The concept of a link between mood changes and the female reproductive organs is not new. Although Hippocrates had proposed peregrinations of the uterus as the underlying cause of hysterical behaviour in women, it was not until the advent of the twentieth century advances in endocrinology that the link between gonadal hormonal status and mood changes in women was demonstrated unequivocably. The recognition that steroid metabolites of progesterone were neuroactive and could pass through the blood–brain barrier further strengthened the link between female hormones and behaviour. Recent studies of progesterone-linked plasticity of GABAA receptor subunit expression are now beginning to offer an insight into the underlying causes of premenstrual symptoms and related hormone-linked disease states.
Many women experience psychological changes during the luteal phase of their menstrual cycle. The late luteal (premenstrual) phase, when symptoms become most severe, is characterized by declining levels of ovarian progesterone. In female rats, withdrawal from prolonged dosing with progesterone leads to upregulation of α4 and δ subunits of the GABAA receptor in several brain regions. During the oestrous cycle of the rat, the natural fall in progesterone that occurs in late dioestrus is associated with a parallel increase in expression of α4, β1 and δ GABAA receptor subunits in neurones in the periaqueductal grey matter (PAG), suggesting that new receptors of the α4β1δ composition have been formed. Recombinant α4β1δ receptors display a low EC50 for GABA, which is consistent with activation by extracellular levels of GABA. They are also likely to be located extrasynaptically and to carry tonic currents. In the PAG, a region involved in mediating panic-like anxiety, α4, β1 and δ GABAA receptor subunits are located principally on GABAergic interneurones. On-going GABAergic neuronal activity normally limits and controls the excitability of the panic circuitry. During late dioestrus, when expression of α4, β1 and δ subunits on GABAergic interneurones is upregulated, the increase in tonic current would be expected to lead to a reduction in the activity of the GABAergic population. Thus the panic circuitry would become intrinsically more excitable. It is suggested that during the menstrual cycle in women, plasticity of GABAA receptor subunit expression in brain regions such as the PAG, which are involved in mediating anxiety behaviour, may underlie some of the changes in mood that occur during the premenstrual period.
Premenstrual syndrome (PMS) describes a constellation of psychological and somatic symptoms that appear in the latter half of the menstrual cycle in women. They include depression, irritability, anxiety, aggression, avoidance of social activity, breast tenderness and bloating (Steiner, 1997). Typically, the symptoms appear gradually during the luteal phase, worsen during the late luteal phase and have remitted by the end of the menstrual flow. Most women of reproductive age experience one or more somatic or behavioural symptoms, which in some individuals can be sufficiently aversive to interfere significantly with their working, social and family life. The symptoms of a number of disease states are also exacerbated during the premenstrual period. Women who suffer from panic disorder, catamenial epilepsy or irritable bowel syndrome report an increase in the severity of their symptoms during this time (Ensom, 2000). Premenstrual symptoms present a considerable social and economic burden to society that is often overlooked. The cost to employers in the USA alone has been estimated to be US$4392 per woman per year (Borenstein et al. 2005). The recent finding indicating that fluctuations in plasma steroid hormone levels can influence GABAA receptor expression in the brain may be a major contributory factor in the premenstrual syndrome and point the way for the identification of new therapeutic management strategies.
Relationship between premenstrual symptoms and plasma progesterone
The appearance of premenstrual symptoms is conditional on ovulation (Bäckström et al. 2003). Symptoms begin to appear as plasma progesterone levels rise during the luteal phase of the menstrual cycle but become most severe during the 5 days prior to menstruation, as progesterone levels decline (Bäckström et al. 2003). Although cyclical changes occur in secretion of all reproductive hormones, only the rise and fall in progesterone levels during the luteal phase coincides with the appearance of premenstrual symptoms, suggesting that changing progesterone levels may be an important factor in the generation of the premenstrual syndrome.
A behavioural state analogous to PMS in women can also occur in non-human primates (Rapkin et al. 1995). Rodents too may be similarly affected. Increased levels of anxiety and aggression have been detected in rats during dioestrus, when plasma progesterone concentrations rise to reach peak levels similar to those that occur during the luteal phase in women, before falling to almost undetectable levels (McLaughlan et al. 1987; Watanabe et al. 1990; Mora et al. 1996; Olsson et al. 2003). Surprisingly, these findings were not replicated in mice, in which increased levels of anxiety were reported during oestrus rather than dioestrus (Maguire et al. 2005).
Synthesis and neuronal effects of progesterone
Progesterone is formed peripherally by the corpus luteum and adrenal gland and centrally by oligodendrocytes (Hu et al. 1987). The lipophilic nature of the steroid facilitates its passage across membranes so that changes in brain levels parallel those in the plasma (Paul & Purdy, 1992). At the cellular level, progesterone acts by binding with intracellular receptors that act as transcription factors in the regulation of gene expression within the cell (Falkenstein et al. 2000). Progesterone also exerts a rapid, non-genomic action via its metabolite allopregnanolone (ALLO), which acts at membrane-bound receptors to potentiate neuronal responses to GABA at GABAA receptors (Lambert et al. 2001).
Acute systemic administration of progesterone or ALLO has anxiolytic effects (Bitran et al. 1995). Anxiety levels in females might therefore be expected to be highest at the time when plasma levels of neuroactive steroids are low. In fact, the reverse is true. During the menstrual cycle, plasma levels of progesterone and hence ALLO are lowest during the follicular phase, when symptoms are absent (Bäckström et al. 2003). In the rat too, anxiety levels are low during proestrus (Olsson et al. 2003) when plasma progesterone is at its lowest level, prior to the preovulatory surge (Butcher et al. 1974; Watanabe et al. 1990). This apparently paradoxical situation may be resolved by recent evidence, which indicates that falling progesterone levels, rather than a steady high or low concentration of the steroid, can lead to significant changes in the inherent excitability of GABA-containing circuits. These changes could underlie some of the alterations in mental state that occur during the oestrous or menstrual cycles.
Plasticity of GABAA receptor subunit expression
GABAA receptors are pentameric structures surrounding a chloride channel. Each receptor typically comprises two α, two β and one γ or δ subunit, assembled from a pool of at least 18 subunits (Mehta & Ticku, 1999). The subunit composition of a receptor is an important determinant of its functional properties (Maitra & Reynolds, 1999). Receptors containing the δ subunit, for example, are highly sensitive to the modulatory effects of neuroactive steroids (Belleli et al. 2002). The distribution of GABAA receptor subunits within the brain is also region specific (Pirker et al. 2000). Thus the actions of GABA are likely to differ significantly at different receptor subtypes and in different parts of the brain. The level of expression of several of the GABAA receptor subunits appears to be linked to ambient levels of progesterone. In cultured cerebellar granule cells, chronic exposure to progesterone significantly reduced the abundance of mRNAs encoding for α1, α3, α5, γ2L and γ2S subunits but, interestingly, had no effect on α4, β1 or β2 subunit mRNA (Follesa et al. 2000). The lack of effect on α4 subunit expression is perhaps surprising, given the considerable plasticity of this subunit in other brain regions (see below). However, expression of α4 subunits in the granule cell layer of the cerebellum is particularly low (Pirker et al. 2000) and may have been been replaced functionally by α6 subunits.
Of relevance to the menstrual cycle is the observation that changes in the level of progesterone, rather than the absolute ambient concentration, provide a potent stimulus for initiating increases in GABAA receptor subunit expression. In hippocampal tissue, a transient increase in expression of α4 subunit mRNA occurred during the second and third days of starting a long-term dosing regimen with progesterone (Gulinello et al. 2001). More recent work has shown that 48 h exposure to progesterone or its metabolite allopregnanolone resulted in increases in both α4 and δ subunit protein in the hippocampus (Shen et al. 2005). In contrast, other studies have shown that acute decreases in systemic progesterone levels can also act as a potent stimulus for upregulation of GABAA receptor subunit expression. The α4 subunit of the GABAA receptor appears to be particularly susceptible. In female rats, withdrawal from a prolonged period of systemic dosing with progesterone initiated an increase in expression of α4 subunit mRNA in the hippocampus (Smith et al. 1998a,b) and amygdala (Gulinello et al. 2003b) and of α4 subunit protein in the periaqueductal grey matter (PAG; Griffiths & Lovick, 2005a) and hippocampus (Smith et al. 1998a,b). This effect was maximal 24 h after progesterone withdrawal (Smith et al. 1998a). Interestingly, a similar increase in subunit expression could also be precipitated in progesterone-treated rats by administration of a 5α-reductase inhibitor to prevent breakdown of progesterone to its metabolite allopregnanolone (Smith et al. 1998b). Thus plasticity of GABAA receptor subunit expression was initiated when progesterone levels remained high but allopregnanolone levels were falling, indicating that allopregnanolone rather than progesterone was the active agent for initiating upregulation of receptor subunit expression. A transient increase in expression of α4 subunits has also been reported in cultured cerebellar granule neurones and cerebral cortical neurones following progesterone withdrawal (Follesa et al. 2000, 2001). Plasticity of GABAA receptor subunits therefore appears to occur in diverse cell types and in different brain regions. Interestingly, although both short-term exposure to progesterone and withdrawal from prolonged dosing with the steroid initiate plasticity in subunit expression, the onset of the withdrawal effect (< 24 h; Smith et al. 1998a) was much faster than the response to an increase in steroid levels (48 h; Gulinello et al. 2001).
In addition to the α4 subunit, plasticity of expression of other subunits has also been reported during withdrawal from progesterone. Increased expression of δ but not γ2 subunit mRNA was reported in hippocampal tissue (Smith et al. 1998a; Sundstrom-Poromaa et al. 2002), whilst in the PAG, an increase in the number of neurones expressing β1 and δ subunits has been described (Griffiths & Lovick, 2005a). In cultures of cerebral cortex, withdrawal from progesterone produced an increase in α1 as well as α4 subunit mRNA, whilst γ2S levels were reduced (Follesa et al. 2001).
Plasticity of subunit expression during the oestrous cycle
The natural fluctuations in endogenous plasma progesterone levels that occur during the oestrous cycle can also stimulate plasticity of GABAA receptor subunit expression. In the latter half of dioestrus (dioestrus II) in rats, when progesterone levels are falling (Butcher et al. 1974; Watanabe et al. 1990) and vaginal smears are characterized by the presence of disintegrating leucocytes (Marcondes et al. 2002; Brack et al. 2006), increases have been reported in the numbers of neurones within the PAG that show immunoreactivity for α4, β1 and δ subunit protein, although numbers of α1 subunit immunoreactive cells remained unchanged (Lovick et al. 2005; Griffiths & Lovick, 2005b). These changes were both qualitatively and quantitatively similar to those reported after withdrawal from an exogenous progesterone dosing regimen (Griffiths & Lovick, 2005a). Interestingly, no changes in subunit expression were seen in the early part of dioestrus (dioestrus I), when progesterone levels are rising, or following the brief preovulatory surge in progesterone prior to oestrus (Butcher et al. 1974; Watanabe et al. 1990). However, increases in GABAA receptor subunit expression in response to high levels of dosing with exogenous progesterone were not seen until plasma progesterone levels had remained elevated for 48 h (Gulinello et al. 2001). In rats, the preovulatory surge lasts only 6–8 h and during dioestrus I, plasma progesterone levels rise for only 24 h before falling (Butcher et al. 1974; Watanabe et al. 1990). Thus it is unlikely that there would be time for any significant change in subunit expression to occur.
In mice, an increase in expression of δ subunit protein has been reported in the hippocampus during late dioestrus, although no change in expression of α4 subunits was detected and γ2 subunit protein was downregulated (Maguire et al. 2005). However, the unusually long 7 day oestrous cycle of the mice used and the large circadian variation in brain levels of ALLO that occur in the mouse compared to the rat (Park & Ramirez, 1987; Corpéchot et al. 1997) make it difficult to make a direct comparison with data obtained from rats.
Functional consequences of changes in GABAA receptor subunit expression
Both experimentally induced and naturally occurring decreases in systemic progesterone levels in rats are associated with an increase in anxiety or aggression levels (Mora et al. 1996; Smith et al. 1998a; Gulinello et al. 2003a,b; Olsson et al. 2003). At the cellular level, differences in the pharmacological properties of GABA modulatory agents emerge after withdrawal from progesterone. The modulatory effects of benzodiazepines are reduced and the benzodiazepine antagonist flumazenil evokes agonist-like effects (Costa et al. 1995; Smith et al. 1998a,b; Follesa et al. 2000; Gulinello et al. 2002). These characteristics are consistent with the appearence of new receptors with the α4βγ2 conformation (Wafford et al. 1996). Other studies suggest that new receptors of the α4βδ type are expressed. After withdrawal from progesterone, the maximum current generated in CA1 pyramidal neurones by the GABA agonist gaboxadol increased, becoming even larger than currents gated by saturating concentrations of GABA (Sundstrom-Poromaa et al. 2002; Gulinello et al. 2003a). This effect of gaboxadol is a defining characteristic of the α4βδ receptor isoform of the GABAA receptor (Brown et al. 2002). The results from immunohistochemical studies also point to expression of new GABAA receptors of the α4β1δ subtype when progesterone levels fall. Parallel increases in the numbers of neurones immunoreactive for α4, β1 and δ subunits occurred in the PAG (Griffiths & Lovick, 2005a,b; Lovick et al. 2005). Although colocalization of different subunits within individual cells was not investigated in these studies, it is likely that at least some of the new subunits were coexpressed within the same cells and assembled into functional receptors.
Receptors that contain α4 and δ subunits have been shown to be located at extrasynaptic sites on hippocampal and cerebellar neurones (Nusser et al. 1998). Their activation by GABA spillover (Wei et al. 2003) is consistent with a role in mediating the tonic level of inhibition on GABAergic neurones (Mody, 2005). Recombinant α4β1δ GABAA receptors expressed in oocytes from Xenopus laevis show exceptional sensitivity to GABA, with an EC50 that is within the concentration range for GABA within the extracellular fluid (Lerma et al. 1986; Lovick et al. 2005). An increase in expression of extrasynaptic α4β1δ GABAA receptors would be associated with an increase in tonic current carried by the cells. However, the functional consequences of such changes could differ markedly depending on the neuronal phenotype in which the new receptors are expressed. Within the PAG, α4, β1 and δ GABAA receptor subunits are expressed mainly by GABAergic neurones. In late dioestrus (dioestrus II). When subunit expression increases, more than 90% of the GABAergic neurones became immunopositive for these subunits (Griffiths & Lovick, 2005b). On-going activity in the GABAergic population normally restricts the excitability of output neurones in the PAG (Schenberg et al. 1983). An increase in tonic current carried by the GABAergic interneurones would be expected to depress their activity, leading to a reduction in the level of GABAergic inhibition on the principal (output) neurones in the PAG. As a consequence, the overall excitability of the circuitry should go up (Fig. 1). In support of this hypothesis, electrophysiological studies in vivo indicate that GABAergic tone on output neurones in the PAG is indeed reduced during late dioestrus (Brack et al. 2004).
Other studies in hippocampus have proposed that the increase in αβδ GABAA receptor expression seen following 48 h exposure of rats to progesterone represents a subunit switch rather than an increase in the total GABAA receptor population (Shen et al. 2005). In the supraoptic nucleus too, a change in the ratio of α1 to α2 subunits was reported to occur around the time of parturition (Brussaard et al. 1997). With respect to oestrous cycle-related changes of GABAA receptor subunit expression in the PAG, further studies at the cellular level will be necessary to determine whether neurosteroid-evoked subunit switching can occur in this structure.
The PAG contains neural circuitry that is involved in the generation of panic-like anxiety. Panic attacks are episodes of intense irrational fear accompanied by pronounced autonomic disturbances (DSMIV, 1984). In women with panic disorder, susceptibility to panic increases during the premenstrual period (Ensom, 2000). In line with these clinical findings, recent studies in female rats have indicated that autonomic responsiveness to challenge with a systemically administered panicogenic agent is increased during late dioestrus (Brack et al. 2006). This effect may be linked to a lowering of the threshold for activation of the panic circuitry in the PAG, as a consequence of the decrease in on-going levels of GABAergic inhibition.
In summary, withdrawal from prolonged dosing with exogenous progesterone appears to reproduce the behavioural and cellular changes that occur naturally during the oestrous cycle. The physiological significance is not clear but presumably represents a response of steroid-sensitive GABAA receptors to changing steroid levels. Indeed, the presence of new steroid-sensitive receptors located at extrasynaptic sites should maintain responsiveness to neuroactive steroids when the local concentration falls. However, in those brain regions where the new GABAA receptors are located on GABAergic interneurones that show on-going activity, an enhanced responsiveness to the GABA present in the extracellular space would lead to a decrease in local GABAergic activity and an overall increase in neuronal excitability. In regions such as the PAG, amygdala and hippocampus, such changes could make a major contribution to the development of the psychological changes that characterize the premenstrual period in women.
Financial support from the Wellcome Trust and the British Heart Foundation is gratefully acknowledged.