Why We Develop Food Allergies
Coached by breast milk and good bacteria, the immune system strives to learn the difference between food and pathogens before the first morsel crosses our lips
No peanuts. No dairy. No eggs or shellfish or soy. No wheat or corn, no tree nuts or fin fish, no sesame seeds or spices of any kind. Few people have a diet this restrictive, but allergies to foods affect at least 1 in 20 young children and about 1 in 50 adults in industrialized countries. The numbers are rising: According to a recent study, the prevalence of peanut allergy—which accounts for the majority of emergency-room visits and deaths related to food allergies each year—doubled between 1997 and 2002.
The story of food allergy is a story about how the development of the immune system is tightly linked to the development of our digestive tract or, as scientists and physicians usually refer to it, our gut. A human being is born with an immature immune system and an immature gut, and they grow up together. The immune system takes samples of gut contents and uses them to inform its understanding of the world—an understanding that helps safeguard the digestive system (and the body that houses it) against harmful microorganisms.
The many-layered defenses of the immune system are designed to guard against invaders while sparing our own tissues. Food represents a special challenge to this system: an entire class of alien substances that needs to be welcomed rather than rebuffed. An adult may pass a ton of food through her gut each year, nearly all of it distinct at the molecular level from her own flesh and blood. In addition, strains of normal, or commensal, bacteria in the gut help with digestion and compete with pathogenic strains; these good microbes need to be distinguished from harmful ones. The body's ability to suppress its killer instinct in the presence of a gut-full of innocuous foreign substances is a phenomenon called oral tolerance. It requires cultivating a state of equilibrium, or homeostasis, that balances aggression and tolerance in the immune system. Intolerance, or failure to suppress the immune response, results in an allergic reaction, sometimes with life-threatening consequences.
An infant floating in the womb enjoys warmth, nutrition and an environment free of microorganisms. During the birth process—even before she takes her first breath—a baby begins to encounter microbes and other foreign substances, collectively called antigens, that can stimulate her new immune system. Most of these immunological challenges take place on mucosal surfaces such as the gut and airways.
The first line of defense in a newborn's gut is the system of immune exclusion, which uses exported antibodies to bind germs and potentially harmful compounds on the mucosal surface. Antibodies coat the pathogens to prevent them from invading the gut wall, and they bind to unfamiliar cell fragments or macromolecules to regulate their passage into the body. The class of antibodies known as secretory immunoglobulin A(SIgA) is most responsible for immune exclusion; it is a nice antibody that is actively pumped out to the surface and seldom elicits inflammation when it goes to work.
New babies, however, produce little or no SIgA. They depend on other types of antibodies during the first vulnerable months of life, primarily residual IgG from the mother and small amounts of mucosal IgM. The only significant source of SIgA antibodies during this period is breast milk, which helps protect the newborn until her immune system is established. In developed countries, the child's ability to produce SIgA is quite variable, being completed between one and ten years of age. Babies in developing countries often establish secretory immunity much earlier, presumably because of greater exposure to stimulating microbes.
In addition to their job of binding up troublesome antigens, SIgA antibodies help the gut to develop by enhancing the barrier function of the epithelial lining. The gut mucosa of most infants matures during the first months of life. But in some children, the mucosal barrier remains inadequate for several years, and incomplete secretory immunity can contribute to the delay. Not surprisingly, genetically manipulated (knockout) mice that lack SIgA and SIgM have leaky mucosal membranes.
The SIgA system seems to be important in setting an individual's threshold for adverse reactions to food. The risk of food allergy is higher when the development of IgA-producing cells is retarded or when SIgA-dependent development of the gut barrier is insufficient. On the positive side, babies who breastfeed exclusively for at least the first four months appear to have fewer allergies. This effect may be the product of IgA-directed gut maturation, but human milk also contains immune cells, immune-regulating cytokines and growth factors that exert positive biological effects.
Immune cells are woven into the fabric of the gut rather than being restricted to one place, but there are also discrete structures for immune surveillance. Dotting the prairie of tiny villi that lines the gastrointestinal tract are swollen domes called Peyer's patches. These regions, part of a larger system of gut-associated lymphoid tissue or GALT, are covered by an epithelial-cell layer containing specialized M cells (the M stands for membrane or microfold), which constantly scan the stream of passing antigens and transport them to the principal cell types in the immune system—B cells (from the bone marrow), T cells (from the thymus) and antigen-presenting cells (APCs) such as macrophages and dendritic cells. It is here that mucosal immunity is induced and regulated.
What follows the identification of an antigen is a complicated ballet of cells, secreted signals and movement from one compartment of the body to another. The keys to the system are the APCs, the "decision makers" in the immune system, which link innate and adaptive immunity. APCs process chunks of antigen brought in by M cells and then show the pieces, along with a selection of co-stimulatory signals, to so-called naive T cells, which have never met their cognate antigens before. Those specific T cells whose antigen receptors match one of the pieces become primed or activated; they then release cytokines (hormone-like regulatory proteins) and growth factors that instruct B cells to proliferate, differentiate and begin producing IgA. Activated T and B cells migrate to nearby lymph nodes to receive additional biological signals; most of those cells then enter the bloodstream. Many will return to the lamina propria of the gut, the tissue layer beneath the surface epithelium, or to mammary glands in lactating mothers through a kind of chemical navigation system. There, depending on what antigen-induced "second signals" the B cells receive, they may undergo one last, or terminal, differentiation to become plasma cells, which produce antibodies in quantity (about 10,000 molecules per second).
The system works differently in newborns who have never encountered microbes. Very few IgA-producing B cells circulate in the blood of newborns, although this number is approximately 75 times higher after the first month of life, a period of continuous stimulation of GALT by microbial antigens.
In the GALT structures, APCs need to receive certain "danger signals"—fragments of commensal bacteria from the digestive tract—to provide the right mix of co-stimulatory signals that prime helper T (Th) cells to aid the B cells. Without this timely inoculation with bacteria, the IgA system fails to develop normally. Bacteria from the genus Bacteroides and certain strains of Escherichia coli seem to be particularly good at stimulating the mucosal immune system. Lactic-acid-producing bacteria (lactobacilliand bifidobacteria) also contribute. These microbes help establish and regulate the epithelial barrier as well.
At least in mice, many of the beneficial effects of the commensal microbiota come from the binding of bacterial components by pattern recognition receptors on the surface of or inside the epithelial cells. This binding starts a back-and-forth, homeostasis-enhancing exchange of signals between epithelial cells and cells in the underlying lamina propria, including macrophages and dendritic cells. Experiments in mice suggest that before birth, cells lining the gut can detect certain parts of chewed-up microbes—particularly the component of the bacterial cell wall known as endotoxin or lipopolysaccharide (LPS)—because the cells contain an intracellular receptor for this common bacterial signature. Exposure to LPS in the mother's vaginal tract during birth modulates the gut epithelium so that it becomes tolerant to microbial patterns after birth. In remarkable contrast, mice delivered by caesarean section do not show signs of epithelial tolerance. These observations may be relevant to humans: Children who have a genetic predisposition to produce excess IgE (as indicated by mothers who suffer from various allergic reactions—a condition called atopy) are at least eight times as likely to develop food allergy when delivered by caesarean section.
Oral tolerance is not a single process but a complex series of events that contribute to intestinal and systemic immunosuppression. Many variables influence the development of oral tolerance (and therefore of food allergy): genetics, age, the dose and postnatal timing of fed antigens, the structure and composition of those antigens, the integrity of the epithelial barrier, and the extent to which nearby immune cells are simultaneously activated.
Human milk helps the gut tolerate certain food antigens early in life. Antibodies to gluten peptides from wheat are present in breast milk, and breastfeeding has been shown to protect significantly against the development of gluten-triggered celiac disease in children. This observation hints that mixed feeding, rather than abrupt weaning, may promote greater tolerance to food proteins in general.
This tolerance depends in part on the mothers' own immune function. In a study of breastfed infants, the ones whose mothers had low levels of anti-bovine antibodies were more likely to develop cow's-milk allergy later in life. Human milk also contains cytokines and growth factors that might account for its tolerance-promoting properties by modulating the activation of GALT and enhancing the function of the epithelial barrier. Most epidemiological studies support the view that breastfeeding protects against asthma and atopic dermatitis, or eczema, although this notion remains controversial. Nonetheless, the reinforcing effect of breast milk on mucosal barrier function in infants is robust and has special significance in families with a history of allergy.
As we currently understand it, oral tolerance is effected mainly through T-cell maturation events, such as anergy (a kind of cellular hibernation), clonal deletion (which removes T cells with undesirable targets) and, particularly, amplification of the immune system's voice of reason, the regulatory T (Treg) cell. As a result, healthy people have hardly any hyperactivated effector T cells (Teff) in their gut mucosa, scant mucosal production of proinflammatory IgG, and only low levels of IgG antibodies to food antigens in serum.
Food allergies vary in their severity and how swiftly symptoms appear. The immediate, life-threatening reactions experienced by some people (most often to peanuts) happen when the allergen binds to IgE-type antibodies, which then trigger the release of histamine, the compound responsible for acute inflammation with itching, sneezing and other allergy symptoms. Other types of food allergies result from IgG or IgM antibodies, or from so-called delayed-type hypersensitivity (not depending on antibodies). The latter reaction is typified by gluten-triggered celiac disease and may involve local dysregulation of both innate and adaptive immune functions. Delayed-type reactions may not show the hallmarks of classical inflammation that characterizes faster reactions.
Food allergies can be serious enough by themselves, but they can also announce the start of an "allergic march" that leads to antigen-triggered respiratory diseases. People who inherit a predisposition to atopy are at particular risk. Asthma and other atopic respiratory diseases have certainly become more common in developed countries during the past two decades.
Upon encountering a novel antigen, the immune system must decide whether the antigen is pathogenic (meriting a so-called productive immune reaction, which tries to eliminate the antigen) or harmless (leading to a suppressed response). If commensal bacteria have in the past modulated APCs (macrophages and dendritic cells) via their pattern-recognition receptors, then these cells are more likely to express co-stimulatory molecules and secrete the types of cytokines that encourage Treg cell development and, therefore, tolerance.
In a healthy gut, the APCs are continuously exposed to components of the normal gut microbiota. Like a sleepy sheriff in a peaceful town, a dendritic cell raises only a negligible alarm of proinflammatory cytokines on getting a whiff of microbial LPS. In this quiescent state, the dendritic cells carry antigens from the gut to nearby lymph nodes, where, in a normal maturation process, the cells become further conditioned for tolerance and drive the expansion of Treg cells. Some of the Treg cells travel through lymphatic and blood vessels back to the gut mucosa to maintain homeostasis.
Altogether, the body avoids unnecessary hyperactivation of its immunological sentinels (along with potentially harmful inflammation) in two ways: initially, with the restrained alarm, and also later when the activated Treg cells migrate to the mucosa to exert homeostatic control. At the same time, the macrophage deputies in town, which retain their ability to engulf and kill microbial invaders, continue to do the work of getting rid of commensal bacteria that sneak past the gut lining.
In the gut-associated lymph nodes, conditioning for oral tolerance depends on the menu of microbial components that the dendritic cells receive. Tolerance to food proteins is more likely to develop in the presence of telltale components from certain commensal bacteria as well as from harmless bacteria native to soil and from surface water or parasites such as flatworms or flukes. (It's nice to know these pests are good for something.) This evidence supports the extended hygiene hypothesis, which argues that a too-hygienic lifestyle in industrialized countries can prevent the mucosal immune system from maturing, leading to inadequate secretory immunity and fewer Treg cells. In a way, you could say that our immune system has lost its stimulating "old friends." Supporters of the hypothesis speculate that this inadequacy could help explain the increasing incidence of allergy and other immune-mediated inflammatory disorders in Westernized society.
Several clinical studies have tested the hygiene hypothesis by evaluating the effect of probiotic preparations, which deliver to the gut new colonists—certain strains of commensal bacteria or intestinal parasites from other species. (Eggs from the pig whipworm Trichuris suis, which don't pose an infection risk to humans, are the stimulants of choice in the latter experiments.) Reports from studies in humans and animals indicate that lactobacilli and bifidobacteria enhance the production of SIgA, apparently in a T cell-dependent manner. In a double-blind study of infants with a family history of atopy, babies who received a daily dose of a probiotic (Lactobacillus GG strain) for the first six months of life had 50 percent less atopic dermatitis at age two than did babies who received a placebo. Allergy prevalence still differed between study groups four years later. It's unclear whether this remarkable result is the product of reinforced barrier function from SIgA, enhanced oral tolerance or both.
Microorganisms existed billions of years before the first immune systems. Rather than waging war on them, our immune systems evolved a mutually beneficial partnership (mutualism) with certain bacterial strains that would compete for resources, in the environment of the gut or on other body surfaces, with more harmful microbes. An average adult carries 1014—100 million million—bacteria in his gut, or about 10 times more bacterial cells than there are human cells in the body. Our mechanisms of defense are shaped by this mutualism.
According to the original hygiene hypothesis, reduced or aberrant microbial exposure early in infancy doesn't provide enough stimulation to the so-called helper T cell type 1 (Th1). As a result, Th1 cells don't sufficiently antagonize the other type of helper T cell, Th2. Without this suppression, Th2 cells release cytokines that induce B cells to produce too much IgE, leading to atopy. Thus, the right commensal microbiota promotes mucosal homeostasis by helping to shift the newborn's immune system from a state dominated by Th2 signals (the allergy track) to one in which the cytokine profile is more balanced.
The extended hygiene hypothesis postulates that Treg cells, which have identifying proteins called CD25 receptors on their surfaces, are an important part of the homeostatic mechanism. Treg cells limit Th1 and Th2 cells when they act as proinflammatory effector T cells, thereby avoiding inflammation and tissue damage. Treg cells also suppress immune responses indirectly, either by reducing APC function or by secreting suppressive cytokines.
The window for fine-tuning a baby's mucosal immune system is relatively narrow, starting when the infant is colonized with vaginal and intestinal bacteria from the mother's birth canal. In healthy individuals, this initial exposure shifts the Th2-skewed cytokine profile of the newborn toward a Th1 profile, a sign of immunological maturation. But in atopic children, cytokines from Th2 cells continue to predominate, increasing the output of IgE, which predisposes the newborn to later allergy. Fortunately, the system retains some plasticity. Infants may be able to correct their ratio of Th2 to Th1 responses, and most children with overt food allergy outgrow it. (Some reactions, such as peanut sensitivity, are more likely to persist.) As alluded to above, there is hope for the future in that Treg cells can be stimulated intentionally through bacterial or parasitic products.
Clearly some stimuli are better than others at balancing the immune system. Several studies have reported that atopic infants have more of the intestinal bacterium Clostridium and less bifidobacteria in their stools than non-atopic controls. Similarly, another study found that children in Sweden tended to have more clostridia as well as allergies, whereas an age-matched group in Estonia had fewer allergies and high levels of lactobacilli and eubacteria. This research raises the possibility that various feeding and treatment regimens (particularly antibiotics) could exert long-term effects on the developing immune system through the composition of the gut microbiota. Certainly the possibility of promoting immune homeostasis through probiotic adjustment of the gut microbiota deserves further research.
Four out of five children who are allergic to cow's milk outgrow the problem before school age, which makes this disorder a good model for exploring the complexities of oral tolerance. Experiments in mice suggest that oral tolerance is brought about mainly by the actions of CD25+ Treg cells, although other mechanisms may also be involved.
In a recent study, our research team at the University of Oslo looked at a group of children who were all initially allergic to cow's milk. After a two-month dairy-free period, we gave the children cow's milk for up to one week. The challenged kids who then had outgrown the allergy (13 of 21) showed numerically more, and functionally better, Treg cells in peripheral blood than did children who remained allergic. And when we studied effector T cells from the children in vitro, those from the first group did not react as strongly to cow's-milk protein as did cells taken from children whose allergy persisted. However, in the same blood samples, removing the CD25+ cells (including Treg cells) caused a five-fold increase in the immune response to milk protein, which suggested to us that this subset of T cells had contributed to the developed tolerance to cow's-milk antigens.
Our study provided the first human data that link the induction of oral tolerance to the development of CD25+ Treg cells. This insight could prove useful as a diagnostic tool, and Treg cells might someday be candidates for preventing or treating allergy.
Blood Will Tell
One common tool for studying neonatal immune responses is blood taken from the umbilical cord at birth. The blood cells are from the baby rather than the mother, and such cord blood mononuclear cells, or CBMCs, can be studied in culture as a proxy for the fetal immune system.
We wanted to examine how exposure to LPS from bacteria during early antigen encounters might influence the responsiveness of neonatal T and B cells, including the activation of Treg cells by a food antigen. We also wondered whether it would be possible to use this measure to distinguish neonates with a high risk of allergy (because of family history) from controls with no hereditary risk. In fact, the stimulation with cow's-milk protein did cause greater (less controlled) proliferation of CBMCs from infants predisposed to atopy, suggesting that this test might predict later allergies. Various subsets of T cells, as determined by their immunological cell-surface markers, were also distinct between groups. After stimulation with a combination of milk antigen and LPS, the cells from babies with a family history of atopy expressed less of the markers overall, a trait that implied delayed development of a balanced immune system. The induction of Treg cells was also significantly impaired.
These data support the idea that induction of immunity should normally be modulated very early in life under the influence of genes and microbes. However, the CBMC model can only reveal small pieces of the mechanistic puzzle. Immunological events in the gut are much more complex, and mucosal homeostasis probably involves a multitude of processes.
Zooming out from all this complexity, the phenomenon of oral tolerance rests on a few primary processes: SIgA antibodies, the barrier function of the gut epithelium, the timing and dose of inoculation with commensal bacteria, and family history. These variables are interdependent, and no single factor predominates in maintaining mucosal homeostasis. There is no single cause of food allergy.
From an evolutionary perspective, intolerance to certain dietary antigens is not too surprising. It has not been long since human beings began growing and preparing their food rather than hunting and gathering it. And evolution is slow when the undesirable phenotype is so seldom deadly.
We must also keep in mind that the current epidemic of allergy in industrialized countries is a small price to pay for the remarkable reduction of infant mortality provided by the elimination of pathogens through improved hygiene. Having too few microbes in our immediate environment seems to be problematic, but having many pathogens is far, far worse. Nevertheless, the pace of research raises hope that future therapies will compensate for the missing good microbes needed to develop homeostasis of mucosal immunity.
- Benn, C. S., J. Wohlfahrt, P. Aaby, T. Westergaard, E. Benfeldt, K. F. Michaelsen, B. Björksten and M. Melbye. 2004. Breastfeeding and risk of atopic dermatitis, by parental history of allergy, during the first 18 months of life. American Journal of Epidemiology 160:217-223.
- Bischoff, S., and S. E. Crowe. 2005. Gastrointestinal food allergy: New insights into pathophysiology and clinical perspectives. Gastroenterology 128:1089-1113.
- Brandtzaeg, P. 2002. Current understanding of gastrointestinal immunoregulation and its relation to food allergy. Annals of the New York Academy of Sciences 964:13-45.
- Brandtzaeg, P. 2002. Role of local immunity and breast-feeding in mucosal homeostasis and defence against infections. In Nutrition and Immune Function, eds P. C. Calder, C. J. Field and H. S. Gill, 273-320. No. 1 of Frontiers in Nutritional Science. Oxford, UK: CABI Publishing.
- Brandtzaeg, P. 2006.The changing immunological paradigm in coeliac disease. Immunology Letters 105:127-139.
- Eggesbø, M., G. Botten, H. Stigum, P. Nafstad and P. Magnus. 2003. Is delivery by cesarean section a risk factor for food allergy? Journal of Allergy and Clinical Immunology 112:420-426.
- Guarner, F., R. Bourdet-Sicard, P. Brandtzaeg, H. S. Gill, P. McGuirk, W. van Eden, J. Versalovic, J. V. Weinstock and G. A. W. Rook. 2006. Mechanisms of disease: The hygiene hypothesis revisited. Nature Clinical Practice Gastroenterology & Hepatology 3:275-284.
- Haddeland, U., P. Brandtzaeg and B. Nakstad. In press. Maternal allergy influences the proliferation of neonatal T cells expressing CCR4, CXCR5 or CD103. Clinical & Exp er imental Allergy.
- Haddeland, U., A. B. Karstensen, L. Farkas, K. L. Bø, J. Pirhonen, M. Karlsson, W. Kvåvik, P. Brandtzaeg and B. Nakstad. 2005. Putative regulatory T cells are impaired in cord blood from neonates with hereditary allergy risk. Pediatric Allergy and Immunology 16:104-112.
- Holt, P. G., and C. A. Jones. 2000. The development of the immune system during pregnancy and early life. Allergy 55:688-697.
- Ivarsson, A., O. Hernell, H. Stenlund and L. A. Persson. 2002. Breast-feeding protects against celiac disease. American Journal of Clinical Nutrition 75:914-921.
- Karlsson, M., M. R. Karlsson, J. Rugtveit and P. Brandtzaeg. 2004. Allergen-responsive CD4+CD25+ regulatory T cells in children who have outgrown cow's milk allergy. Journal of Experimental Medicine199:1679-1688.
- Rakoff-Nahoum, S., J. Paglino, F. Eslami-Varzaneh, S. Edberg and R. Medzhitov. 2004. Recognition of commensal microflora by toll-like receptors is required for intestinal homeostasis. Cell 118:229-241.
- Rautava, S., O. Ruuskanen, A. Ouwehand, S. Salminen and E. Isolauri. 2004. The hygiene hypothesis of atopic disease—an extended version. Journal of Pediatric Gastroenterology and Nutrition 38:378-388.
- Sicherer, S. H., A. Muñoz-Furlong and H. A. Sampson. 2003. Prevalence of peanut and tree nut allergy in the United States determined by means of a random digit dial telephone survey: A 5-year follow-up study. Journal of Allergy and Clinical Immunology 112:1203-1207.
- Sicherer, S. H., and H. A. Sampson. 2006. 9. Food allergy. Journal of Allergy and Clinical Immunology 117(2 Supplemental Mini-Primer): S470-475.
- van Odijk, J., I. Kull, M. P. Borres, P. Brandtzaeg, U. Edberg, L. Å. Hanson, A. Høst, M. Kuitunen, S. F. Olsen, S. Skerfving, J. Sundell and S. Wille. 2003. Breastfeeding and allergic disease: A multidisciplinary review of the literature (1966-2001) on the mode of early feeding in infancy and its impact on later atopic manifestations. Allergy 58:833-843.