Intrauterine environment and fetal allergic sensitization

The possibility that raised cord blood IgE levels are either a risk factor or predictor of subsequent allergic disease has been a focus of research and subject of discussion for many years [ 1–3]. The conclusion to date has been that raised cord blood IgE is highly specific for subsequent atopic disease but has low sensitivity because most atopics did not have raised levels. Most recently, Edenharter et al. [ 4] have tried to resolve this issue. In their large German multicentre allergy study, elevated cord blood total IgE was identified as a strong risk factor for sensitization of infants (detected by the presence of > 0.35 kU/L allergen-specific IgE to one of nine test allergens) at 12 months of age, with a weaker positive association at 60 months, Unfortunately, the predictive performance was poor being described by the authors as ‘useless as a basis for preventive measures’. Remarkably, they found elevated cord blood IgE was a significant protective factor for early onset atopic dermatitis, a finding in conflict with that of Halonen et al. [ 5]. This finding is very difficult to explain. However, irrespective of the low sensitivity of cord blood IgE in predicting atopic disease, its presence in some subsequent atopics indicates that intrauterine allergic sensitization can occur. The search for risk factors and predictors of allergic disease detectable at birth extends beyond the frequently studied IgE. More recent observations have focused on antigen-specific proliferation by umbilical cord blood mononuclear cells which occurs in response to a variety of antigens, including allergens [ 6–8]. Thus, raised proliferative responses to β-lactoglobulin (βLG) and ovalbumin (OVA) at birth were associated with subsequent development of eczema at 1 year of age in children who had a positive skin-prick test to these allergens and whose symptoms improved upon withdrawal of either cow's milk or egg from the diet [ 6]. The ability to demonstrate proliferative responses at birth indicates that prior, therefore intrauterine, exposure to allergen has occurred leading to the generation of immunological memory. Such observations raise two main questions. What is the pathway of intrauterine exposure to allergen, and secondly, given the ubiquitous nature of allergens to which intrauterine exposure may occur, why do only some children undergo allergic sensitization upon allergen exposure in utero? A number of mechanisms have been postulated to account for intrauterine exposure to allergen but space limitations prevent discussion here. With regards to the second question, two recent studies using non-selected populations (for family history of atopy) highlight that a large proportion of newborns show evidence of intrauterine exposure to allergens. In a study by Szepfalusi et al. [ 7], 87% and 74% of cord blood mononuclear samples showed proliferative responses to the cow's milk allergens α-lactalbumin and β-lactoglobulin, respectively. Prescott et al. [ 8] similarly demonstrated that 47% of cord blood samples showed proliferative responses to house dust mite while 42% responded to OVA. Although neither of these studies have analysed the relationship between antigen-specific proliferative responses at birth and the onset of clinical symptoms of allergic disease in the child as in the study by Warner et al. [ 6], they do reveal that sensitization to antigen may be a common feature of pregnancy. Thus, allergen exposure at a vulnerable time during gestation may not in itself be sufficient for allergic sensitization, other factors must determine if the sensitization that does occur is of an allergic nature. One of the principle ways in which antigen-specific responses are controlled is via the prevailing cytokine milieu. The result of many years of study are that the cytokines IFNγ and IL-12 are of importance in generating TH1-type responses [ 9, 10] whereas IL-4 is important for TH2-type responses [ 11]. Thus IL-4, in addition to IL-5 and IL-13, has a central role in induction and maintenance of the allergic response. Despite the identification of NK cells and macrophages as sources of the cytokines required during antigen priming for the development of TH1-type responses [ 12, 13], the source of IL-4 at this crucial time has not been identified. Although mast cells [ 14], basophils [ 15], eosinophils [ 16] and NK1.1+ T cells [ 17] are well characterized producers of IL-4 (at least in adults), in vivo studies reveal that these populations are not essential for the development of TH2-type responses [ 18]. Where could TH2-favouring IL-4 originate during fetal development? In 1993 Thomas Wegmann et al. postulated that a predominance of Th2-type cytokines at the materno-fetal interface during gestation is vital to successful pregnancy [ 19]. This was based on murine studies and the observation that patients with rheumatoid arthritis report improvement in symptoms during pregnancy. The available data suggest that fetally derived human gestation-associated tissues also preferentially express Th2-type cytokines with IL-4 immunoreactivity evident in both the fetally derived amnion epithelium [ 20] and the placenta [ 21]. There are, however, a large number of other immunoregulatory molecules produced by maternally and fetally derived cell populations within gestation-associated tissues. The role-call of cytokines present at this site is dominated by those considered regulators of antigen presenting cell accessory function and/or differentiation of T helper cell populations, namely GM-CSF, TNFα, IL-1, IL-6, IL-10, TGFβ, IFNγ and, of course, IL-4. Maternally derived decidual tissue (the endometrium of pregnancy) is a source of bioactive TNFα in the first trimester and at term (> 37 weeks of gestation) [ 22] as is the human placenta [ 22, 23], with both placental macrophages [ 22] and trophoblast [ 23] identified as sources. Decidual and placental tissue are also a source of IL-1α, IL-1β, IL-6 and GM-CSF. Late gestation decidual tissue has been identified as a source of IL-10 and IFNγ [ 24], while placental trophoblast produces biologically active IL-10 [ 25]. TGFβ1 immunoreactivity is evident in the extracellular matrix of first trimester decidua and syncytiotrophoblast in the chorionic villi of the placenta [ 26]. We have recently identified IL-13 immunoreactivity in the human placenta [ 27] and Dearly et al. [ 28] have identified first trimester trophoblast as the source of both protein and transcripts for IL-13. IFNγ is also detectable by immunohistochemistry in the human placenta but its expression is limited to the first trimester [ 29]. Although these cytokines are likely to have biological functions additional to those already accredited to them, the pregnancy-associated immunological environment can still be viewed as dominated by many of the cytokines that are key regulators of immunological reactivity. These cytokines may influence pregnancy success and the development of the fetal immune system. The variety of cytokines present in gestation-associated tissues implies that an interacting balance in the levels, and therefore function, of these pleiotropic molecules exists at the materno–fetal interface. Upsetting this balance may have detrimental effects on the success of pregnancy and, in particular, the health of the developing fetus. As an example of this the administration of either IL-2 or IFNγ to pregnant mice was found to induce resorption of the developing mouse embryo by the mother [ 30]. The consequences of upsetting this balance in non-experimental systems may be more subtle. Could cytokines produced by gestation-associated tissues influence the development of the fetal immune system and, consequently, could a perturbation in the levels of these cytokines mediate altered immune reactivity in the developing human? It is relatively easy to conceptualize that placenta-derived cytokines could enter the fetal circulation as placental tissue surrounds the fetal blood vessels of the placenta. In support of this, transplacental transfer of cytokines has been demonstrated in vitro [ 31]. However, at first glance it is much more difficult to determine how cytokines that are produced by decidual and amnion tissue which are quite distant from the foetus influence its immune development, that is until you consider the amniotic fluid in which the fetus ‘floats’ throughout pregnancy. The amniotic cavity containing the amniotic fluid and fetus is lined by the fetally derived chorioamnion (comprising a layer of chorion and the amnion epithelium separated by an expanse of extracellular matrix) which is juxtaposed to the maternal decidua. Importantly, transfer of cytokines across fetal membranes has been demonstrated albeit under laboratory conditions [ 32]. Cytokines produced by cells within these tissues, as well as the placenta, could therefore be present in the amniotic fluid. Indeed, a large number of cytokines have been detected in human amniotic fluid, including many of those mentioned above. IL-4, IL-10 and IFNγ [ 24] have all been detected in amniotic fluid collected at term. TNFα has been detected in samples collected by amniocentesis at 16 weeks of gestation with levels higher at term and increasing during labour [ 33, 34]. IL-1, IL-6 and IL-8 are also detectable at term [ 34, 35]. Biological activity has been demonstrated for TNFα, IL-1 and IL-6, primarily because of the existence of well characterized bioassays for these. TGFβ1 and TGBβ2-like activity has also been observed in human amniotic fluid and these are activated in vivo, although latent TGFβ is also present [ 36]. As the study of amniotic fluid cytokines stems from interest in the onset of labour, particularly in the setting of infection-associated preterm labour, much of the work has focused on comparing levels of cytokines in the laboured and nonlaboured setting. However, data is now emerging on the association between amniotic fluid cytokine levels and various clinical conditions. For example, elevated TNFα levels were significantly associated with respiratory distress in preterm infants even after adjustment for birth weight [ 37]. Similarly, higher median concentrations of TNFα, IL-1β and IL-6 were found in amniotic fluid from mothers of neonates with brain white matter lesions (a risk factor for cerebral palsy) [ 38]. Of course, these elevated levels of cytokines may not be causal and serve simply as markers of disease. How does fetal exposure to amniotic fluid cytokines occur? Protein turnover in the amniotic fluid occurs at a rate of 70% each day with much of this removal via fetal swallowing [ 39]. As the fetus also aspirates amniotic fluid both the respiratory and gastrointestinal tract, in addition to the skin, are exposed to components of the amniotic fluid. The fetal gut is more permeable than that of either the child or the adult, and the skin is highly permeable during early pregnancy. Thus, cytokines in amniotic fluid could be absorbed by the fetus in immunoreactive forms. It is even possible that antigen itself may be present in the amniotic fluid. Of the lungs, gut and skin the fetal gut is the best studied. Given that the first manifestations of allergic disease are often food associated the interaction between the gut and molecules in the amniotic fluid may be critical to allergic sensitization. This interaction may continue postnatally especially during breast feeding. Human breast milk not only provides the best source of nutrients for the infant but it may also be a source of immunomodulatory molecules. G-CSF, M-CSF, IL-6 and IL-8 are all detectable in human breast milk and are produced locally in the mammary gland [ 40–42]. Other preformed cytokines present in human breast milk include TGFβ [ 43], IFNγ [ 44] and IL-10 [ 45]. Interestingly, many of these cytokines occur as high molecular weight forms which has been postulated to protect them from proteolytic cleavage in the gastrointestinal tract. Another source of cytokines in breast milk may arise from the passage of leucocytes from mother to infant during breast feeding. A high proportion of maternal milk T helper cells express the B cell costimulatory molecule CD40 ligand [ 46] and if IL-4 production is favoured by these cells the appropriate microenvironment for isotype switching by B cells would be generated. Thus both amniotic fluid and breast milk are sources of cytokines that could influence the developing fetal and neonatal immune systems. Interestingly, cells of the neonatal immune system have been recognized as deficient in their capacity to produce many of the cytokines. This apparent immaturity may be counterbalanced by alternative sources of these cytokines during pregnancy. Exposure to inappropriate levels of these cytokines may have dramatic consequences for the infant. To date, much of the study of cytokines in amniotic fluid has been performed on samples collected from full term pregnancies. However, allergen-specific responses are demonstrable from 22 weeks of gestation [ 47] and, more recently, allergen exposure prior to the sixth month of pregnancy was deemed necessary for allergic sensitization [ 48]. Thus, the contribution of gestation-associated tissues and amniotic fluid to the cytokine milieu that the developing fetus is exposed to during the time period at which it is most vulnerable to allergic sensitization remains to be evaluated. How might cytokines in the amniotic fluid influence allergic sensitization by the foetus? TNFα for example has been identified as required for the migration of Langerhans cells from the skin to the draining lymph nodes [ 49]. IL-10 has been described as favouring tolerance induction by dendritic cells [ 50]. IL-6 may promote the growth and differentiation of fetal T and B cells and, as discussed above, a relative abundance of either IL-4 or IFNγ could affect the development of TH2 and TH1 responses, respectively. Most importantly, IL-4 favours the induction of IgE-producing B cells while IFNγ inhibits this [ 51]. Perhaps measuring total IgE levels at birth or in early infancy would serve better as a risk factor or predictor if we measured levels in faeces instead of cord blood. To this end, Sasai et al. [ 52] found total IgE levels in faecal samples of 1-month-old infants higher in infants born into families with a history of atopy compared to history-negative families. Alternatively, as pointed out by Edenharter et al. [ 4], IgE-specific IgG is protective [ 53] and perhaps both total IgE and IgG anti-IgE should be measured with the balance being more informative. Although perturbed cytokine levels at the materno–fetal interface may mediate allergic sensitization of the fetus it is likely that a complex interaction of this with genotype and events ex utero will dictate whether clinically significant allergic disease manifests in the child and, later, the adult. Perhaps this is why cord blood IgE serves as a strong risk factor for sensitization but such a poor predictor. Early life programming of the developing fetal immune system may determine how our immune systems respond to antigen throughout life. Thus, an understanding of the interaction between the gestation-associated cytokine milieu and allergic sensitization of the fetus may have far-reaching implications for future preventive therapeutic strategies. CAJ is supported by a National Health & Medical Research Council (Australia) CJ Martin Post-Doctoral Training Fellowship.