The glucocorticoid contribution to obesity

Abstract

Obesity is fast becoming the scourge of our time. It is one of the biggest causes of death and disease in the industrialized world, and affects as many as 32% of adults and 17% of children in the USA, considered one of the world's fattest nations. It can also cost countries billions of dollars per annum in direct and indirect care, latest estimates putting the USA bill for obesity-related costs at $147 billion in 2008. It is becoming clear that the pathophysiology of obesity is vastly more complicated than the simple equation of energy in minus energy out. A combination of genetics, sex, perinatal environment and life-style factors can influence diet and energy metabolism. In this regard, psychological stress can have significant long-term impact upon the propensity to gain and maintain weight. In this review, we will discuss the ability of psychological stress and ultimately glucocorticoids (GCs) to alter appetite regulation and metabolism. We will specifically focus on (i) GC regulation of appetite and adiposity, (ii) the apparent sexual dimorphism in stress effects on obesity and (iii) the ability of early life stress to programme obesity in the long term.

Keywords::

• Adipose
• HPA axis
• metabolism
• perinatal
• sex
• stress

Glucocorticoid regulation of appetite and adiposity

Psychological stress is a significant risk factor for obesity

Severe psychological trauma can clearly contribute significantly to obesity in later life. Studies have shown that post-traumatic stress disorder, a debilitating anxiety disorder that develops after exposure to a single or multiple extremely traumatic events, is a significant predictor of obesity (Vieweg et al. 2007; Kozaric-Kovacic et al. 2009; Perkonigg et al. 2009). Similarly, exposure to war-related events during childhood has been associated with a higher body mass index (BMI) 10 years later (Llabre and Hadi 2009). More mild psychological stress is also closely associated with excess weight gain. For instance, female macaques exposed to the chronically stressful social subordination or social reorganization have increased visceral fat deposition and adverse metabolic characteristics such as hyperglycaemia (Shively and Clarkson 1988; Jayo et al. 1993; Shively 1998; Shively et al. 2009). Numerous studies in humans have established a link between an environment of stress and excess body weight. Social status (defined by employment grade) is inversely related to central adiposity and the metabolic syndrome in men and women (Brunner et al. 1997). Perceived psychosocial stress in multiple domains, such as job-related demands and difficulty paying bills, has also been linked with weight gain over an extended period in adult men and women (Block et al. 2009; Fowler-Brown et al. 2009). In addition to, job stress being associated with obesity, it has also been associated with negative eating behaviours such as eating to satiety and to curb irritability (Nishitani et al. 2009). Psychological stress also contributes to obesity in children. Children from families reporting high levels of stress across a number of domains are significantly more likely to be obese than those from “low-stress” families (Koch et al. 2008; Moens et al. 2009), with serious events such as divorce, parental unemployment or death of a family member having particular impact.

While a background of psychological stress can confer a propensity to develop obesity, the condition itself is linked to exacerbated responses to stress. For instance, the cortisol response to public speaking stress is enhanced in obese women, as is heart rate and diastolic blood pressure (Benson et al. 2009). Obese people are also more likely to suffer from stress, anxiety and depression (Doyle et al. 2007; Scott et al. 2008; Abiles et al. 2010). Whether this phenomenon is responsible for or symptomatic of obesity cannot be deduced from these studies. However, a more direct link between psychological stress, a hyperactive hypothalamic–pituitary–adrenal (HPA) axis and obesity in humans has recently been established. Thus, women who developed obesity following a stressful event were found to have gained weight significantly faster than those who developed obesity without a prior stressful event, and this was associated with higher basal cortisol levels (Vicennati et al. 2009). It is, therefore, emerging that dysfunctional regulation of the stress response, particularly of the HPA axis, after severe or prolonged stress is significantly involved in the pathogenesis of obesity.

Glucocorticoid regulation of appetite

The HPA axis is the key endocrine axis responsible for mediating the body's response to stress. Other components of the body's response to stress such as sympathomedullary activation undoubtedly also influence obesity, but are beyond the scope of this review. Briefly, when an organism encounters a stressor, medial parvocellular cells in the paraventricular nucleus of the hypothalamus (PVN) are activated, leading to the release of corticotropin-releasing hormone (CRH) and arginine vasopressin (AVP) at the median eminence into the hypothalamo-hypophysial portal blood vessel system. Adrenocorticotropic hormone is then released from the anterior pituitary into the systemic circulation, and acts at the adrenal cortex to stimulate glucocorticoid (GC: i.e. cortisol in humans, corticosterone in rodents) synthesis and release. GCs have a number of functions to assist in combating the stressor, including to feed back and prevent further HPA axis activation. This axis has been reviewed extensively elsewhere (Sapolsky et al. 2000; Papadimitriou and Priftis 2009). There are several mechanisms by which dysregulation of the HPA axis and of GC function may affect energy balance and potentially contribute to obesity.

GCs are, for instance, important in appetite regulation (Figure 1). Acutely after stress, appetite is suppressed. This anorexia occurs centrally via a CRH-mediated mechanism, whereby CRH (Heinrichs and Richard 1999; Richard et al. 2002) and other CRH-like molecules such as urocortin (Weninger et al. 1999; Richard et al. 2002) act in the brain to suppress appetite, probably via an inhibition of neuropeptide Y (NPY)-stimulated food intake (Heinrichs et al. 1993; Currie 2003). This CRH-induced anorexia is likely to involve a number of brain regions including the PVN, perifornical and ventromedial regions of the hypothalamus, lateral septum and parabrachial nucleus (Richard et al. 2002), but recent evidence also indicates an important role for the indirect action of CRH on the dorsal portion of the anterior bed nucleus of the stria terminalis (Ciccocioppo et al. 2003). Although stress may acutely suppress appetite in this way, GCs are released in the hours to days following a stressful event and these stimulate feeding (Santana et al. 1995; Dallman et al. 2004). Indeed, George et al. (2010) were recently able to show that elevated cortisol concentrations are correlated with food intake as little as 1 h after a peripheral injection of CRH in humans, and that the amount of food consumed is directly related to the size of the peak cortisol response (George et al. 2010).

Figure 1.  An overview of the concepts discussed for GC regulation of appetite and adiposity. GCs stimulate food intake via direct action on orexigenic NPY/AGRP neurons in the ARC, as well by stimulating nucleus accumbens (NAcc)/VTA pathways that reinforce the rewarding nature of food. GC may also stimulate food intake by enhancing the appetite- and reward-stimulatory effects of ghrelin. In addition, GC can prevent the actions of satiety hormones leptin and insulin; reducing the sensitivity of the brain to these hormones and contributing to leptin and insulin resistance. GC also contributes to enhancing visceral fat WAT. They activate HSL, enhancing lipolysis and LPL, promoting fat storage. Because there is a greater density of GRs in visceral fat, as well as greater LPL activity, fat storage may be preferentially promoted in this tissue. See text for details.

In an acutely stressful situation, this is probably a highly adaptive phenomenon. The stimulation of feeding would serve to replace lost energy and to facilitate a “stocking up” in anticipation of the next stressor. However, if the stressor continues too long or there are multiple sequential stressors, GC levels can become chronically elevated, leading in humans to a chronically stimulated appetite, increased feeding and consequently to obesity (De Vriendt et al. 2009).

GCs stimulate food intake via a series of complex interactions on various appetite-regulating targets. They stimulate the actions of orexigenic peptides NPY and agouti-related peptide (AGRP) (Savontaus et al. 2002; Konno et al. 2008), up-regulating gene expression of NPY and AGRP in the arcuate nucleus (ARC) by increasing AMP-activated protein kinase signalling (Shimizu et al. 2008). GCs also influence the function of anorexigenic leptin. Although GCs stimulate leptin release from adipose tissue, which would normally lead to appetite suppression, they also reduce the sensitivity of the brain to leptin, contributing to leptin resistance (Zakrzewska et al. 1997, 1999; Jequier 2002).

In addition to leptin, GCs also stimulate insulin secretion from the pancreas (Strack et al. 1995), which has complex effects on appetite. Insulin usually acts at the hypothalamus to reduce food intake and at the ventral tegmental area (VTA) to reduce the dopaminergic neuron-mediated rewarding nature of food (Figlewicz et al. 2008). However, when food choice is available, it enhances the preference for “comfort foods”, particularly those high in fat and sucrose (la Fleur et al. 2004; Warne et al. 2006, 2009). In addition, chronically elevated GC concentrations can contribute to insulin resistance, reducing its ability to inhibit feeding-stimulatory pathways in the brain, as is the case with leptin (Asensio et al. 2004).

An emerging, but still poorly elucidated mechanism by which GC may further influence appetite regulation during stress is via ghrelin. Ghrelin is a gastrointestinal peptide usually associated with feeding regulation, acting to signal initiation of a meal (Hosoda et al. 2006). Circulating levels of ghrelin are increased during stress (Kristenssson et al. 2006) and activate CRH and AVP release to potentiate the stress response, probably doing so either directly by acting at the PVN or indirectly by activating brainstem noradrenergic inputs to the PVN (Cummings 2006; Kawakami et al. 2008). It is, therefore, conceivable that chronic or severe stress resulting in prolonged elevated GC secretion could lead to chronically up-regulated ghrelin signalling, culminating in increased food intake.

Stress and GCs tend to stimulate appetite specifically for high-energy foods, with insulin having a key role in this regard (la Fleur et al. 2004; Warne et al. 2006, 2009; Dallman 2010). It has been established that chronically stressed animals prefer calorically dense foods and that these can actually ameliorate HPA axis responses to further stress (Pecoraro et al. 2004; Foster et al. 2009). Thus, rats exposed to chronic restraint stress voluntarily ingested more “comfort food” (lard and sucrose) than non-restrained rats, and HPA axis responses to the restraint were reduced in rats that had access to these high-calorie foods (Pecoraro et al. 2004). Given the strongly rewarding nature of this cycle, it is not surprising that this preference for highly palatable foods during stress is mediated through some of the same pathways that mediate responses to drugs of addiction (Nieuwenhuizen and Rutters 2008; Coccurello et al. 2009). Dopaminergic, opioid and glutamatergic transmission within the nucleus accumbens, for instance, appear to be crucial for mediating reward-associated information, whether it be regarding drugs (Di Chiara and Imperato 1988; Koob and Le Moal 2001; Ito et al. 2004), sex (Balfour et al. 2004) or palatable food (Berridge 2009). This reward circuitry has long been known to be activated by eating (Bassareo and Di Chiara 1997, 1999) and, interestingly, the propensity to ingest highly palatable food has been correlated with preference for drugs of abuse (Gosnell 2000). For instance, ghrelin, which usually serves to stimulate food intake (Hosoda et al. 2006) and subserves the rewarding nature of food (Egecioglu et al. 2010), is also essential for the rewarding effects of alcohol and psychostimulant drugs (Jerlhag et al. 2009, 2010).

Glucocorticoid regulation of adiposity

Stress can clearly contribute to the propensity to obesity via GC-mediated up-regulation of appetite, particularly for fatty foods. However, GCs also play a role in the regulation of lipid homeostasis (fat mobilization from adipose tissue and fat deposition and storage; Figure 1).

Excess GC has been particularly associated, in primate models (Shively and Clarkson 1988; Jayo et al. 1993; Shively 1998; Shively et al. 2009) and in humans, with visceral fat accumulation (Marin et al. 1992b; Rosmond et al. 1998; Epel et al. 2000), a phenotype that is strongly linked to an increased risk for cardiovascular disease (Grassi et al. 2004; Hamdy et al. 2006; Mathieu et al. 2008). GC receptors (GRs) are present in visceral adipose tissue in greater density than in other adipose depots (Rebuffe-Scrive et al. 1990) and may serve to relatively enhance the effects of GC specifically in this tissue (Bjorntorp 2001). Additionally, type 1 11β-hydroxysteroid dehydrogenase (11β-HSD1), an enzyme which generates active GC from inactive GC metabolites within tissues, has been shown in mice to be linked to visceral adiposity, with over-expression of the 11β-HSD1 gene leading specifically to visceral fat accumulation (Masuzaki et al. 2001). Conversely, transgenic mice expressing specifically in adipocytes type 2 11β-HSD (11β-HSD2), which inactivates GC, protects against this phenotype (Kershaw et al. 2005). In humans, increased visceral fat 11β-HSD1 oxoreductase activity is also associated with increased accumulation in this depot, as well as adipocyte hypertrophy, increased lipolysis (the release of non-esterified fatty acids from adipocytes) and increased lipoprotein lipase (LPL) activity in this tissue (Veilleux et al. 2009). It appears that elevated 11β-HSD1 may be a cause rather than a consequence of obesity, as inhibiting the enzyme can improve metabolic parameters and reduce weight (Tiwari 2010). Furthermore, disruption of 11β-HSD1 leads to protection from high-fat diet-associated weight gain and a fat distribution profile away from the visceral depot (Morton et al. 2004). These studies clearly demonstrate a relationship between GC activity and adiposity; however, the mechanisms underpinning this relationship are not clearly understood.

The action of GC on fatty acid metabolism and storage is complex and GCs have a variety of direct and permissive effects (Macfarlane et al. 2008; Xu et al. 2009). It is generally seen that GC acutely (hours) enhance lipolysis. That is, short-term infusion of GC in vivo stimulates the release of non-esterified fatty acids from adipocytes (Divertie et al. 1991) through the activation of hormone-sensitive lipase (HSL) (Slavin et al. 1994), a key enzyme responsible for enhancing fatty acid mobilization (Figure 1). This increase in circulating free fatty acids may restrict glucose utilization and encourage insulin resistance (Arner 2002). GCs have also been shown to enhance adipose LPL activity, which should promote fat storage (Bjorntorp 1996, 2001). Thus, it appears that GC may enhance lipid metabolism through its effects on both the turnover and uptake of fatty acids in adipose tissue. Since LPL activity is greater in visceral adipose tissue than in other adipose depots (Marin et al. 1992a), GCs may, in this way, contribute to a redistribution of fat such that it accumulates preferentially in visceral depots (Figure 1). Insulin appears to be essential for this process as it is able to markedly potentiate the GC affect on LPL (Ashby and Robinson 1980).

GCs also play a critical role in the differentiation of pre-adipocytes into mature adipocytes (Hauner et al. 1987; Tomlinson and Stewart 2002), a role which is again insulin related (Hauner et al. 1989; Gregoire et al. 1992; Suryawan et al. 1997), potentially further accentuating the accumulation of fat. They are also known to up-regulate the NPY Y2 receptor in abdominal fat, leading to stimulation of proliferation and differentiation of adipocytes, again culminating in enhanced fat accumulation (Kuo et al. 2007, 2008). Although the distinct molecular mechanisms by which GCs may affect adiposity remain to be determined, it is clear that active GCs are strong modulators of adipose tissue metabolism and distribution in rodents and humans, contributing to enhanced fat deposition particularly in the high-risk visceral adipose depot.

Sexual dimorphism in stress effects on obesity

Sexual dimorphism in stress effects on white adipose tissue

Chronic stress and the increased GC level that results can thus predispose an individual to obesity or exacerbate an already obese phenotype. However, there are indications that there is some sexual dimorphism in these effects.

Obesity affects men and women very differently. Women are more likely to develop obesity than men (Lovejoy and Sainsbury 2009), and are more likely to develop metabolic syndrome if they do become obese (Razzouk and Muntner 2009). They are also less likely to lose weight with calorie restriction or exercise regimes than men (Meijer et al. 1991; Westerterp and Goran 1997; Sartorio et al. 2005). Men, by contrast, have a higher risk of accumulating fat in the visceral depot, leaving them more vulnerable to associated complications such as cardiovascular disease (Bjorntorp 1996).

Several studies have suggested that women may be more sensitive to the obesity-promoting effects of psychological stress. For instance, women who are moderately overweight (BMI = 25.0–29.9) are more likely than normal weight women to report having experienced stressful life events. In men, obesity and extreme obesity are associated with more stressful life events, but a moderately overweight phenotype is not (Barry et al. 2008). Similarly, a 13-year follow-up study of black American adults revealed high levels of perceived stress is correlated with greater percentage increases in BMI in women, but not in men (Fowler-Brown et al. 2009). Our own studies in rats have suggested that females, but not males, that became obese due to overfeeding as neonates have enhanced HPA axis responses to psychological stress (Spencer and Tilbrook 2009).

BMI may, however, be a less useful predictor of the health outcomes associated with obesity than adipose distribution. Obese men and post-menopausal women have a significantly greater risk of developing obesity-associated diseases such as type II diabetes mellitus and cardiovascular disease than pre-menopausal women owing to a profile of enhanced visceral fat (Bjorntorp 1996). This profile in particular may be associated with differential effects of GC at the adipose tissue level (Tomlinson et al. 2004).

As mentioned, GCs may contribute to a predominantly visceral fat distribution. In women, lipolysis is stimulated more effectively in subcutaneous than in visceral adipocytes, whereas there are no differences between the depots in men. The synthetic GC, dexamethasone, is also able to increase the ability of a synthetic cAMP analogue to stimulate lipolysis in subcutaneous adipose tissue in women, with a similar tendency in visceral adipocytes, but not in men (Lundgren et al. 2008). Conceivably, the major action of GC in females is at subcutaneous rather than visceral fat to stimulate lipolysis, whereas in men GC would act at subcutaneous and visceral fat equally and promote visceral fat distribution via greater activation of LPL in this tissue.

A key player in this adipose depot specific action of GC is also 11β-HSD1. As mentioned, higher 11β-HSD1 activity in visceral vs. subcutaneous adipose tissue is associated with preferential visceral fat accumulation and associated metabolic alterations (Veilleux et al. 2009). Rat and human studies demonstrate significantly higher activity of 11β-HSD1 in males compared with females (Albiston et al. 1995; Weaver et al. 1998; Toogood et al. 2000), which may contribute to increased visceral adiposity in males in the face of chronic stress.

Sexual dimorphism in stress effects on brown adipose tissue

In addition to the action of GC at white adipose tissue (WAT), other sexually dimorphic peripheral effects of GC may contribute to sexually dimorphic effects of stress on obesity, for instance at the level of brown adipose tissue (BAT). BAT is responsible for dissipating energy as heat primarily via an uncoupling protein (UCP)1-mediated uncoupling of oxidative phosphorylation (Cannon and Nedergaard 2004). Impairments in BAT function can, therefore, result in reduced thermogenesis and thus increased energy storage leading to obesity. In this regard, BAT dysfunction is certainly a consistent factor across many models of obesity. Ablation of BAT will result in obesity even in the absence of hyperphagia (Lowell et al. 1993) and various genetic models of obesity, such as the ob/ob and db/db mice, are also found to have reduced levels of BAT thermogenesis (Commins et al. 1999; Masaki et al. 2000).

There is some evidence that BAT functions differently in males and females. In rodents, basal BAT thermogenesis is greater in ad libitum fed females than in males in a thermoneutral environment (Justo et al. 2005; Valle et al. 2007). Female rats are also capable of deactivating BAT thermogenesis to a greater degree than males under calorie-restricted conditions and this is associated with resistance to the loss of lean mass (Valle et al. 2005). In addition, we have recently found that females made obese by overfeeding during the neonatal period maintain an obese phenotype due to a down-regulation of BAT function, whereas males show no such alterations in BAT activity (Stefanidis, Tilbrook and Spencer unpublished observations, 2009–2010).

Until recently, it was thought that adult humans did not possess substantial amounts of active BAT. This dogma has now been dispelled by several conclusive high-profile studies revealing that BAT depots exist in the neck anterior to the thorax, particularly the cervical–supraclavicular region (Nedergaard et al. 2007; Cypess et al. 2009; van Marken Lichtenbelt et al. 2009; Virtanen et al. 2009; Zingaretti et al. 2009). It is noteworthy that these studies have revealed differences in the frequency with which men and women present with functionally active BAT—defined regions of active BAT being present more frequently in women than men (Cypess et al. 2009). These findings lead to the suggestion that the role of BAT in human obesity is also sexually dimorphic, or at least presents a sexually dimorphic target for therapeutic intervention.

With respect to the effects of stress on this tissue, GCs are among the strongest down-regulators of UCP1 expression in BAT (Arvaniti et al. 1998), contributing to a reduction in thermogenesis and a conservation of energy that could, in a chronic setting, promote obesity. The greater predominance of BAT in women and the sexually dimorphic sensitivity of BAT to environmental manipulations suggest that stress may also influence the development and maintenance of obesity in a sexually dimorphic manner via its effects at this tissue.

The role of sex steroids in obesity and the impact of stress

Obesity may also occur by different mechanisms in males and females due to the differential involvement of sex steroids. Sex steroid hormones, sexually dimorphic after puberty, act centrally to regulate feeding as well as peripherally to influence adipose tissue metabolism, and their actions may be stress sensitive. Receptors for oestrogen, progesterone and androgens are all present in WAT and BAT and the sex steroids can act as transcriptional regulators of various metabolic genes. For instance, oestrogen, the predominant sex steroid in females, can exert effects on LPL to alter lipid accumulation, playing an essential role in fat distribution (Shi et al. 2009) and leading to greater inguinal fat deposition in females (Bjorntorp 1996; Pedersen et al. 2004). Thus, ovariectomy in rodents increases food intake and promotes weight gain, and replacing oestradiol is sufficient to normalize this (Asarian and Geary 2002). Similarly, menopause in human females, leading to a reduction in oestrogen levels, is also associated with increased weight gain and a shift towards visceral fat deposition (Svendsen et al. 1995; Tchernof et al. 2000). Levels of progesterone and testosterone are also sexually dimorphic and may contribute to variation in energy balance (Lobo et al. 1993; Mayes and Watson 2004; Saad and Gooren 2009). For instance, testosterone also inhibits LPL activity and adipogenesis (De Pergola 2000; Blouin et al. 2010), and hypotestosteronaemia in human males has been associated with increased adiposity (Mah and Wittert 2010). Sexual dimorphism in levels of sex hormones or their receptors may also contribute to sexual dimorphism in BAT thermogenic capacity. Females have lower levels of BAT androgen receptors and oestrogen receptor α than males, both of which are involved in promoting thermogenesis (Rodriguez-Cuenca et al. 2007). These differences could potentially leave females more vulnerable to changes in concentrations of the ligand.

Stress and chronically elevated GC levels can have pronounced effects on androgen activity. In male rats, the circadian rise in corticosterone concentrations are associated with a dip in testosterone levels, and the normal bimodal circadian pattern of testosterone release can be abolished with the administration of synthetic GC (Waite et al. 2009). Thus, there is an antagonistic relationship between the activation of the HPA axis and that of the HPA axis, with chronic stress-suppressing sex hormone secretion and interfering with reproductive function (Retana-Marquez et al. 2003; Kinsey-Jones et al. 2009). The hypotestosteronaemia that results from an overactive HPA axis or from chronic stress in males may contribute in turn to increased adiposity and an obese phenotype (Mah and Wittert 2009), as may a chronic reduction in circulating oestrogen levels in females with HPA axis activation (Kinsey-Jones et al. 2009).

Stress, obesity and neonatal programming

Prenatal stress programmes a predisposition to obesity

Foetal and neonatal environmental risk factors are increasingly recognized as relevant determinants of adult physiology and may contribute to an obese phenotype in adulthood. It is likely that stress, at critical windows of development, is one of these risk factors.

Adaptations during pregnancy ensure that the foetus is relatively protected against increases in GC. Elevated levels of progesterone and its metabolite, allopregnanolone, in pregnancy ensure that there is an enhancement of the allopregnanolone-mediated inhibition of the noradrenergic input to the PVN from the nucleus of the solitary tract (Brunton et al. 2005, 2009; Brunton and Russell 2010). Thus, pregnant females respond to stressors with less of an increase in GC than do non-pregnant females (Brunton et al. 2005, 2009; Slattery and Neumann 2008). Placental 11β-HSD2, responsible for deactivating GC, also plays a crucial role in ameliorating the impact of maternal GC (Lucassen et al. 2009). However, in cases of severe or chronic maternal stress, excess GC may still affect the foetus.

Studies in humans and in animal models have revealed myriad long-term effects on the offspring of maternal exposure to stress or GC during pregnancy. Stressful events during pregnancy can impact upon foetal brain development leading to alterations in HPA axis (Henry et al. 1994; Rossi-George et al. 2009), anxiety behaviours (Vallee et al. 1997), learning and memory (Lordi et al. 1997; Entringer et al. 2009), sensitivity to drug abuse (Morley-Fletcher et al. 2004; Thomas et al. 2009) and, crucially, obesity. For instance, maternal bereavement immediately before or during pregnancy is associated with a significantly increased risk of excess weight gain in childhood (Li et al. 2010).

One major factor contributing to the aetiology of our current obesity epidemic may, therefore, be excessive exposure to stress or GC at specific time points during prenatal or post-natal development. Prenatal (gestational days 8, 10 and 12) exposure to dexamethasone (a synthetic GC), or to the physical stressor (see (Dayas et al. 2001; Keay and Bandler 2001) for discussion on physical vs. psychological stress) of exposure to the cytokines interleukin 6 or tumour necrosis factor α, was seen to increase adiposity throughout life by up to 40% (Dahlgren et al. 2001). Such effects appear to be somewhat gender specific. Thus, while dexamethasone and these cytokines affected offspring adiposity in both males and females (Dahlgren et al. 2001), doses of lipopolysaccharide (LPS; 0.79 mg/kg) at the same gestational ages led to adiposity and insulin resistance in male but not female offspring (Nilsson et al. 2001).

However, prenatal stress alone may not necessarily be sufficient to induce obesity in the offspring (D'Mello and Liu 2006; Baker et al. 2008), but may still confer an increased susceptibility to obesity (Tamashiro et al. 2009). For instance, the offspring of rats exposed to chronic variable stress during the third week of gestation do not become obese if weaned onto a normal rat-chow diet. However, obesity and impaired glucose tolerance in adulthood are manifested if these offspring are weaned onto high-fat chow (Tamashiro et al. 2009). This susceptibility may be related, in part, to prenatal stress causing intrauterine growth restriction (reflected in lower birth weights (Lesage et al. 2004)) which is then compensated for by catch-up growth if sufficient high-energy food is available after birth (Hales and Barker 1992; Gluckman and Hanson 2004).

Post-natal events also influence development, and post-natal stress can also lead to a predisposition to obesity in adulthood. Early psychological trauma in humans, ranging from sexual abuse to the death of a mother or parental stress, is a significant risk factor for the development of obesity in later life (Koch et al. 2008; D'Argenio et al. 2009). In rodents, the effects of post-natal stress are less clear. It appears that exposure to stress in the early post-natal period may not directly lead to obesity but may confer an enhanced susceptibility. In rodents, separation from the mother is a significant psychological stress to the neonate and has long-lasting physiological effects, including potentially enhancing susceptibility to changes in energy balance and metabolism. Thus, neither separation from the mother as a neonate nor social isolation post weaning alone significantly affect body weight, but social isolation in rats that had previously undergone maternal separation leads to significant, long-lasting increases in food intake and weight gain (Ryu et al. 2009).

An important point highlighted by several of these investigations is that timing is critical. Chronic exposure to dexamethasone in early gestation in the common marmoset does not adversely affect the animal, while the same type of challenge late in gestation leads to early indices of metabolic syndrome such as high fasting plasma glucose and triglyceride levels (Nyirenda et al. 2009). Restraint stress experienced in rats in mid-late gestation produces long-term changes in the body weights of offspring of stress-sensitive dams, whereas early gestation restraint stress does not (Mueller and Bale 2006). Similarly, the physical stressor of an immune challenge with LPS experienced at either post-natal days 3, 7 or 14 does not affect long-term growth or adiposity in the pair-housed rat (Spencer et al. 2007), whereas a similar challenge presented at day 10 does appear to mildly increase adult body weight (Iwasa et al. 2010). Interestingly, it appears that the programming effects of stress may not be confined to the perinatal period. Chronic social stress has also been shown to lead to long-term alterations in HPA axis function and fat distribution if it occurs during adolescence (Schmidt et al. 2009).

Equating developmental stages between species is fraught with controversy. However, it is generally considered that late gestation and the early post-natal period in the rodent are roughly equivalent to the third trimester of pregnancy in the human, particularly with regard to metabolic systems (Grove et al. 2005). Thus, it appears that, for humans, stress encountered during the third trimester may have more lasting impact on the offspring's propensity to develop obesity than stress encountered at other times. Indeed, this appears to be reflected in the number of elaborate adaptive mechanisms that are present at this time to protect the foetus (Brunton et al. 2005, 2009; Slattery and Neumann 2008; Spencer et al. 2008).

Mechanisms for perinatal stress effects on body weight regulation

One mechanism by which prenatal stress (or GC exposure) may result in a long-term susceptibility to obesity is by altering HPA axis function. Prenatal restraint stress leads to altered corticosterone responses to mild stress in the adult rat offspring (Henry et al. 1994; Maccari et al. 1995, 2003; Koehl et al. 1999; Rossi-George et al. 2009) and reduced hippocampal levels of mineralocorticoid receptor and GR (Henry et al. 1994; Maccari et al. 1995). Prenatal stress can interfere with synaptic pruning during brain development in regions that are important for HPA axis control, the prefrontal cortex (Spencer et al. 2005) and hippocampus (Jacobson and Sapolsky 1991). In particular, it has recently been determined that growth-associated protein of 43 kDa (GAP-43), an intracellular protein involved in the establishment and reorganization of synaptic connections during development (Pfenninger et al. 1991; Larsson 2006), is up-regulated in the second week after birth and reduced in adulthood in offspring from restraint-stressed dams (Jutapakdeegul et al. 2010). These findings indicate that prenatal stress can produce lasting changes in the connectivity of these regions, potentially affecting HPA axis sensitivity in the long term. A super-sensitive HPA axis may lead, in turn, to greater susceptibility to the effects of GC on appetite and weight regulation discussed above.

Peripheral alterations in GC function may also come into play. A recent study has determined that 11β-HSD1 may play a crucial role in these programming effects. Thus, chronic prenatal exposure to dexamethasone during late, but not early, gestation led to persistent elevations in 11β-HSD1 mRNA expression and activity in adipose tissue in marmosets. These animals went on to develop early indices of metabolic syndrome (Nyirenda et al. 2009).

The mechanisms by which perinatal stress may cause these changes in HPA axis function are not well understood. It is clear that maternal GCs are able to cross the placenta and influence foetal GC levels and receptor expression (Edwards et al. 1993) as well as levels of 11β-HSD2 (Clifton et al. 2006). Foetal 11β-HSD2 levels are important in programming subsequent stress responses and 11β-HSD2 − / − mice have greater anxiety levels and lower birth weights than wild-type littermates (Holmes et al. 2006). Another potential mechanism by which stress may programme later HPA axis function is by DNA methylation. Meaney and colleagues have shown that offspring that were nursed with low intensity (low licking and grooming) develop to be more anxious with exacerbated HPA axis activation in response to stress (Caldji et al. 1998; Champagne and Meaney 2001). These animals probably have less of a grooming-associated serotonin-mediated increase in nerve growth factor inducible factor A (NGFI-A) expression. NGFI-A binds to its recognition sequence on the GR resulting in increased histone acetylation, which facilitates demethylation of the GR promoter and thus receptor activity. In the case of the low maternal attention groups, this activity would be reduced. The opposite occurs with high maternal attention (Fish et al. 2004; Meaney and Szyf 2005). Although there is no evidence that these animals develop obesity, such alterations to the GR could predispose the animal to it given high-fat or low-exercise conditions.

Clearly, factors affecting the development of the HPA axis can have long-lasting effects on energy balance within the body. These findings highlight just how important is a stress-free early life environment to long-term body weight control in the offspring.

Perinatal nutritional stress

Inadequate or inappropriate nutrition is a unique type of stressor that has particular long-term impact on the HPA axis and body weight regulation. In the winter of 1944–1945, particularly harsh weather and the final battles and aftermath of Second World War meant that the people of Holland experienced a severe famine. While devastating to the population, several insightful findings on the importance of nutrition have emerged as a result of this disaster. These studies have shown that inadequate nutrition in the first trimester of gestation (but not the last trimester) led to significant obesity in young adult males (Ravelli et al. 1976, 1999) and middle-aged females. This group also has an altered plasma lipid profile (Roseboom et al. 2000a) and a higher risk of coronary heart disease (Roseboom et al. 2000b), complications closely associated with obesity. Similar findings have been seen in animal models, with intrauterine growth restriction leading to obesity and related health problems (Jimenez-Chillaron and Patti 2007).

As previously suggested, this outcome may be due to a mismatch between the developmental and subsequent environments, with “catch-up growth” leading to an increased susceptibility to obesity. The “predictive-adaptive hypothesis” suggests that an animal adapts specifically to its early life (in utero or immediately post-natal) environment and this may have negative physiological consequences when it is exposed to a different nutritional environment later, i.e. if the “prediction” about the future environment turns out to be inappropriate (Gluckman and Hanson 2004; Wadhwa et al. 2009). Thus, rapid weight gain in the first weeks or months post partum is a significant risk factor for obesity in later life (Stettler et al. 2005). Remarkably, for every 100 g of weight gained in the first week of life, the risk of becoming obese as an adult increases by 28% (Stettler et al. 2005). Furthermore, investigation in the rodent has suggested that if this catch-up growth is prevented, the risk of obesity disappears (Ross and Desai 2005).

Not surprisingly, perinatal overnutrition is also a risk factor for obesity in later life (McCance 1962; Plagemann 2006; Chen et al. 2008; Morris and Chen 2009; Spencer and Tilbrook 2009). Although not usually constituting a stressor in itself, perinatal overnutrition does affect the HPA axis (Boullu-Ciocca et al. 2005; Spencer and Tilbrook 2009), making animals more responsive to stress and potentially contributing to an ongoing cycle of HPA axis hyperactivity and increased weight gain.

Conclusions and future directions

Obesity is currently epidemic within our society and among the many risk factors contributing to the aetiology, stress plays a significant role. Although eliminating stress from our everyday lives is unlikely to occur in the near future, we can work to further our understanding of the mechanisms by which stress may predispose an individual to obesity and potentially target some of these mechanisms.

As discussed, stress affects myriad appetite and adiposity regulating factors that may provide potential targets for treatment of obesity in people with an overactive HPA axis. Ghrelin is one such candidate. Acute and chronic stress both lead to an increase in circulating ghrelin, stimulating food intake (Kristenssson et al. 2006; Lutter et al. 2008). Ghrelin antagonists or inverse agonists are regarded as promising, but yet unproven, candidates for obesity treatment (Chollet et al. 2009). It is possible that these will also ameliorate responses to stress, affecting weight gain by inhibiting GC production as well as via a direct effect on appetite. Another promising target is 11β-HSD1. Mice that do not express 11β-HSD1 are resistant to obesity, and 11β-HSD1 inhibitors have been shown, in rodents, to improve glucose and insulin homeostasis (Morton et al. 2004; Stimson and Walker 2007; Macfarlane et al. 2008; Tiwari 2010). Encouragingly, carbenoxolone, a non-specific inhibitor of 11β-HSD1 and 11β-HSD2, has been shown to improve hepatic insulin sensitivity in diabetic and healthy lean patients (Walker et al. 1995; Andrews et al. 2003). It can also suppress lipolysis in healthy subjects (Tomlinson et al. 2007). Clinical trials with specific 11β-HSD1 antagonists are currently underway. Early results report successful inhibition of adipose 11β-HSD1 (Courtney et al. 2008), but effects on adiposity and indices of metabolic syndrome remain to be determined (Andrews et al. 2003; Fotsch and Wang 2008).

Recent critical findings have indicated that the predisposition to obesity in stressed animals may be differentially expressed depending on their sex. For instance, adiposity is differentially distributed in males and females, with males more likely to exhibit visceral adiposity (Hamdy et al. 2006). We have shown that female rats, but not males, that were overfed as neonates have enhanced HPA axis responses to psychological stress (Spencer and Tilbrook 2009). We also have some recent data to suggest that the mechanism by which obesity is maintained in neonatally overfed rats is different in males and females (Stefanidis, Tilbrook and Spencer unpublished observations, 2009–2010). These findings illustrate that the interplay between responses to stress and subsequent obesity is complex and probably subtly different between the sexes. They also highlight the need for sex-specific treatments for obesity.

The perinatal environment is also crucially important in programming the circuitry that regulates body weight and metabolism. It will be necessary for future studies to define the impact of and identify the mechanisms for stress effects on long-term body weight and metabolism so that we can develop strategies to ameliorate these effects.

Acknowledgements

This work was supported by the National Health and Medical Research Council (NHMRC) of Australia, the Australian Research Council (DP109339), and the Clive and Vera Ramaciotti Foundation. S. J. S. holds a Peter Doherty Fellowship awarded by the NHMRC of Australia (465167) and an Endocrine Society of Australia Postdoctoral Fellowship.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.