, starting from E20) and measured the consequences at term. Cell-attached recordings revealed major differences between treated and age-matched control pups, including strongly depolarizing values of DFGABA (Fig. 1D, red circle) and a high proportion of cells excited by GABA (Fig. 1A, red circle). Similar effects were also observed in neocortical neurons, in which DFGABA was –9.4 ± 3.4 mV in control rat pups (n =8) and 28.3 ± 3.7 mV (n = 9) in pups of mothers pretreated with SSR126768A (Fig. 1D). Thus, maternal oxytocin is necessary and sufficient to trigger the near-term switch in GABA action in the hippocampus and neocortex.

What are the mechanisms underlying the oxytocin-mediated reduction of [Cl]i at term? Because the NKCC1 chloride inward cotransporter is responsible for elevated [Cl]i in immature neurons (4, 7, 29, 30), we tested the effects of the selective NKCC1 antagonist bumetanide. We found that bumetanide (10 μM) produces a robust negative shift of DFGABA and occludes the effects of oxytocin (fig. S5). The short latency of oxytocin actions (Fig. 3F) and the observation that blockade of oxytocin receptors does not alter chloride transporters gene expression (fig. S6) suggest that the hormone down-regulates NKCC1 activity.

Hypoxic-ischemic brain damage is a principal cause of newborn death and neurological impairment (1). We hypothesized that the excitatory-to-inhibitory switch in GABA signaling reduces neuronal activity and metabolic demand, thus helping to protect fetal neurons from hypoxic insults (31). Episodes of anoxia and aglycemia were induced by superfusion with a solution in which oxygen was substituted for nitrogen and glucose was substituted for sucrose. The onset of anoxic depolarization (AD) that is an electrophysiological marker of neuronal death (31, 32) was measured by using extracellular field potential recordings in the intact hippocampi of E21 rats in vitro after various treatments in vivo to the mother and/or to the fetuses intracardially before the in vitro experiment (Fig 4). In control fetuses, AD occurred after 55.6 ± 1.4 min (n = 30) of perfusion with the anoxic-aglycemic solution (Fig. 4). In the hippocampi prepared from the fetuses intracardially perfused with atosiban (5 μM; n = 15) or from the fetuses whose mothers received SSR126768A (1 mg/kg; n = 19), AD onset was significantly accelerated to 44.1 ± 1.1 min (n = 34; P < 0.01). Bumetanide (10 μM) applied in the presence of the oxytocin receptor antagonists delayed AD to the control values (53.4 ± 1.7 min; n = 18; P < 0.01). Bumetanide (10 μM) occluded the effects of the oxytocin receptor antagonists and restored the control delays (53.4 ± 1.7 min, n =18; P < 0.01). These data indicate that maternal oxytocin exerts a neuroprotective action on fetal neurons during parturition and that this action is likely due to a reduction of [Cl]i.

Fig. 4.

Blockade of oxytocin receptors decreases fetal brain resistance to anoxia-aglycemia at birth. (A) Representative traces to illustrate the effects of anoxic-aglycemic solution on extracellular field potential recordings from E21 intact hippocampi. Arrows indicate terminal AD that marks neuronal death. AD occurs earlier in the presence of SSR126768A (middle trace). Addition of bumetanide (10 μM) occludes the effects of the antagonist and restores the initial delay (bottom trace). (B) Summary plot of the onset of AD in control, in the presence of the oxytocin receptors antagonists atosiban (AT, 5 μM, fetal intracardial perfusion and SSR126768A (1 mg/kg to the mother), and after further addition of bumetanide (10 μM). Each circle corresponds to one hippocampus (n = 82 intact hippocampi at E21). Error bars indicate SEM.

Our results suggest that oxytocin, in addition to its well-established role in labor and lactation and its multiple effects in the adult central nervous system (1925), also exerts a powerful action on fetal neurons. This mechanism adds a previously unknown facet to the plasticity of GABA signaling via modulation of [Cl]i (4, 33). The dual action produced by a single messenger in the mother and fetus enables a perfect timing for adaptation of fetal neurons to delivery.

Supporting Online Material

www.sciencemag.org/cgi/content/full/314/5806/1788/DC1

Materials and Methods

Figs. S1 to S6

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article