Emotion and Mood Adaptations in the Peripartum Female: Complementary Contributions of GABA and Oxytocin

Abstract

Peripartum hormones and sensory cues from young modify the maternal brain in ways that can render females either at risk for, or resilient to, elevated anxiety and depression. The neurochemical systems underlying these aspects of maternal emotional and mood states include the inhibitory neurotransmitter GABA and the neuropeptide oxytocin (OXT). Data from laboratory rodents indicate that increased activity at the GABA_A receptor contributes to the postpartum suppression of anxiety-related behaviour that is mediated by physical contact with offspring, whereas dysregulation in GABAergic signalling results in deficits in maternal care, as well as anxiety- and depression-like behaviours during the postpartum period. Similarly, activation of the brain OXT system accompanied by increased OXT release within numerous brain sites in response to reproductive stimuli also reduces postpartum anxiety- and depression-like behaviours. Studies of peripartum women are consistent with these findings in rodents. Given the similar consequences of elevated central GABA and OXT activity on maternal anxiety and depression, balanced and partly reciprocal interactions between these two systems may be essential for their effects on maternal emotional and mood states, in addition to other aspects of postpartum behaviour and physiology.

Introduction

Some of the most dramatic behavioural adaptations seen in adult mammals occur across the transition from late pregnancy to the early postpartum period. For example, by the time females give birth, any aversion toward neonates is replaced by avidity and relative passivity is supplanted by pugnaciousness (1–6). A critical influence on these and probably most other peripartum behavioural adaptations is the mother's emotional and mood states. It has long been thought that emotional hyper-reactivity in laboratory animals, including high anxiety-related behaviours, can interfere with maternal caregiving and aggression (1,7). More recently, some attention has been paid to the possibility that emotional hyporeactivity, involving abnormally low anxiety or possibly high depression-like behaviours, may be equally detrimental to rodent mothering (8,9). In humans, emotional and mood states that are neither too high (e.g. anxiety, mania), nor too low (e.g. lack of concern, depression) are similarly optimal for mothers to best attend to the needs of their infants (10,11).

A plethora of neurochemicals including steroid hormones, neuropeptides and classic neurotransmitters, fluctuates across late pregnancy and the early postpartum period to regulate maternal physiology, caregiving behaviours, cognition, emotions and mood (12–18). Our purpose here is not to exhaustively review aspects of these literatures instead, to focus on the studies explicating how maternal anxiety- and depression-related behaviours in laboratory rodents are influenced by central nervous system activity of the inhibitory neurotransmitter, GABA, and the neuropeptide oxytocin (OXT). We also highlight some of the many compelling studies implicating these neurochemical systems in regulating anxiety and depression in human mothers. Because it will be clear that elevated GABA and OXT signalling can each reduce symptoms of anxiety and/or depression in both nonhuman animals and humans, we conclude by proposing that the balanced activity of these two systems, and perhaps reciprocal interactions between them, may be essential for their positive effects on maternal emotions and mood.

Anxiety and depression across pregnancy and the early postpartum period

Humans

Anxiety

There are considerable individual differences in the trajectory of anxiety across pregnancy and the postpartum period in women because it is strongly influenced by social and experiential factors. Nevertheless, for many women, these times of the reproductive cycle fortunately involve either little change or even a reduction in their pre-existing anxiety symptoms (19–21). A variety of hormones and other neuro-chemicals released at parturition and when human mothers receive suckling or other tactile inputs from their infants, including GABA and OXT, are presumed to help protect against elevated anxiety (13,22–24). Unfortunately, not all postpartum women are so well protected. A recent review of the literature revealed that rates of diagnoses for obsessive–compulsive disorder [OCD; characterised by obsessive or intrusive thoughts and compulsive behaviours that interfere with everyday life; Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV] (25) and generalised anxiety disorder (GAD; characterised by excessive and irrational worry for longer than 6 months) (25) are higher in pregnant or postpartum women compared to the general population (rates of OCD = 3–4% versus 1–2%; GAD = 4–8% versus 3–4%) (26). These increases partly reflect both the higher susceptibility for relapse in women who are particularly at-risk for anxiety disorders due to their history of high anxiety (19,21,27) and cases of new onset of OCD during the peripartum period (26). Elevation in these anxiety disorders is of particular concern because they produce tremendously detrimental effects on mothers' general well-being, their ability to provide sensitive caregiving, and numerous aspects of infant development (28,29). Importantly, the lack of regular screening for anxiety in the peripartum population and an emphasis of public mental health campaigns on peripartum depression lead to significant underreporting of women's anxiety symptoms (30). Consequently, the problem of high peripartum anxiety is often considered to be much greater than it may at first appear, with subclinical and clinical elevations perhaps affecting up to 20–30% of women (26,27,31,32).

Depression

Depressive symptoms are also influenced by reproductive events in women. Postpartum ‘blues’, a transient and mild condition characterised by mood disturbances beginning a few days after parturition and lasting less than 2 weeks (therefore not fulfilling the time criterion for major depression), is extremely common, with prevalence estimates suggesting that it affects up to 84% of childbearing women (33,34). Research on postpartum depression, however, is particularly hampered by inconsistency of its definition with regard to the time criterion of the maximum interval between parturition and onset of the major or minor depressive episode. The definitions have ranged from 4 weeks (DSM-IV, specifier: depressive episode ‘with postpartum onset’) (25), to over 6 weeks [International Classification of Diseases (ICD)-10, depression specifier: ‘as associated with the puerperium’] (35), and up to 12 months (‘perinatal depression’) (36). Given this heterogeneity in postpartum depression criteria, the multitude of instruments used to assess it, and the lack of differentiation between point-, 1-month or other period prevalences, it is probably not surprising that reported rates of peripartum depression vary quite widely across studies.

Nonetheless, meta-analyses indicate that the point prevalence of depression is highest in the third month postpartum (but based on large confidence intervals of the estimates) and that the incidence of depression (i.e. the percentage of women with depressive episodes that begin during pregnancy or postpartum) is comparable between pregnancy and the first 3 months postpartum. This notably underscores the relevance of the postpartum period as a time of increased risk for depression onset because the duration of pregnancy is approximately 300% longer than the first 3 months postpartum (36,37). Additionally, when directly comparing samples of childbear-ing women from conception until 12 months postpartum with mothers who had not recently given birth to their children, there were no substantial differences in depression prevalence and incidence estimates, with the exception of one study indicating an increased risk of newly developing depression during the first 5 weeks postpartum (36,38). Initially, this contradictory picture may be explained not only by mere inconsistencies between studies, but also by a longer history of depression in the nonrecent mothers, whose depression onset was triggered by their longer-ago deliveries (38). Similar to anxiety, women with a pre-existing mood disorder (depression or bipolar disorder) are especially at risk for relapse of symptoms during the first few weeks after giving birth (39–41). Also similar to anxiety, depression is not regularly screened for and surely under-reported in the peripartum population, so even though the risk for depressive episodes during the postpartum period does not appear to be considerably increased compared to nonchildbearing times, postpartum depression is without doubt a common condition.

A neuroendocrine basis for postpartum depression involving the withdrawal of hormones at parturition has often been presumed, so it is perhaps surprising that adoptive mothers of infants or toddlers also show higher depressive symptoms soon after becoming a parent (42,43). Depression is also common for new fathers (44). Endocrine flux can occur in nonparturient humans when they become parents and interact with infants (45), although the changes are certainly less dramatic than those occurring in recently parturient women. Thus, both non-endocrine factors (e.g. environmental and personality factors) and endocrine factors even unrelated to reproductive state are relevant for depression in new human parents.

Laboratory rodents

Anxiety-like behaviours

The temporal progression of anxiety-related behaviour in most reproductive female laboratory rats is generally characterised by an increase during some points of pregnancy followed by a postpartum decrease to levels that fall even below those seen in cycling virgins. Specifically during pregnancy, anxiety is higher during the second week and again a few days before parturition compared to before pregnancy or during very early pregnancy (46–50). There are a few reports, however, showing no difference between late-pregnant and nulliparous rats in their anxiety-related behaviour (51,52) or a decrease in anxiety on some days of the last week of pregnancy (53,54). There may also be species differences in anxiety profiles across pregnancy because both early and late pregnancy are associated with reduced responses to novelty in ewes (55).

Regardless of the reported changes in anxiety during pregnancy, most studies find that, soon after giving birth, most female rodents and sheep show lower indices of anxiety compared to nulliparous females (12,13,52). The anxiolytic effect of current maternal state in primiparous rats appears to last only through the first postpartum week and requires recent physical contact with the pups (either suckling or nonsuckling) (56). Importantly, not every mother rat responds to offspring touch with a reduction in anxiety because dams with the lowest trait anxiety are more anxious after having spent time with their litter, indicating important individual differences in the factors influencing postpartum anxiety, which can be overlooked when examining mothers as a whole (9). Importantly, unlike recently parturient female rats that as a whole show relatively low anxiety, pseudopregnant rats that are ovariectomised and virgins given a regimen of pregnancy-like ovarian hormones that is rapidly terminated show either increased (57), decreased (58) or unchanged (59) anxiety-related behaviour. Such results make the obvious point that these experimental models do not accurately mimic the endocrine, sensory and neural events occurring with natural pregnancy and parturition.

Depression-like behaviours

A burgeoning but still relatively small animal literature indicates that depression-like behaviours in female rodents (most often tested in the forced swim and sucrose preference tests) are particularly low during the beginning of the third week pregnancy, but that these behaviours mostly do not differ among females tested at the very end of pregnancy, during the early postpartum period, or when nulliparous (47,50,60–65). These studies can be somewhat difficult to interpret because the depression paradigms used do not take into consideration the pregnancy-specific physiological and behavioural alterations, including increased body surface and body fat relevant for interpreting behaviour in the forced swim test or altered food demands relevant for the sucrose preference test, and so they might not provide a uncomplicated reflection of depression-like behaviour. Studies using tests of saccharin rather than sucrose preference may avoid the metabolic issues and, indeed, pregnant and lactating rats have been observed to have a lower preference for saccharin than do virgins (66). This is consistent with the findings that late pregnant and early postpartum female rats also have reduced interest in other nonpup rewards, such as cocaine (67,68).

Rodent models attempting to mimic the endocrine flux of parturition by using ovariectomised nulliparous females treated with progesterone and/or oestradiol followed by their abrupt withdrawal show increased or occasionally decreased depression-like behaviours (59,69–73). Valuable insight can be gained by the fact that the depression-like behaviours (and anxiety-like behaviours; see above) observed in most steroid-hormone withdrawal studies are dissimilar to that found in naturally postpartum animals. It is likely that other neurochemicals changing during the very early postpartum period in response to interactions with neonates (including GABA, OXT and prolactin) prevent any increase in anxiety- and depression-like behaviours in parous females after physiologically occurring steroid withdrawal (14,74,75). Furthermore, the rapid withdrawal of progesterone in most pharmacological steroid studies is temporally dissimilar to the less-abrupt progesterone decline normally occurring at parturition. This is relevant because slow progesterone withdrawal elicits less dramatic changes in females' anxiety-related behaviours compared to what is found after the hormone's sudden absence (76,77) and the same may be true for depression-like behaviors.

Modification of central GABA release during the peripartum period

GABA is the primary inhibitory neurotransmitter in the brain and there is a vast scientific literature on its involvement in anxiety in humans and other animals (78,79). GABA acts on at least three distinct transmembrane receptors, the ionotropic GABA_A receptor (GABA_AR) that is comprised of five of up to 19 receptor subunits, a metabotropic GABA_B receptor existing as hetero- or homodimers comprised of two receptor subunits, and the relatively poorly studied ionotropic GABA_C receptor that is similar to the GABA_AR in that it is also made from five subunits. Activity of GABA_ARs is traditionally considered to be of the utmost importance for modulating anxiety, although there is emerging evidence for a role of the GABA_B receptor (80,81). Expectedly, the density of GABA_ARs and expression of its receptor subunits most relevant for anxiety are particularly high in brain regions underlying anxiety (82,83).

Humans

It is intuitive that changes in central GABA release would be related to changes in anxiety across the peripartum period in humans and nonhumans animals, although there is a paucity of work on this, especially in women. Women's cerebrospinal fluid (CSF) levels of GABA have been seen to drop during late pregnancy (24) and increase during labour (84), although the importance of such changes for women's anxiety is questionable because the relationship between CSF GABA and general anxiety in humans is often not significant (85–87). Determining GABA concentrations and their effects on GABA_ARs in specific brain regions would probably be more fruitful for understanding peripartum changes in human anxiety. The only study to do so used proton magnetic resonance spectroscopy (¹H-MRS) and found that GABA levels in the occipital cortex were lower in postpartum women (up to 6 months after parturition) compared to women who had not recently given birth and examined during the follicular phase of the menstrual cycle (88). The postpartum women who were studied closest to parturition tended to have the lowest occipital GABA levels. The occipital cortex is not typically implicated in depression, and at no time point were the women's occipital GABA levels related to the presence of postpartum depression (anxiety was not assessed), so the relevance of this finding for peripartum mood or emotional regulation is unknown. Nonetheless, this study demonstrates that reproductive state affects cortical GABA in women and it will be important in future studies to examine cortical and subcortical (especially limbic) regions more closely associated with anxiety or depression.

Laboratory rodents

In laboratory rats, CSF concentrations of GABA are high in lactating rats that interact with pups, but are almost nondetectable in dams whose pups have been removed for as little as 6 h. The high GABA levels are restored after mothers and pups are reunited (89). Because anxiety-related behaviours in postpartum rats are not affected by ovariectomy, hypophysectomy or inhibiting steroidogenesis (56,90,91), the postpartum increase in GABA levels is presumably not mediated by the mother's current endocrine state but is instead is the result of rapid GABA release when mothers interact with offspring. Pregnancy and later interaction with offspring influence GABA synthesis and release in numerous specific forebrain sites involved in postpartum behaviour. In ewes, parturition and subsequent interaction with lambs increases maternal GABA release in the olfactory bulbs, medial preoptic area and bed nucleus of the stria terminalis (92,93). Altered GABA synthesis, as indicated by expression of glutamate decarboxylase (GAD), is also found in the late-pregnant and early postpartum rat olfactory system (94,95) and in the mouse rostral lateral septum (96). Baseline GABA release is lower in the basolateral amygdala of pregnant rats compared to cycling rats, which may disinhibit amygdalar output neurones and result in the increased anxiety observed during some points of pregnancy (97). In the cerebral cortex, late pregnancy and the early postpartum period in mice is associated with reduced GABA turnover compared to nulliparous females (98), but postpartum laboratory rats have higher basal GABA release and turnover in the medial frontal cortex compared to virgins (99,100). The latter finding is consistent with recent data indicating that the medial frontal cortex of postpartum rats has higher expression of both the 65-kDa molecular weight isoform of GAD (GAD₆₅) and the vesicular GABA transporter compared to dioestrous virgins, suggesting greater potential for cortical GABA synthesis and release in mothers (101) (Fig. 1).

Fig. 1.

Highly schematic representation of neuronal GABA synthesis and release. GABA is synthesised from glutamate (GLU) by two isoforms of glutamate decarboxylate (GAD). GAD₆₇ is found throughout the cytoplasm and produces the pool of GABA necessary for intracellular functions, whereas GAD₆₅ is more localised to neuronal terminals and synthesises the majority of the GABA that will be packaged by vesicular GABA transporters into synaptic vesicles for release into the synapse.

Modification of central GABA_A receptors during the peripartum period

Concomitant with the above-mentioned changes in central GABA synthesis and release across pregnancy and the early postpartum period, many studies in laboratory rodents have demonstrated dramatic plasticity of GABA_ARs and their capacity to mediate GABAergic inhibition. There is a significant increase in the affinity of total forebrain GABA_ARs for GABA in mid-to-late pregnant rats compared to cycling females, with a further increase in postpartum females despite the fact that they have reduced receptor density (102). The cortex must not drive this change in affinity because its GABA_ARs have been reported to have decreased affinity for their ligand during late pregnancy (103). Brain sites contributing to the postpartum reduction in total forebrain GABA_AR density are also unknown because [³H]flunitrazepam and [³H]muscimol binding (indicating the densities of benzodiazepine and GABA_A receptor binding sites, respectively) in numerous forebrain regions involved in anxiety-related behaviours do not differ among virgin, pregnant and early postpartum female laboratory rats (104,105).

As noted above, GABA_ARs are comprised of five of up to 19 known potential receptor subunits, and the expression of many of these subunits is affected by female reproductive state (106,107). Peripartum changes in the expression of specific GABA_AR subunits reflect a homeostatic mechanism that maintains ideal levels of inhibition in the face of elevated neurosteroids. Given that neurosteroids can act as positive allosteric modulators on GABA_ARs to enhance GABAergic inhibition and, at high levels, even directly gate these receptors, increased neurosteroid levels such as that occurring during pregnancy have the potential to dramatically alter GABAergic signalling. Remarkably, plasma and cerebral progesterone levels increase by almost 200-fold during pregnancy and this is accompanied by elevations in its metabolites allopregnanolone and tetrahydrodeoxycorticosterone (108). These neurosteroid concentrations during pregnancy can be sufficient to cause sedation in nonpregnant animals (109). Thus, GABA_AR plasticity during pregnancy is necessary to offset these incredibly elevated neurosteroid concentrations. Considerable attention has been given in this context to expression of the GABA_AR δ subunit, which is uniquely sensitive to neurosteroid modulation (110,111). δ subunit expression is down-regulated in the hippocampus during late pregnancy, and rebounds upon restoration of ovarian hormone/neurosteroid levels the early postpartum period (64,112–114). In addition to maintaining normal GABA_AR-mediated inhibition in the hippocampus in the face of elevated neurosteroids during pregnancy, plasticity of the δ subunit also occurs through the first postpartum week, when neurosteroid levels are quite low (103). During this time, we found that δ subunit expression is also up-regulated in the midbrain periaqueductal gray (PAG; considered to be a ‘final common pathway’ for anxiety and fearful behaviours and is also highly sensitive to tactile inputs from offspring) (115) in the face of what appears to be decreased GABA synthesis and release (101). Together, these changes may modify tonic inhibition of GABAergic interneurones in the PAG to help blunt anxiety- or fear-related behaviours in new mothers (116).

There are only a few studies of peripartum changes in GABA_B receptor signalling. GABA_B subunit expression in the hypothalamic supraoptic (SON) and paraventricular (PVN) nuclei does not significantly differ between virgin and lactating rats (117), although examining other sites could still be an interesting endeavour given this receptor's emerging role in anxiety and depression (80,81) and that GABA_B receptor signalling affects other postpartum processes in rats such as maternal behaviour and suckling-induced milk letdown (118,119).

GABAergic signalling and maternal anxiety- and depression-like behaviours

The homeostatic plasticity described above would require fine-tuned alterations in GABA release, GABA_AR activation, GABA_AR subunit expression, and GABA_AR affinity and density. Disruptions of this balanced regulatory process has been proposed to contribute to abnormal anxiety- and depression-like behaviours during the postpartum period (107,113). In support, treating postpartum rats with GABA_AR antagonists, including bicuculline and pentylenetetrazol, prevents their normal anxiolytic phenotype (120-122). Because nulliparous females are often at or near a ceiling for anxiety behaviours, these antagonists have considerably less of an effect on them (121,122). The specific brain sites mediating the anxiogenic effects of GABA_AR antagonism in dams include the ventrocaudal periaqueductal gray (cPAGv), where bicuculline infusion increases anxiety in mothers to levels typical of nulliparous rats (121) (Fig. 2). By contrast, the ventromedial hypothalamus and amygdala do not appear to be sites where GABA_AR activity is involved in postpartum anxiolysis (123).

Fig. 2.

Top: Peripheral injection of a higher dose of the GABA_A receptor antagonist bicuculline (+BIC) reduces the percentage of time female rats spend in the open arms of the elevated plus maze, with a greater effect in postpartum rats, as indicated by the hash symbol (#). This peripheral effect of bicuculline in dams is reproduced by site-selectively infusing bicuculline methiodide (BM) into the ventrocaudal periaqueductal gray (cPAGv), but not if the infusions missed the cPAGv. *Main effect of reproductive state. Adapted from Miller et al. (121).

Failure to properly regulate central GABA_AR subunit expression during pregnancy and postpartum has been implicated in the pathophysiology of postpartum emotional and mood dysregulation. Mice deficient in the GABA_AR δ subunit (Gabrd^-/- mice) exhibit increased depression-like behaviours during the first few days postpartum, as indicated by increased passive coping in the forced-swim test and a lower preference for sucrose over water compared to wild-type females. They also exhibit more behaviours indicative of anxiety (digging, burrowing and circling) when exposed to a novel environment (64). Importantly, such differences are not seen between virgin Gabrd^-/- and wild-type females or males (64,124), and so they may be specific to sex and reproductive state. Gabrd^-/- dams are also inadequate mothers and fail to build nests or retrieve scattered pups (Fig. 3). As a result, pups born to Gabrd^-/- mothers are more likely to die as a result of cannibalism and/or neglect. Cross-fostering experiments confirm that the increased mortality rate is a result of the abnormal maternal behaviours of Gabrd^-/- mice rather than the characteristics of the pups or an interaction between the two factors (64). Abnormal mothering is also seen in heterozygous Gabrd^-/- mice and can be rescued by selective agonism of GABA_ARs that contain δ subunits (64).

Fig. 3.

Top: Gabrd^-/- mice exhibit deficits in maternal care, including failure to build a nest and keep the pups in close proximity (right) compared to wild-type dams (left). Bottom: schematic representation of the loss of the GABA_AR δ subunit in corticotrophin-releasing hormone neurones in the paraventricular nucleus of Gabrd^-/- mice, which is associated with elevated stress reactivity, depression-like behaviours and the deficits in maternal care Adapted from Maguire and Mody (64) and Sakar et al. (226), as well as J Maguire and I. Mody, unpublished data

These data suggest that the inability to regulate δ subunit-containing GABA_ARs results in increased anxiety- and depression-like behaviours and abnormal mothering restricted to the postpartum period. Because interactions with pups reduce anxiety- and depression-like behaviours in rats (56,74), it could be possible that the reduced maternal behaviour in the Gabrd^-/- or Gabrd^-/- mice contributes to their increased anxiety- and depression-like behaviours. In addition, Gabrd^-/- mice exhibit higher hypothalamic-pituitary-adrenal (HPA) axis reactivity to stress as a result of the loss of GABAergic control of corticotrophin-releasing hormone (CRH) cells in the PVN (Fig. 3) (125). Dysregulation of the HPA axis has been implicated in the pathophysiology of depression in nonpostpartum (126,127) and postpartum humans (14,128). Indeed, blocking CRH signalling with antalarmin from days 14-21 of pregnancy in Gabrd^-/-mice decreases their depression-like behaviours postpartum and increases the survival rate of their pups (J. Maguire and I. Mody, unpublished data).

Modifications of central OXT during the peripartum period

Humans

One of the hallmarks of parturition and lactation is elevated OXT release. This occurs from the neurohypophysis into the general circulation, which facilitates parturition and milk ejection, and also is thought to occur in all mammals centrally within hypothalamic and imbic brain regions that influence maternal behaviour, anxiety- and depression-related behaviours, as well as stress coping (12,15). Comparable neurobiological mechanisms are assumed to underlie human and rodent maternal adaptations, although our information regardng changes in the OXT system in human mothers comes only from assessments of OXT in blood or CSF. Plasma OXT levels may gradually ncrease during the course of pregnancy, although some reports are inconsistent with this observation (129,130). This may be because single-point assessments of plasma OXT concentrations greatly vary between subjects as a result of rhythmic or episodic OXT release superimposed on their tonic OXT release, leading to short-term OXT fluctuations consistent with the short half-life of OXT in the peripheral circulation (129). In a study assessing OXT concentrations immediately following normal vaginal delivery, OXT levels were elevated at 15, 30 and 45 min after delivery, and returned to prepartum levels by 60 min (131). During the postpartum period, plasma OXT concentrations are mostly driven by suckling-induced bursts of OXT released during breast-feeding, with no baseline differences between breastfeeding and nonbreastfeeding women (132). With regard to the CSF, it has been suggested that its OXT may reflect centrally released OXT from dendrites and perikarya within the hypothalamus or from axon terminals in other limbic regions (15,133). In pregnant women, CSF OXT concentrations do not appear to be higher than in nonpregnant women, although there is some evidence for a rise in CSF OXT levels during labour (24,134,135).

Laboratory rodents

The much larger literature from laboratory rodents reveals that OXT synthesis is elevated within the SON and PVN (the primary neuronal sources of OXT) during the peripartum period beginning with late pregnancy (136,137). Local release of OXT within the SON and PVN, septum, preoptic area, bed nucleus of the stria terminalis and olfactory bulb is strongly stimulated by acute pelvic and other sensory stimuli received during parturition, and later by the suckling and other tactile inputs that mothers receive from their offspring (93,138,139). Such contemporaneous release of central and peripheral OXT helps ensure coordination among maternal physiological and behavioural functions requisite for offspring survival (14,140). These reproduction-related events also stimulate other neurochemical systems such as GABA and prolactin, which are up-regulated during the peripartum period (141,142). Thus, OXT is likely to act in concert with these systems to ensure not only birth, milk ejection and maternal care, but also the finely-tuned peripartum changes in anxiety- and depression-related behaviours.

In addition to OXT release, OXT receptor (OXTR) mRNA in rodents is elevated in the lateral septum, amygdala and medial preoptic area during pregnancy, and within the olfactory bulb, BNST and ventromedial hypothalamus during parturition (143,144). Consistent with these increases in mRNA expression, elevated OXTR binding is found in most of these brain regions (145). Dramatic alterations in circulating ovarian steroids, including progesterone and oestrogens, before parturition underlie the changes in OXT and OXTR expression in some brain regions (58).

OXTR-mediated signalling and maternal anxiety-related behaviour

Humans

There has been considerable interest in elucidating the role of OXT for the aetiology of anxiety disorders and other mental processes in humans (146,147). Advancing knowledge in this field is hampered by the fact that OXT concentrations in human plasma may or may not reflect brain OXT activity and therefore have to be interpreted with caution (15,147,148). Similarly, in human studies, OXT is often administered intranasally, although ultimate evidence regarding its uptake into the brain is lacking (149) but see in rats and mice (150). There are also sex differences in the relationships between OXT signalling and emotional processing in humans (151–153), emphasising that caution is warranted when extrapolating data obtained in men to women of any reproductive state. With these caveats in mind, there is a negative association between plasma OXT and aspects of anxiety in peripartum women (154,155), which can be associated with breastfeeding and physical contact with the infant (23). Fascinatingly, an association between OXT and anxiety was not found in a large sample of women who were not recently parturient (156), and so the relationship between OXT and anxiety even differs across reproductive state within sex.

Laboratory rodents

Similar to humans, the anxiolytic properties of OXT in laboratory rats and mice depends on sex and reproductive state. Acute i.c.v. administration of synthetic OXT produces inconsistent effects on anxiety-related behaviour, probably because it depends on the basal stress level of the animal. Thus, in male mice that received an i.c.v. infusion of OXT within hours after stereotaxic surgery, an anxiolytic effect is found compared to vehicle-treated surgery-stressed mice (157), whereas no such effect is found in unstressed rats tested 5 days after implantation of a guide cannula (158,159). Importantly, local infusions of OXT into either the central amygdala (140,160,161) or PVN (159,162) consistently produce anxiolytic effects in male and virgin female rats or mice. Interestingly, an involvement of endogenous brain OXT on anxiety is only found under conditions of elevated brain OXT system activity. Such a condition is a given during the peripartum period when there is high availability of OXT and its receptor in the brain (52,159). Thus, blockade of OXTR-mediated effects by i.c.v. infusion of an OXTR antagonist has anxiolytic effects in pregnant and lactating rats but not in virgin females (52). Similarly, in males, such anxiolytic properties of brain OXT are only found after mating-induced stimulation of central OXT release (163). Furthermore, chronically infusing synthetic OXT using osmotic minipumps to increase brain OXT availability in virgin, ovariectomised steroid-primed rats attenuates their emotional and neuronal responses to an acute noise stress (164,165). In mothers, one of the many sites other than the amygdala and PVN where brain OXT acts to produce its anxiolytic effect is the midbrain cPAGv. Antagonism of OXTRs in the cPAGv increases dams' anxiety-related behaviours on the elevated plus maze, whereas infusion of OXT restores low anxiety in mothers that have not recently been in physical contact with their pups (166).

The OXTR-mediated intraneuronal signalling cascades underlying the anxiolytic effect of OXT acting at least within the PVN include the mitogen-activated protein (MAP) kinase pathway. This pathway is activated within 5–10 min after OXT administration to male and virgin female rats, as indicated by increased phosphorylation of various kinases, including MAP kinase kinase (MEK). Pharmacological blockade of the MAP kinase (MAPK) cascade (i.e. blockade of MEK kinase activity) prevented the anxiolytic effect of OXT administered into the PVN of male and virgin female rats, demonstrating that OXTR-mediated activation of the MAP kinase pathway, specifically phosphorylation (p) of extracellular signal-regulated kinase (ERK) (males) (162) and MEK1/2 (females) (159) is essential for acute OXT effects on anxiety. Importantly, and in agreement with the increased presence of OXT in the local extracellular fluid as a result of pup-related stimuli and increased intra-PVN release during lactation, there is a general up-regulation of the MAPK pathway in the postpartum PVN: cytosolic pMEK1/2 levels are approximately 25% higher on postpartum day 8 compared to what is found in virgins. These high pMEK levels and subsequent nuclear translocation of pERK1 are necessary for the anxiolytic phenotype typically observed during lactation (159) because blockade of the MAPK pathway results in increased anxiety-related behaviours in dams but not virgins. Another specific peripartum adaptation in this system is that that further elevation of OXT availability by local infusion of synthetic OXT bilaterally into the PVN of postpartum rats does not result in further increase in MEK1/2 phosphorylation and/or produce the anxiolytic effect seen in male and virgin female rats (159,162). Thus, the MAPK pathway is an important intracellular mechanism activated by high extracellular OXT in the PVN and probably elsewhere in the brain to mediate postpartum anxiolysis. In this context, it is worth noting that this pathway, which is also activated by OXT in the hippocampus during lactation, has been related to improved spatial memory in postpartum mice (167).

OXTR-mediated signalling and maternal depression-related behaviour

Humans

Lower plasma OXT is associated with more depressive symptoms during the third trimester of pregnancy and at 8 weeks postpartum in women (but not at 2 weeks postpartum) (168). As evidence for a link between pre- and postpartum psychobiological processes, there is an inverse association between plasma OXT during pregnancy and postpartum depressive symptoms, even after controlling for depressive symptoms during pregnancy (Fig. 4) (148). This finding highlights the potential of plasma OXT levels during pregnancy as prepartum predictors of risk for developing postpartum depression, and it may then be speculated that the window of opportunity to modify OXT activity to reduce postpartum depression may be before parturition. Moreover, it provides evidence that, somewhat similar to the infant HPA axis response to stress, the relationship of the OXT system with maternal depression may be subject to ‘peripartum programming’ in that there is a link between maternal prepartum neuroendocrine activity and postpartum behavioural or mental processes (169,170). In line with the above-mentioned finding, OXT during pregnancy predicts maternal bonding behaviours (i.e. positive affect and gaze during interactions with infant) and cognitive attachment representations of the child, both of which are vulnerable to postpartum depression (131). On the other hand, it has been found that intranasal OXT does not make mothers with postnatal depression happier (instead it makes them sadder), even though their perception of the relationship with their baby improves (171). However, the latter finding should be interpreted with caution because the size of the sample of that study was small and individual variation in sensitivity to OXT (e.g. as a result of early experiences) may modulate the biobehavioural effects of intranasally applied OXT (171–174).

Fig. 4.

Individual oxytocin (OXT) concentrations, as well as group means, in a group of women at risk and a group of women without risk for developing postpartum depression (PPD). OXT values are shown on a logarithmic scale. Women at risk have lower OXT concentrations. Adapted from Skrundz et al. (148).

The neurobiological mechanisms underlying OXT-related mood adaptations during the peripartum period in humans are mostly unknown. Some insight can be garnered from studies of nonperi-partum humans, in which intranasal OXT application is related to mostly increased amygdala responses to negatively-valenced and other stimuli in women, whereas decreased amygdala responses are found in men (151,175–180). However, nulliparous and postpartum women differ in how OXT affects their amygdala responses to negative pictures, with nulliparous women displaying greater sensitivity to the attenuation of amygdala responses by OXT, bringing them down to levels comparable to those observed in a group of post-partum breastfeeding women (181). These differences are presumably a result of the differences between these groups in endogenous OXT levels, which are elevated in postpartum women (182). Hence, peripartum changes in depressive symptoms may be a consequence of changes in the release of and sensitivity to OXT, driven by changes in the hormonal milieu during the peripartum period (11).

Laboratory rodents

Additional insight into the neurobiological mechanisms underlying OXT-related mood adaptations during the peripartum period in humans can also be garnered from rodent studies. There is growing evidence, at least in male rats and mice, for antidepressant-like effects of acute or repeated systemic administration of OXT in the forced swim test (183,184), in the learned helpless test (185) and the tail suspension test (157). Antidepressant-like effects of OXT have also been seen after intracerebral administration in male mice (157). These results suggest that systemic or central administration of OXT has antidepressant-like properties (186). However, these potential antidepressant-like effects of OXT could not be confirmed in a psychopathologic animal model (male or female rats selectively bred for high innate anxiety and comorbid depression-related behaviour), as neither acute nor chronic i.c.v. OXT infusion their immobility in the forced swim test (186).

To our knowledge, there is no direct evidence for the effects of high OXT availability on depression-related behaviour in pregnant or lactating rodents. However, there are several symptoms of depression that are likely to be the result of a dysregulated OXT system. For example, an important symptom of depression in humans is impaired social interaction, especially with the child, and general social withdrawal. OXT has been shown to promote naturally occurring social preference, as well as maternal behaviour (187-189). Therefore, we suggest that imbalanced OXT system activity may contribute to social withdrawal and the lack of strong mother-infant bonding in postpartum depression. Furthermore, OXT itself was shown to have rewarding properties, given that both centrally applied OXT as well as drugs of abuse increase dopamine release in the nucleus accumbens as part of the brain's reward circuitry (190,191). Interestingly, in lactating dams, suckling increases the functional magnetic resonance imaging activity in regions of the reward circuitry, including the accumbens-prefrontal cortical pathway (11,192). Because this activation was prevented by pre-administration of an OXTR antagonist, it may be that suckling is rewarding for the mother via central OXT release, encouraging her to continue engaging in nursing behaviour. This finding is also consistent with the observation that, during early lactation, dams find pups more rewarding than cocaine (193). Taken together, elevated OXT either via exogenous application or endogenous release in response to specific stimuli results in a positive hedonic state. Consequently, impaired OXT system activation may underlie the disrupted mother-infant bonding characteristic of postpartum depression as a result of the lack of stimulation of the reward circuitry during mother-infant interactions.

OXT likely modulates maternal depression by regulating the CRH system. As noted above, the depression-like behaviours and poor mothering of postpartum mice with null mutation for the delta subunit of the GABA receptor (Gabrd^-/- mice) are associated with high HPA axis responsiveness, and their phenotype can be rescued by CRH receptor antagonism (J. Maguire and I. Mody, unpublished data). There are similar interactions between OXT and components of the stress system that can influence depression-like behaviours in peripartum laboratory rats. OXT is a robust inhibitor of the HPA axis (52), particularly the brain CRH system (165). Given that CRH triggers depression-like symptoms in nulliparous rodents, and is a putative causal factor for major depression (194,195), up-regulation of neuronal OXT (and prolactin) (196) activity during the peripartum period likely dampens the CRH system and its depressive actions. Interestingly, chronic stress during pregnancy can prevent this peri-partum adaptation in the OXT system. In a chronic psychosocial stress paradigm for pregnant rats that combined restraint stress (twice daily for 1 h) and overcrowding with unknown conspecifics between pregnancy days 4–16, stressed dams did not show the rise in OXT mRNA expression within the PVN that is typical of the peri-partum period (197). Moreover, pregnancy stress prevented the anxiolytic effect of motherhood and resulted in an abnormally high frequency of arched-back nursing by the stressed dams (197). These effects of pregnancy stress on emotional responses and maternal behaviours may not both be mediated by stress-induced glucocorticoid release because, although chronic administration of corticosterone to pregnant or postpartum rats does increase postpartum depression-like behaviours, it reduces maternal behaviour (198,199). In humans, both impaired and increased maternal care and infant attachment have been reported after chronic stress. In this context, it would be also of interest to study the effects of pregnancy stress on other relevant neurochemicals systems, such as vasopressin, which not only promotes maternal care and aggression toward potentially harmful intruders to the nest, but also exerts anxiogenic effects postpartum (200).

Moreover, the dramatic physiological (in particular neuroendocrine) changes occurring during the peripartum period also affect hippocampal plasticity. Specifically, the continuous high levels of circulating glucocorticoids are responsible for reduced cell proliferation, neurogenesis and dendritic architecture during the postpartum period (18,201,202). Reduced hippocampal neurogenesis has often been related to major depression and other stress-related psychopathologies (203,204) and may contribute to mothers' increased risk of developing psychiatric disorders, including postpartum depression. However, it has been found that stress during pregnancy reverses the postpartum reduction in hippocampal plasticity (202). In this context, it is also of interest to note that OXT contributes to hippocampal cell proliferation and neurogenesis (205), but to what extent this neuropeptide is involved in these hippocampal processes during the peripartum period remains to be determined.

Interactions between central GABA and OXT systems

The similar positive consequences of elevated GABA and OXT on peripartum anxiety and depression might suggest redundancy between these neurochemical systems to ensure an optimal maternal emotional state. One can instead conjecture that such similar consequences suggest functional interactions (if not interdependency) between these systems. The literature on interactions between central GABA and OXT systems is quite small, although interactions between them are already known to be critical for other peripartum adaptations of the brain. A particularly salient example involves the large population of OXT-synthesising cells of the hypothalamic SON. At the end of pregnancy, there is a decrease in the ratio of α₁ : α₂ GABA_AR subunit expression on SON OXT cells, which alleviates neurosteroid-potentiated GABAergic inhibition and permits increased activity of these OXT cells necessary for the bolus release of OXT characteristic of parturition and milk letdown (206). Dramatic remodelling of glial morphology and neurone-to-neurone communication among OXT-synthesising cells occurs peri-partum (207), which results in elevated heterosynaptic inhibition of local GABA transmission (208). This increased excitability of OXTergic cells facilitates synchronisation of cell firing during times of high peripheral and central OXT release. In turn, increased local OXT release within the SON rapidly elicits the formation of new GABAergic synaptic contacts and up-regulates GABAergic inhibitory input onto these OXT cells, which may be necessary for their quiescence during times of low OXT release and characteristic bursts of firing at other times (209,210).

Outside the SON and presumably in brains of mostly male laboratory rodents, it has been observed that OXT increases GABA release and GABA_AR-mediated inhibition in brain sites as divergent in function as the lateral hypothalamus (211), hippocampus (212,213), cortex (214) and hypoglossal nucleus (215). Probably relevant to maternal anxiety and depression is the finding that stimulating endogenous OXT release or otherwise activating OXT receptors in the central amygdala of ovariectomised or cycling female rats suppresses the display of emotion-related behaviours (160,161). This results from OXT-sensitive GABAergic cells in the lateral central amygdala inhibiting neural firing in the medial central amygdala, an effect that can be blocked by the GABA_A receptor antagonist bicuculline (216,217). Some other central effects of OXT can also be prevented by bicuculline (218,219). Furthermore, OXT synergises with benzodiazepines to inhibit activity in the medial central amygdala (220) and these cells have been found to project to the cPAGv (221). If such findings extend to peripartum females, which would not be surprising because their GABA and OXT systems are already up-regulated and highly sensitive compared to males or nulliparous females, it would demonstrate an important limbic-midbrain pathway through which GABA and OXT could interact to affect mother's emotional and mood-related behaviours.

Conclusions

Appropriately balanced adaptations of the neurochemical systems regulating peripartum anxiety- and depression-like behaviours are essential for overall maternal well-being and caregiving abilities, and thus the normal development (if not survival) of their offspring. It is clear from the review above that, although adaptations in both the GABA and OXT systems are reasonably well-studied for roles in regulating the anxiety and depressive behaviours of peri-partum rodents, knowledge of peripartum changes in the central GABA system in women and its influence on their peripartum anxiety and depression is almost completely lacking. Furthermore, the work on OXT in peripartum rodents focuses only on anxiety whereas in peripartum women, it focuses more on depression. Filling these gaps will tremendously enhance the translational value of the animal research and broaden our perspective on which systems would be best targeted to improve both mental health concerns of peripartum women. Nonetheless, one can conclude that high cerebral GABA and OXT signalling are likely crucial for optimising maternal emotions and mood during the peripartum period, and that these neurochemicals may depend on each other for their success. Interactions between GABA and OXT refine neural network activity by altering signal-to-noise ratios (213). If the same is true for brain sites where OXT and GABA_A receptor activity produce similar effects on anxiety- and depressive-like behaviours in postpartum rats (which for anxiety includes the cPAGv) (166,121), OXT released in response to interactions with offspring could alter already elevated local GABAergic activity to refine how output neurones respond to emotion-and mood-relevant stimuli. Of course, GABA and OXT are not the only neurochemical systems involved in this important function. As noted above, there are some interesting interactions between GABA and OXT with the CRH system. Prolactin is also very well-known for regulating peripartum emotional and caregiving behaviours (142,222,223) and interacts with both GABA and OXT (224,225). Thus, dysregulation of the normally finely-tuned adaptations of the maternal brain GABA, OXT, CRH and prolactin systems likely underlies the high incidence of emotional and mood disorders during the peripartum period, with adverse consequences for maternal caregiving behaviours and the development of offspring.