This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
A comprehensive number of epidemiological and animal studies suggests that prenatal and early life events are important determinants for disorders later in life. Among them, prenatal stress (i.e. stress experienced by the pregnant mother with impact on the fetal ontogeny) has programming effects on the hypothalamic-pituitary-adrenocortical axis, brain neurotransmitter systems and cognitive abilities of the offspring. This review focuses on the impact of maternal stress during gestation on the immune function in the offspring. It compares results from different animal species and highlights potential mechanisms for the immune effects of prenatal stress, including maternal glucocorticoids and placental functions. The existence of possible windows of increased vulnerability of the immune system to prenatal stress during gestation is discussed. Several gaps in the present knowledge are pointed out, especially concerning the time when prenatal stress effects are expressed during postnatal life, why this expression is delayed after birth and whether prenatal stress predisposes to immune-related pathologies later in life.
Key words: prenatal stress, immune system, gestation, placenta, glucocorticoids
A growing body of scientific literature of epidemiological and animal studies suggests that besides genetic factors, environmental events acting prenatally on the developing fetus are important determinants for disorders later in life (for reviews, see Barker et al., 1993; Huizink et al., 2004; Fowden et al., 2006). Due to the fact that these disorders are often long-lasting even until adulthood, this phenomenon has been denoted as 'fetal programming'. Maternal gestational stress (i.e. stressful situations experienced by the mother with impact on the fetal ontogeny) is one of these factors (Whadwa 2005). The modulation of physiological systems by environmental stimuli during ontogeny has been proposed to be a mechanism favouring the adaptation of the developing offspring to the post-partum environment (Fowden et al., 2006). However, these developmental adaptations could be detrimental if the organizational effects by external stimuli exceed the tolerable range or if the later environment is not conform to the 'expectancy' (Coe and Lubach, 2005).
Stressful situations for pregnant females have long-term consequences on physiology and disease risk of their offspring. In rodents and non-human primate species, it was found that prenatal stress causes alterations of the hypothalamic-pituitary-adrenocortical (HPA) axis and brain neurotransmitter systems in the offspring (for reviews, see Kofman, 2002; Maccari et al., 2003), increases anxiety and emotionality and decreases motor development and learning abilities (for reviews, see Welberg and Seckl, 2001; Huizink et al., 2004). In humans, evidence has been provided that maternal psychosocial stress reduces the length of gestation and concomitantly the weight at birth, but the impairment of fetal growth itself is still controversial (Nordentoft et al., 1996; Wadhwa 2005; Precht et al., 2007). Because low birth weight is correlated with a higher susceptibility to coronary heart disease and type 2 diabetes in later life (Barker et al., 2002), prenatal stress is suspected to increase the susceptibility to these diseases (Wadhwa, 2005).
Little attention has been devoted to the influence of prenatal stress on the developing immune system of the human infant. Maternal stress has been proposed to have implications for the development of atopic diseases (von Hertzen, 2002). For example, self-reported maternal nervousness during gestation has been reported to be correlated with high levels of immunoglobulins E in cord blood, which is used as a predictor for atopic diseases in early childhood (Lin et al., 2004). Furthermore, recent data show that pregnant women experiencing stress display increased serum levels of pro-inflammatory cytokines (Coussons-Read et al., 2005 and 2007), and alterations in maternal circulating cytokine levels during pregnancy are suspected to be a cause for a higher risk of allergy for the infant later in life (Hamada et al., 2003; Prescott et al., 2005; Pincus-Knackstedt et al., 2006). Given the scarcity of human data, the aim of this review is to present results from animal studies in order to highlight common or contradictory findings on the effects of prenatal stress on immune function, potential mechanisms and mediators, as well as possible windows of increased vulnerability.
Consequences of prenatal stress on the immune function of the offspring
2.1 Relevance of animal studies for humans
Although animal studies provide useful information, it has to be kept in mind that the generalization of findings from animals to humans is limited by differences in gestational length, in utero developmental time-line and inter-species differences in physiology. For example, primates are the only species that produce placental corticotropin-releasing hormone (CRH) during pregnancy (Robinson et al., 1989). To compare maturational landmarks of the ontogeny of the immune system among species, the use of arbitrary time units has been proposed (Holladay and Smialowicz, 2000). Since the gestational length in mice is approximately 20 days and 40 weeks in humans, a specific maturational landmark occurring on week 7 in humans or on day 10.5 in mice can be 'translated' in arbitrary units, such as 0.18 for humans and 0.53 for mice. However, this tool only partly solves the problem: a stressful event occurring for example at 0.50 of gestational length will probably have different effects in humans and rodents because the timing of maturation of the immune system to birth is also highly species specific (Fig 1). In general, animals that give birth to precocious offspring (sheep, pigs, guinea pigs, primates) display development of the immune system predominantly in utero (Holladay and Smialowicz, 2000; Holsapple et al., 2003). By contrast, in species that give birth to non-precocious offspring (rats, rabbits, mice), a large proportion of the development occurs during late gestation and the postnatal period (Holladay and Smialowicz, 2000; Holsapple et al., 2003). As an example, a stressful event has to occur around 0.18 of gestation length in humans and around 0.53 of gestation length in mice to target the beginning of fetal liver hematopoiesis (Holsapple et al., 2003).
2.2 Effects of prenatal stress on the immune function of the offspring - results from animal studies
The effects of maternal psychological stress on the innate immunity of the offspring are generally inhibitory. In mice born from mothers exposed to psychological stress during the third part of gestation, macrophage and neutrophil functions such as spreading and phagocytosis were inhibited in two-month-old offspring (Palermo-Neto et al., 2001; Fonseca et al., 2002). Prenatal stress during the same gestational period in mice or rats (between 0.66-0.80 and 0.95) decreased cytotoxicity of blood and spleen natural killer (NK) cells and in vivo resistance to experimentally-induced tumors (Klein and Rager, 1995; Kay et al., 1998; Palermo-Neto et al., 2001). By contrast, prenatal stress at a comparable gestational age did not affect NK cytotoxicity in pigs (Tuchscherer et al., 2002). The contradiction between pig and rodent studies might be explained by the fact that NK cell ontogeny occurred at the end of the stress period in mice, while it was accomplished before the stress treatment in pigs (see Fig. 1). Several studies have also investigated the capacity of leukocytes to produce inflammatory cytokines following exposure to lipopolysaccharide (LPS). In rhesus monkeys, blood cells from babies that were stressed either during the second or the last third of gestation released significantly less tumor necrosis factor ï¡ (TNF-ï¡) and interleukin 6 (IL-6) upon LPS stimulation in vitro (Coe et al., 2002). By contrast, prenatal stress in rats has been shown to increase corticosterone and fever responses to LPS (Hashimoto et al., 2001) and allergic airway inflammation (Pincus-Knackstedt et al., 2006).
Thymus size and morphology at birth can be affected in prenatally stressed rodents (Hashimoto et al., 2001) and pigs (Tuchscherer et al., 2002). The alteration of thymic function is illustrated by a decreased number of total lymphocytes (Llorente et al., 2002) as well as CD4+ and CD8+ lymphocytes (Götz and Stefanski, 2007) in blood of adult rats born from mothers stressed during either late (0.63 to 1 of gestation length) or entire gestation. Another study reported no alterations in T lymphocyte subsets in the spleen and blood of rats (Kay et al., 1998). The reason for these discrepancies may be that the stressor was of shorter duration and occurred later during gestation in this last study (gestational period 0.77 to 0.80). Proliferation assays performed in vitro to assess functional capacity of T cells give variable results. In rats, response to concanavalin A (ConA) was decreased in animals of 14 days (Sobrian et al., 1997) but not in animals of 1 to 3 months (Klein and Rager, 1995). Response to phytohemagglutinin A (PHA) was increased in 21 days of age but not in 2-month-old rats (Sobrian et al., 1997; Kay et al., 1998). Kinetic studies bring some explanation to these contradictory results and reveal that the effect of prenatal stress on lymphocyte response to mitogens strongly depends on the age of the offspring. In pigs, restraint during the last third of gestation decreased the response to ConA until 3 days of age but this effect was not observed any more at 7, 21 or 35 days (Tuchscherer et al., 2002). In rats, the increase in response to PHA in prenatally stressed animals develops after birth and becomes observable not before 3 or even more than 7 weeks after birth depending on the study (Sobrian et al., 1997; Vanbesien-Maillot et al., 2007).
Prenatal stress also modifies the response of T lymphocytes to specific antigens. The capacity of T cells from 3-day-old monkeys to respond to non-self antigens in mixed lymphocyte cultures is decreased in babies whose mothers were stressed during mid-late pregnancy (0.62 to 0.87 of gestation length) and increased in babies whose mothers were exposed to the same stressor during early pregnancy (period 0.30 to 0.54, Coe et al., 1999). This reveals that the direction of the alteration depends on the time of application of the stressor during pregnancy. Detrimental effects of prenatal stress on antigen-specific T-cell dependent responses are also illustrated by the reduction of the delayed-type hypersensitivity (DTH) response in two-month-old rats (Sobrian et al., 1997).
Prenatal stress in rats has been shown to increase the proliferation of splenic B cells in response to pokeweed mitogen in pups aged less than one month (Sobrian et al., 1997). However, the proliferation is decreased in two and five-month-old offspring (Kay et al., 1998; Götz and Stefanski, 2007), suggesting again that the effect of prenatal stress on B-cell proliferation depends on the age of the offspring. Antibody production by B cells in response to an antigenic challenge has been found to be increased for the primary immunoglobulin M (IgM) as well as for the secondary immunoglobulin G (IgG) responses and this is a long-lasting effect of prenatal stress, since it can be observed in adult rats (Klein and Rager, 1995). However, a decrease in antibody production in prenatally stressed mice was observed in another study (Gorczynski, 1992) and decreased levels of neutralizing antibody were found when rats were challenged with a live virus (Sobrian et al., 1997). All these studies were performed in rats or mice, with a stressor occurring during the gestational period from 0.63-0.68 to 0.95, but they differed in the kinds of stressors that were used, suggesting that the nature of the stressor may play a major role in the immune outcomes.
Important gaps remain in the present knowledge concerning the effects of prenatal stress on the immune function. Especially, there is no explanation yet why the time when these effects appear after birth is variable between studies and why this expression is often delayed in time (Sobrian et al., 1992; Tuchscherer et al., 2002; Fonseca et al., 2002; Vanbesien-Maillot et al., 2007). Furthermore, more data are required to determine whether prenatal stress predisposes to immune-related pathologies later in life.
Consequences of prenatal stress on the transfer of passive immunity to the offspring
Prenatal stress may affect the acquisition of maternal IgG differently in species where IgG are transferred to the offspring mainly through the placenta (rodents, primates) or via colostrum intake (ongulates). In rats, inescapable electric shocks during the last fourth of gestation were found to reduce fetal IgG levels at birth (Sobrian et al., 1992), while in mice, maternal restraint-and-light stress during the second half of gestation failed to affect the transfer of total as well as specific antibodies to herpes simplex, maintaining a satisfactory protection against a systemic herpes virus challenge (Yorty and Bonneau, 2003 and 2004). Authors proposed that the absence of effect on the transfer of passive immunity in these last studies resulted from the nature of the psychological stressor used, which is supposed to be milder than the electric shocks used in the previous studies. In squirrel monkeys, IgG levels are decreased in male and unaffected in female offspring after chronic social stress during gestation (Coe and Crispen, 2000). This suggests a sex-specific mechanism comparable to behavioural and neuroendocrine effects in many prenatal stress studies (Weinstock, 2007). Furthermore, maternal IgG acquisition seems to be affected only if the stress period covers the end of gestation, which corresponds to the period of active transplacental transfer of maternal IgG, while a stress occurring during mid-gestation has no effect on IgG acquisition (Coe and Crispen, 2000).
In pigs, heat-stress during the last week of pregnancy or psychological stress during the last third of pregnancy decreased (Machado-Neto et al., 1987; Tuchscherer et al., 2002) while cold stress during the last two days increased (Bate and Hacker, 1985a) the circulating IgG levels in neonates. As equal IgG levels were found in neonates before the first colostrum intake (Machado-Neto et al., 1987), the decrease in prenatally stressed piglets may result from a lower IgG content in maternal colostrum. Indeed, lower levels of immunoglobulins were reported in the colostrum of sows and cows stressed during late gestation (Machado-Neto et al., 1987; Nardone et al., 1997). Another explanation could be that the intestinal absorption of IgG is altered in neonates (see section 3.2.2).
Maternal molecular mediators of prenatal stress and mechanisms of action
3.1 Hormonal or neuroendocrine mediators
Stress in adults leads to numerous cardiovascular and endocrine changes, including increase in plasma adrenocorticotropic hormone (ACTH), ï¢-endorphin, glucocorticoids and catecholamines (Harbuz and Lightman, 1992; Johnson et al., 1992). The release of these hormones in maternal bloodstream could act on fetal development both directly, via regulation of fetal gene expression, or indirectly, through changes in placental metabolism. Indeed, hormones play a major role in regulating growth (insulin, thyroid hormones, insulin-like growth factors) and development (thyroid hormones, steroids) of fetal tissues (Fowden and Forhead, 2004). Although the placenta is a relatively impermeable barrier, several maternal hormones can cross it, and the differences in placental structure among species have only a limited influence on this transfer (Fisher, 1998). These hormones are lipophilic hormones such as the steroids and peptide hormones of less than 0.7 to 1.2 kD (Fisher, 1998). Other stress hormones such as CRH or ACTH apparently do not cross the placenta (Dupouy et al., 1980; Mastorakos and Ilias, 2003).
Maternal glucocorticoids are a major candidate for the mediation of prenatal stress to the fetus (for reviews, see Seckl, 2004; Kapoor et al., 2006). They can cross the placental barrier, as shown in rats (Zarrow et al., 1970), sheep (Hennessy et al., 1982) and pigs (Klemcke, 1995), and they exert many organisational effects on prenatal tissue development (Muglia et al., 1995). The access of glucocorticoids to the fetal compartment is regulated by the placental enzyme 11ß-hydroxysteroid dehydrogenase type 2 (11ß-HSD2, Benediktsson et al., 1997; Staud et al., 2006). Under normal conditions, fetal exposure to maternal glucocorticoids is maintained at a low level (Benediktsson et al., 1997), but excessive increases in maternal glucocorticoids (Otten et al., 2004) or suppression of 11ß-HSD2 activity (Staud et al., 2006) can lead to increased exposure. Elevated glucocorticoids in pregnant females result in growth retardation, deregulation of the HPA axis, behavioural alterations and glucose intolerance in the offspring (for reviews, see Welberg and Seckl, 2001; Kapoor et al., 2006). Suppression of glucocorticoid elevation by adrenalectomy abrogates the effects of prenatal stress on the HPA function in rats (Barbazanges et al., 1996).
The involvement of glucocorticoids as mediators for the effects of prenatal stress on immune function has not yet been incontestably demonstrated. Some studies investigated whether artificially elevated levels of glucocorticoids in the mother can mimic the effects of prenatal stress on the immune system of the progeny. Glucocorticoid treatment generally inhibits the immune variables explored, whereas maternal psychological stress has more inconsistent effects (Table 1). Dietert et al. (2003) found that daily dexamethasone application during the two last thirds of gestation in rats induced thymic involution, had no effect on blood lymphocyte subsets numbers, decreased DTH and increased IgG response to keyhole limpet hemocyanin in the offspring. Repeated ACTH administration to pregnant mothers during the last third of gestation decreased lymphocyte functions in rhesus monkey and pig offspring (Coe et al., 1996; Otten et al., 2007). Other studies evaluated the possible deleterious effects of short-term corticosteroid treatment at the very end of gestation in various mammal species. They found that antenatal corticosteroid treatment transiently suppresses several fetal monocyte functions, decreases lymphocyte proliferation, induces thymic involution, increases the CD4/CD8 T cell ratio in the blood and decreases T cell numbers in the spleen (Murthy and Moya, 1994; Bakker et al., 1995; Bakker et al., 1998; Kramer et al., 2004). Neither cell-mediated hypersensitivity nor humoral responses were affected in these studies. The only studies where the effects of maternal stress and maternal elevated glucocorticoid levels were compared in the same experiment and at comparable gestational times were performed in rhesus monkeys. A short dexamethasone exposure at the very end of pregnancy (gestational period 0.85-0.86) had effects comparable to those of a longer late gestational stress (gestational period 0.62-0.87) on mixed lymphocyte proliferation response (Coe et al., 1999). In another study, the ability of two-year-old monkeys to produce IL-6 and a fever response to an injection of IL-1 was decreased in babies from ACTH-treated but not from psychologically stressed mothers (Reyes and Coe, 1997). These discrepancies suggest that glucocorticoids may be one but probably not the only mediator of prenatal stress effects.
Catecholamines and opioids
Catecholamines may influence both placental and fetal units. Indeed, elevated levels of catecholamines in maternal blood lead to constriction of placental blood vessels, decreasing fetal supply of glucose and activating fetal catecholamine release (Gu and Jones, 1986). Most of the catecholamines taken up by the placenta are metabolized by the enzymes monoamine oxidase and catechol-O-methyltransferase (Chen et al., 1974). Nevertheless, several early studies reported that small but detectable amounts (typically 5-10%) of unmetabolized noradrenaline were transferred from the maternal to the fetal side of the human and guinea pig placentas (Morgan et al., 1972; Saarikoski, 1974; Sandler et al., 1963; Sodha et al., 1984). This placental transport may play a role during early fetal development, before maturation of the fetal sympathoadrenal system (Thomas et al., 1995).
A few studies support the hypothesis that opioids may be other candidates for the mediation of prenatal stress. Indeed, fetal thymic cells are sensitive to opioids, which could be involved in T-cell thymic selection (McCarthy et al., 2004). Furthermore, in mice, treatment of the mothers with naloxone, an opioid antagonist, prevented prenatal stress effects on peritoneal macrophage activity (Fonseca et al., 2005). In chickens, prenatal exposure of fetuses to opiates inhibited the fever response to LPS as well as the hypersensitivity response to phytohemagglutinin (Schrott and Sparber, 2004), which resembles the effects obtained in prenatally stressed young mammals.
Cytokines and progesterone
Pro-inflammatory cytokines may be another candidate. Stress increased plasma levels of TNF-ï¡ and IL-6, but decreased IL-10 levels in pregnant women (Coussons-Read et al., 2005 and 2007). Increased pro-inflammatory Th1 cytokines at the materno-fetal interface can lead to rejection of the fetus (Joachim et al., 2003; Blois et al., 2004) and are suspected to be a cause for a higher risk of allergy for the infant later in life (Hamada et al, 2003.; Prescott et al., 2005). Progesterone could be involved in the regulation of maternal cytokine levels during gestation, at least in rodents. In pregnant mice, stress induced a decrease in progesterone levels, which resulted in an increase of Th1 cytokine production at the placental level (Joachim et al., 2003; Blois et al., 2004).
3.2 Possible mechanisms leading to immune alterations by prenatal stress
3.2.1 Direct hormonal action on the ontogeny of immune organs
Maternal hormones and neuromediators released during maternal stress can exert their action via several mechanisms. The first hypothesis is that they reach fetal organs and directly influence the ontogeny of immune cells. This may be the case for glucocorticoids, which play a major role in fetal ontogeny and tissue maturation before birth (Muglia et al., 1995). Their role in the ontogeny of the immune system has been demonstrated in transgenic mice with reduced glucocorticoid receptor (GR) function or adrenalectomized pregnant rats (Sacedón et al., 1999a and b). Epigenetic regulation, e.g. gene silencing via DNA methylation and/or histone deacetylation/methylation, has been proposed as one process by which environmental factors could induce long-lasting regulation of gene expression in immune cells (Sanders, 2006).
3.2.2 Hormonal action on other fetal organs
Immune alterations may be an indirect effect of a deregulation of the HPA axis in prenatally stressed offspring (Matthews, 2002). Indeed, lymphoid organs, and particularly the thymus, display a markedly elevated expression of GR during extra-uterine life (Miller et al., 1998). Immune cells are targets of glucocorticoids which regulate many of their properties including tissue-selective homing, activation, expansion and cell death (Dhabhar and McEwen, 2001; Herold et al., 2006). Immune anomalies in prenatally stressed offspring could be a "side-effect" of HPA dysfunction (Fig. 2). This hypothesis is supported by studies showing that neuroimmune communication is altered in prenatally stressed offspring. The sensitivity of blood mononuclear cells to glucocorticoids has been shown to be decreased using both an in vitro approach in rhesus monkeys (Coe et al., 2002; Coe and Lubach, 2000) and an in vivo approach in pigs (Tuchscherer et al., 2002).
There is also evidence for an effect of glucocorticoids on gut maturation and the resulting acquisition of maternal immunoglobulins via colostrum (Bate et al., 1991; Bate and Hacker, 1985a and b; Machado-Neto et al., 1987). Naturally, fetal glucocorticoid levels increase during the last days before parturition and then decrease continuously after birth until reaching levels comparable to adult levels. This increase in cortisol has been proposed to be the signal for enhanced gut permeability for immunoglobulins, whereas the decrease after birth would start gut maturation and closure (Bate et al., 1991). In late gestation, a transient increase in maternal glucocorticoids due to maternal stress may mimic the natural cortisol variations which occur around parturition and accelerate gut maturation, thus leading to impaired immunoglobulin acquisition after birth (Machado-Neto et al., 1987; Bate et al., 1991).
3.2.3 Action via alterations in placental function
Mediators released during maternal stress may also affect fetal immune development via indirect mechanisms involving placental activity (Fig. 2). The placenta provides an immune interface between the mother and the fetal allograft, transports nutrients and waste products between the mother and the fetus, and is a source of many peptides and hormones that influence fetal, placental and maternal metabolism (Reis et al., 2001). Elevated maternal glucocorticoids can alter placental trophic or endocrine functions. For example, increased maternal cortisol induced changes in placental glucose transporter (GLUT) gene expression, which may permanently alter transplacental glucose transport to the fetus with implications for fetal metabolism and growth (Langdown and Sugden, 2001). Another example has been provided in monkeys and involves the placental transfer of maternal iron to the fetus. This transfer enables the baby monkey to be born with enough iron stores to sustain postnatal growth for 4-6 months. In prenatally stressed babies, anaemia emerges between 4 and 8 months of life, which is related to iron deficiency, and correlates with a drop in NK activity (Coe and Lubach, 2005). A third mechanism involving the placenta has been proposed in primates where the placenta produces CRH. In these species, maternal glucocorticoids are suspected to stimulate CRH production by the placenta, which in turn would activate the fetal HPA axis and hence increase circulating glucocorticoid levels in fetuses (Challis et al., 2000).
3.2.4 Action via maternal care after birth
Finally, maternal gestational stress can have persistent effects in the mother with impact on the maternal behaviour after birth. In mice and rats, repeated daily restraint during gestation decreased maternal care to the pups and increased maternal anxiety until weaning (Maccari et al., 1995; Smith et al., 2004). As adoption reversed the negative effects of prenatal stress on HPA axis function, it was hypothesized that maternal behavioural disturbances, e.g., reduced maternal care, may cause stress for the offspring and contribute to the long-term effects of prenatal stress on the HPA axis of the offspring. Similar mechanisms are suspected for the immune alterations observed in prenatally stressed animals (Gorczynski, 1992).
4. Critical windows of vulnerability?
The immune effects of prenatal stress have been shown to be different depending on the gestational timing of the stress (Coe et al., 1999), and to be more or less persistent after birth depending on the studies (Tuchscherer et al., 2002; Vanbesien-Maillot et al., 2007). The nature and the persistence of prenatal stress effects probably depend on the time of application of maternal stress relatively to the fetal stage of development. Critical developmental windows of vulnerability of the immune system to environmental programming are not known yet, but they probably differ between species due to dissimilar prenatal maturation (Dietert et al., 2003). Three main periods of vulnerability have be proposed by immunotoxiologists: the period of embryonic stem cell formation (gestational period 0.38-0.53 in rodents, 0.13-0.18 in humans), the period of fetal liver development as the primary hematopoietic organ (0.53-0.62 in rodents, 0.18-0.26 in humans) and the period of colonization and establishment of the bone marrow and thymus (0.65-1 in rodents, 0.23-0.31 in humans; Landreth, 2002; Holsapple et al., 2003). To date, rodent studies on prenatal stress do not clarify this question because stressors have been applied during rather broad periods which often cover the entire immune developmental period (e.g., exposure to the stressor during the second or last third of gestation, see Table 2). Therefore, future studies on immune imprinting by prenatal stress should focus on the impact of stress during specific stages of fetal immune system development.
The authors thank Dr Ulrike Gimsa and Dr Armelle Prunier for their critical review of the manuscript.