Oligosaccharides in Human Milk and Bacterial Colonization

Abstract

In the first year of life, a stable population of microflora is established in the human gut as a complex ecology that includes the intestinal mucosa and more than 400 interdependent species of bacteria. During this period, a child is vulnerable to enteric and other diseases, but breast-fed infants have a lower rate of morbidity and mortality than do artificially fed infants. The lower incidence of disease in infants while they are breast-feeding has been attributed to many causes: less exposure to contaminated water, better nutritional status, and psychological factors. Although these factors probably contribute, the most direct determinant of reduced disease incidence among the breast-fed seems to be factors in the milk itself. This protection is thought to result from two major kinds of components in human milk: those that provide a prebiotic effect that enhances colonization by Bifidobacterium bifidum, and others that inhibit the colonization and growth of pathogens during this vulnerable period. Three major classes of human milk components are known to inhibit pathogens: secretory antibodies; multifunctional agents, such as fatty acids and lactoferrin; and homologues of host cell surface glycoconjugates. Evidence is mounting that these homologues, consisting of glycoconjugates and oligosaccharides, have a major role. Glycoconjugates are molecules that have conjugated carbohydrate moieties (glycoproteins, glycolipids, glycosaminoglycans, and mucins); milk oligosaccharides have carbohydrate moieties conjugated to lactose. These milk carbohydrate moieties often share structural motifs with the cell surface glycoconjugates that pathogens adhere to when gaining entry to host target cells. Indeed, specific human milk oligosaccharides and glycoconjugates have been shown to block binding of specific pathogens to host cell glycoconjugate receptors. This article presents three examples: 1) A fucosylated oligosaccharide from human milk inhibits the pathogenicity of stable toxin of Escherichia coli in vivo, and the blocking effect has been demonstrated in vitro. 2) Fucosylated structures also inhibit binding by Campylobacter jejuni to host cell receptors and limit its colonization in vivo. 3) A glycoprotein, lactadherin, inhibits rotavirus in vitro and in vivo. High concentrations of lactadherin in human milk are associated with reduced incidence of symptoms of rotavirus infection in nursing infants. These findings, along with other examples of pathogen inhibition by milk glycoconjugates, suggest that this is a general mechanism of protection by human milk. Furthermore, human milk oligosaccharides survive transit through the intestine of the breast-feeding infant, implying that they are available to protect against gastroenteritis over the entire length of the intestinal tract. However, pronounced individual variation in the expression of oligosaccharides in human milk suggests heterogeneity in the ability of individual human milks to inhibit specific pathogens. Oral supplementation of vulnerable infants with glycoconjugates, oligosaccharides, or their homologues may be feasible. The milks of other species also contain homologous oligosaccharides and glycoconjugates, suggesting that this mechanism of protection in early life may be common to many mammals. The presence of a large spectrum of oligosaccharides and glycoconjugates in human milk may represent a major mechanism whereby breast-feeding protects infants from enteric and other pathogens during the period when the microflora of the gut are becoming established. BREAST-FEEDING PROTECTS INFANTS AGAINST DISEASES Evidence that breast-feeding protects infants against diseases during their vulnerable first year of life was first convincingly reported from the Massachusetts General Hospital in 1935 (1). In this study of 20,000 patients, breast-feeding was associated with a lower incidence of morbidity and mortality, especially of enteric disease, otitis media, and respiratory infection. Although subsequent studies yielded mixed results regarding the efficacy of breast-feeding against these diseases, more recent studies using relatively sophisticated experimental design and data analysis have yielded findings that mostly support the earlier report. A noteworthy example is a study performed in Lahore, Pakistan (2). The population was divided into rural poor, urban poor, and urban upper middle class; the efficacy of breast-feeding against enteric disease was determined by the same type of analysis used for testing the efficacy of drugs (Fig. 1). Breast-feeding was effective in preventing disease among all three populations, with the protection most pronounced in the first 6, and especially in the first 3, months of life. This protection was most pronounced among the poor but was also clearly demonstrated in the upper middle class, whose socioeconomic status and level of hygiene is closer to that of many western societies. A prospective study in a Scottish population on the effect of breast-feeding on morbidity and mortality due to enteric diseases (3) showed that even in developed countries breast-feeding during the first 3 months of life can be demonstrated to confer protection against gastrointestinal illness. Curiously, in this study the benefits persisted beyond the period of breast-feeding itself. This phenomenon is explicable if breast-feeding contributes to establishing a healthy ecology of the gut, the ramifications of which extend beyond weaning.FIG. 1: Efficacy of breast-feeding in preventing enteric disease was high in both rural (village and mud hut) and urban (slum) poor, especially during the first 6 months of lactation, and the protection persisted for the 2-year duration of this study. Note that breast-feeding was also efficacious in preventing enteric disease, albeit to a lesser extent, in the upper middle class, a population whose living conditions are closer to those of developed countries. (Reprinted, with permission, from Kluwer Academic/Plenum Publishers. Hansen et al. Adv Exp Med Biol 1991;310:1-15).COLONIZATION BY MICROFLORA The adult human intestine represents a complex ecosystem of more than 400 species of bacteria, interdependent among themselves and with the human gut per se. The human intestinal microflora typically contains more cells than the entire number of cells in the human body. These intestinal bacteria have many functions, such as digesting food components that are indigestible by human intestinal enzymes, and can be a source of important factors such as vitamins. Moreover, a stable, healthy intestinal microflora is thought to contribute to overall health by excluding foreign, potentially harmful bacteria (4). The formation of a stable, interdependent ecosystem of intestinal microflora during the first year of life requires selective acquisition from among the many bacteria in the environment. Important environmental sources of bacterial exposure include the mother and other individuals with whom the infant has intimate contact (5). The colonization of the gut by new bacteria is influenced by the current resident bacteria. Moreover, colonization by any new strain of bacteria is influenced by intestinal epithelial cells, which in turn are affected by the bacteria that colonize the gut, as is discussed in greater detail in the review in this issue by Walker. Diet greatly influences the selection of intestinal microflora during this first year of life. For example, in the gut of the breast-fed infant, bifidobacteria predominate and enterobacteria are present (6), whereas in the artificially fed infant, bifidobacteria, bacteroides, clostridia, and streptococci are dominant (7). Bacteroides dominate the adult flora (8). Human milk, normally the principal food during this first year of life, contains factors that stimulate the growth of commensal bacteria. Prebiotics (9) enhance colonization by specific bacteria; for example, bifidus factor stimulates growth of Bifidobacterium bifidum (formerly called Lactobacillus bifidum) (10,11). Lactose affects colonization directly, by selecting for bacteria able to use this unique sugar, and indirectly, when fermented to acetate, making the gut acidic and stimulating the growth of some bacteria while inhibiting colonization by many pathogenic organisms. Human milk also contains three types of components that inhibit colonization by harmful bacteria. One is the secretory immunoglobulin A (sIgA), which is produced by the immune system of the mother against enteric pathogens to which she has a history of exposure (12). Another type is the multifunctional agents that serve as nutrients while protecting against a broad spectrum of pathogens. Examples are fatty acids that, when released from the triglycerides of milk in the stomach of the infant (and in expressed human milk that has been stored) destroy envelope viruses (13) and also act against other pathogens such as Giardia(14,15). Lactoferrin, the major protein in human milk, is active against a broad spectrum of pathogenic bacteria, and its digestion products include lactoferricin, a porin that is toxic to many strains of bacteria (16). The third class of milk components that are active against pathogens is molecules containing structural carbohydrates, the glycoconjugates and oligosaccharides. The oligosaccharides are complex carbohydrate structures that are generally attached to lactose (17). The glycoconjugates contain complex carbohydrate structures that are attached to proteins, lipids, and other structures (e.g., glycoproteins, glycolipids, mucins, and glycosaminoglycans). The carbohydrate portion of these molecules is largely indigestible and passes through the intestinal tract of the infant mostly intact. These carbohydrate structures potentially act as decoys to which pathogens bind. It is this fraction of human milk that will be the focus of this review. HUMAN MILK OLIGOSACCHARIDES Oligosaccharides are, collectively, the third largest solid constituent in human milk, after lactose and fat. Human milk oligosaccharides consist of lactose at the reducing end, with a carbohydrate core that often contains a fucose or a sialic acid at the nonreducing end (17). When they were initially discovered and studied, these human milk oligosaccharides were thought to be the accidental consequence of the high levels of glycosyltransferases in mammary epithelial cells (required for formation of the abundant milk glycolipids and glycoproteins) in proximity to the high levels of lactose unique to these cells. However, the idea that these oligosaccharides may have a biologic purpose is supported by two lines of evidence in relatively recent investigations. First, it was discovered that the approximately 100 milk oligosaccharides that had been isolated and characterized represent only a very small portion of the total that remain to be characterized. The potential for many thousands of fucosylated oligosaccharides in human milk was suggested when the total human milk oligosaccharide fraction was analyzed by time-of-flight mass spectrometry: Evidence was found for the existence of oligosaccharides with up to 32 sugars and up to 15 fucoses (18). The huge potential for structural isomers suggests that there may be tens of thousands, or perhaps even hundreds of thousands, of potential human milk oligosaccharide structures. Such a large number of possible structures greatly increases the probability that some are biologically active. A second, independent line of investigation has identified several biologic functions attributable to the milk oligosaccharides. Oligosaccharides have been found in human milk that inhibit the adherence of Streptococcus pneumoniae to target cells in vitro (19). Human milk oligosaccharides also inhibit binding by enteropathogenic E. coli(20), binding and infection by invasive strains of C. jejuni(21), and the toxicity of stable toxin of E. coli(22). The findings from both lines of investigation, taken together, suggest that many more biologically active oligosaccharides remain to be discovered. MAMMALIAN CELL SURFACE GLYCOCONJUGATES ARE PATHOGEN RECEPTORS Mammalian cell membranes contain glycoproteins and glycolipids, and the membrane structure is such that the oligosaccharide moieties of these cell surface glycoconjugates protrude into the extracellular region (Fig. 2). These complex carbohydrate moieties participate in cell-cell interactions and can act as receptors for humoral mediators such as hormones. They can also act as receptors for antibodies, toxins, and pathogens. They are the usual sites where viruses and bacteria adhere to the host target cell to initiate the pathogenic process.FIG. 2: Cell surface interactions of glycoconjugates. Glycoconjugates (glycolipids, glycoproteins, and others) can participate in cell-cell interactions (top). These glycoconjugates can also bind to humoral mediators and can participate in transmembrane signaling (left). These same cell surface glycoconjugates can be used by pathogens (viruses, bacteria, and toxins) for recognizing, binding to, and gaining entry to host cells (right) as the essential first step in pathogenesis.The glycosyltransferases that make human milk oligosaccharides are similar to the glycosyltransferases that synthesize the oligosaccharide moieties of mammalian cell surface receptors. Thus, the milk oligosaccharides are likely to have some structural motifs similar to those found at the cell surface. These human milk oligosaccharides can therefore act as soluble homologues to the cell surface receptors that are the targets of specific pathogens. Serving as decoys, they bind to the pathogens and inhibit their ability to bind to the host cells. Because pathogen binding to the host is an essential first step in pathogenesis, specific milk oligosaccharides can protect against disease. Three examples will be discussed in which specific human milk components inhibit specific pathogens; two of the components are oligosaccharides, and one is a glycoprotein. Stable Toxin of Escherichia coli Infants born in areas where enterotoxigenic E. coli are endemic are at high risk for diarrhea caused by these organisms. E. coli produce both a labile toxin, which resembles cholera toxin in that it binds to ganglioside in the host cell surface, and a stable toxin, which binds to the extracellular domain of guanylate cyclase in intestinal mucosal cells, initiating a different cascade of pathogenic events resulting in secretory diarrhea. Human milk significantly reduced mortality rates in suckling mice given stable toxin of E. coli(22). The protective activity was found to reside solely in the neutral oligosaccharide fraction of pooled human milk and was active at its native concentration. When this neutral oligosaccharide fraction was passed over a Ulex europaeus lectin-affinity column, the activity was found exclusively in the fucosylated oligosaccharide fraction that bound to Ulex(23). When the active fucosylated fraction was separated into its component peaks on an amino column by high-performance liquid chromatography (HPLC), only one of the minor peaks consistently inhibited the diarrhea caused by stable toxin in vivo. When this peak was separated into its components by reversed-phase HPLC, the activity was found exclusively in one of its minor peaks. This peak was isolated to homogeneity and, at 2.5 times its nominal concentration in human milk (to take into account losses in its purification), this fraction inhibited stable toxin in vivo as effectively as the milk from which it was isolated (Table 1). This fraction is present in human milk at a concentration of approximately 40 parts per billion. Its high level of activity at such a low concentration suggests high degrees of specificity and avidity for its receptor.TABLE 1: Total and fucosylated milk oligosaccharides inhibit 125I-ST binding to T84 membranes An in vitro system commonly used to study stable toxin was used to determine the mechanism by which the fucosyloligosaccharide fraction may protect against diarrhea induced by stable toxin. T84 cells, an immortal line descended from intestinal mucosa, were grown under conditions that induce the expression of many characteristics of intestinal mucosal epithelial cells: tight junctions, polarity, expression of microvilli on the mucosal surface, and expression of guanylate cyclase among these microvilli. Guanylate cyclase is a transmembrane protein with an extracellular domain that binds to stable toxin, activating the intracellular domain, and thus increasing the conversion of guanosine triphosphate (GTP) to cyclic guanosine monophosphate (cGMP; Fig. 3). Elevated levels of cGMP result in increased efflux of chloride ions from the cell, accompanied by an efflux of sodium ion and water. This movement of salts and water to the lumen of the gut is the secretory diarrhea induced by stable toxin. The inhibitory fucosyloligosaccharide binds to a site on the extracellular domain of the guanylate cyclase of T84 cells, blocking binding by stable toxin (24). This seems to be the mechanism whereby this oligosaccharide prevents the stable toxin from inducing secretory diarrhea.FIG. 3: The mechanism whereby a milk oligosaccharide inhibits the ability of stable toxin to cause secretory diarrhea. When stable toxin binds to the extracellular domain of guanylate cyclase, the intracellular enzymes are activated, causing an increase in intracellular cyclic guanosine monophosphate (cGMP) and loss of chloride from the cell. The efflux of chloride along with its counter ion, sodium, and water produce secretory diarrhea. By preventing stable toxin from binding to its extracellular receptor, homeostasis is preserved and secretory diarrhea avoided, despite the presence of stable toxin.Campylobacter Campylobacter jejuni, a common pathogen of birds, also infects humans. In areas where domestic fowl live in proximity to humans, children who come in contact with the birds or their droppings are at high risk for Campylobacter diarrhea early in life. Campylobacter infection is often studied in HEp-2 (human epithelial-derived) cells in vitro. Human milk inhibits the ability of C. jejuni to infect HEp-2 cells. This activity resides in the oligosaccharide fraction of human milk and particularly in the fucosylated oligosaccharides that bind to a U. europaeus affinity column (21). At the concentrations found in human milk, these oligosaccharides specifically inhibit the adhesion of pathogenic invasive strains of C. jejuni to HEp-2 cells, but not the adhesion of nonpathogenic strains. A murine model was used to test whether the inhibition of C. jejuni binding in vitro translates into an inhibition of infection in vivo. Groups of mice were given low- (104Campylobacter) or high (108Campylobacter) inocula, either in the absence or the presence of the active human milk oligosaccharide fraction. Figure 4 illustrates that, at both concentrations, the presence of the oligosaccharide fraction resulted in several logs less colonization than in the saline control groups (25). Thus, milk fucosylated oligosaccharides not only inhibited binding in vitro but prevented infection in vivo.FIG. 4: Human milk oligosaccharides inhibit the ability of Campylobacter to colonize the intestines of mice. With both low- (104 bacteria per animal) and high-concentration inocula (108 bacteria per animal), the ability of Campylobacter to colonize the gut is inhibited by the presence of human milk oligosaccharides at one half the concentration found in human milk.These laboratory findings provide two clear examples of human milk oligosaccharides that inhibit human pathogens. The relevance of these human milk glycoconjugates to the expression of disease in a breast-feeding human population has not been tested directly for either Campylobacter or stable toxin-positive E. coli. However, this issue was addressed in a third example of a human milk glycoconjugate that inhibits a specific pathogen: lactadherin inhibition of rotavirus. Lactadherin and Rotavirus Rotavirus is the most common single cause of childhood diarrhea, especially in the developed world, where infection by bacteria and other parasites is much less common. The ability of rotavirus to infect MA104 (green monkey kidney) cells in vitro is inhibited by human milk. Most of this inhibitory activity resides in the protein fraction and has been isolated to a 46-kDa glycoprotein named lactadherin. The inhibition is general for all major strains of human rotavirus and is dose dependent. Removal of sialic acid from this glycoprotein eliminates its inhibitory activity, suggesting that a terminal sialic acid attached to a specific carbohydrate moiety of this glycoprotein is responsible for the inhibitory activity. The lactadherin binds to rotavirus, thereby blocking its ability to bind to its host target cell (26). This ability to inhibit rotavirus infection and its consequent diarrhea has also been demonstrated in vivo in a murine model. The relationship between the lactadherin levels in mothers' milk and presence of rotaviral diarrhea in their infants was determined prospectively in a population of 200 mother-infant nursing pairs. Stools of the infants were regularly cultured to detect rotavirus, samples of milk were regularly obtained for lactadherin analysis, and the infants were checked daily for symptoms of diarrhea. Of the infants who were infected with rotavirus, those with diarrhea were consuming breast milk with significantly lower levels of lactadherin than those who were asymptomatic (27)(Fig. 5). There was no significant relationship between antirotaviral sIgA levels in milk and rotaviral diarrhea among infected children. These data suggest that the expression of lactadherin in the milk varies among nursing women, and that the level of lactadherin expressed is significantly related to the presence of rotaviral diarrhea in breast-fed infants.FIG. 5: Lactadherin concentrations in milk are associated with less rotaviral disease in breast-fed infants. In breast-feeding infants infected with rotavirus, those infants whose mothers' milk was high in lactadherin were less likely to develop symptoms of rotavirus (i.e., diarrhea) than were those infants whose mothers' milk contained low levels of lactadherin. This demonstrates that lactadherin, which binds to rotavirus and inhibits its ability to infect cells in vitro and to cause diarrhea in animals, also inhibits the production of rotavirus-associated diarrhea in humans.It seems likely that large amounts of lactadherin bind to most of the rotaviral receptors for host cell surface glycoconjugates, blocking the first step in pathogenesis. However, when only a limited amount of lactadherin is available, as in mothers who produce low levels in their milk, or when lactadherin is absent, as in infants who are artificially fed, the virus still has adequate free receptors that are able to bind to intestinal mucosal cells and initiate infections. RELATIVE CONTRIBUTIONS OF ANTIBODIES AND GLYCOCONJUGATES It is noteworthy that in the study of lactadherin and rotavirus (27) the relationship between lactadherin and disease was significant, whereas the relationship between rotaviral sIgA and disease was not. This surprising finding implies that protection by glycoconjugates of human milk may be at least as important as protection by antibodies. Although antibodies in human milk were discovered early and are widely recognized as protective agents (12), they seem to represent only one mechanism whereby human milk protects infants. The glycoconjugates and oligosaccharides confer comparable protection at the concentrations found in typical human milk, probably because the incredibly large number of possible structures of these milk fractions allows great specificity and potential ability to protect host cells from a wide array of pathogens. It is reasonable to postulate that human milk offers multiple mechanisms for the protection of infants. The immune systems of infants are premature. Their microflora is just developing and is unstable, and other mechanisms of defense against pathogens are not fully developed. Thus, human milk may be extremely important to the defense of infants from disease and may therefore have evolved multiple, redundant systems conferring protection (28). VARIABILITY OF MILK OLIGOSACCHARIDES The ability of individuals to express glycoconjugates and oligosaccharides varies, in human milk as in other tissues and fluids. Environmental factors, such as diet, have not been observed to influence oligosaccharide expression in human milk, but oligosaccharide expression is heterogeneous, according to genetic variation among individual mothers (29). The two simplest fucosyloligo-saccharides, 2′-fucosyllactose and 3-fucosyllactose are shown in Figure 6. The fucosyl α1,2 linkage seen in 2′-fucosyllactose is synthesized by a different fucosyltransferase than the fucosyl α1,3 linkage seen in 3-fucosyllactose. Similar to practically all human milk oligosaccharides, these fucosyloligosaccharides contain lactose at the reducing end. Similar features can be seen in the lacto- and fucopentaoses, shown in Figure 7. These also have lactose at the reducing end, but the fucose at the nonreducing end is attached by an α1,2 linkage, α1,3 linkage, or α1,4 linkage. Again, the α1,2 linkage is synthesized by one type of fucosyltransferase, whereas the remaining linkages are synthesized by at least one different fucosyltransferase. Because the expression of fucosyltransferases is known to be heterogeneous, fucose-containing glycoconjugates in human milk and elsewhere are heterogeneous as well. This variation of fucosylated oligosaccharides in populations of human milk has yet to be fully defined.FIG. 6: Structures of the two simplest fucosyloligosaccharides in human milk. In these structures, fucose is attached to lactose either by an α1,2 linkage (top) or an α1,3 linkage (bottom). Each of these linkages is synthesized by a different fucosyltransferase.FIG. 7: The structures of three fucopentaoses of human milk. These structures illustrate several features common to fucosyloligosaccharides: They contain lactose (galactose α1,4-glucose) at the reducing end (right) and contain fucose at the nonreducing end of the molecule in either an α1,2 linkage (top), an α1,4 linkage (middle), or an α1,3 linkage (bottom). The α1,2 linkage is synthesized by transferases whose regulation is independent from transferases that synthesize the α1,3 or α1,4 linkage.Variation of oligosaccharides among milk samples from different mothers was measured by an HPLC technique that visualizes oligosaccharides made of eight sugars or fewer as sharp, distinct peaks. This technique accounts for more than 95%, by mass, of human milk oligosaccharides (30). Wide variation was found among mothers' expression of glycoconjugates in their milks. In the majority the fucosyl α1,2 linkages are prevalent, but in some the chromatographic patterns displayed little evidence of any α1,2-linked fucosyloligosaccharides, leaving primarily the α1,3- and α1,4-linked compounds. This suggests differential expression of protective oligosaccharides among individual mothers, depending on their genetic complement of fucosyltransferases. The expression of oligosaccharides over the course of lactation was consistent within the milk samples from a given mother. The overall concentration of oligosaccharides decreased over the course of lactation. Women whose milk oligosaccharide patterns contained a predominance of α1,2-linked fucosyloligosaccharides continued to produce such milk for the first 6 months of lactation; similarly, those whose milk had no α1,2-linked fucosyloligosaccharides produced milk that was deficient in these oligosaccharides during the first 6 months of lactation. However, between 6 months and 1 year of lactation, the milks from both classes of mothers tended to contain both types of oligosaccharides, suggesting some differences in the mechanism of oligosaccharide synthesis late in lactation. These patterns are consistent with the observation that human milk seems to offer its strongest protection against disease during the first 6 months of lactation. These data also suggest a differential ability to protect against specific pathogens by breast-feeding mothers whose genetic expression of fucosyltransferases differ. This phenomenon is a topic of current investigation. FATE OF MILK OLIGOSACCHARIDES AFTER INGESTION The oligosaccharides of human milk represent a major investment of calories by the mother. One might expect a mechanism for their digestion so that this energy input would be fully used. On the other hand, if the primary function of these oligosaccharides is protection from disease, they would be expected to remain intact as they pass through the alimentary canal to protect even distal regions of the gut. To investigate the fate of the milk oligosaccharides after ingestion by breast-fed infants, the oligosaccharide profiles of the urine and feces of the breast-fed infants were compared with those of the milk from their mothers (31). The pattern of oligosaccharides in the urine and feces of the breast-fed infants resembles that in their mothers' milk, suggesting that their origin is primarily human milk. The urine and feces of breast-fed infants were also compared to the urine and feces of matched formula-fed infants. Oligosaccharides in the urine and feces of artificially fed infants have a different pattern from breast milk and from the urinary and fecal patterns of breast-fed infants. The oligosaccharides in feces of artificially fed infants probably arise from endogenous turnover of intestinal epithelial cells and from bacteria. Urinary oligosaccharides in these infants may arise from turnover of renal tubular cells, or, possibly, from the absorption of intestinal oligosaccharides followed by their excretion into the urine. The quantities of oligosaccharides in the urine and feces of breast-fed infants are so much higher than those in formula-fed infants (Table 2) that they may mask any endogenous oligosaccharide excretion. The concentration of milk oligosaccharides in the feces of breast-fed infants is, as anticipated, much higher than in the milk they are consuming, because the other milk components are digested and absorbed as they pass through the gut, whereas the oligosaccharides remain and increase in relative concentration.TABLE 2: Oligosaccharide concentrations in milk, urine, and feces of breast-fed and formula-fed infants An estimate of intake and excretion of oligosaccharides, based on typical values for the volume of milk consumed and for urinary and fecal output, suggests that the bulk of oligosaccharides consumed by breast-fed infants is excreted in the feces intact, whereas a small quantity is absorbed and excreted in the urine. It is interesting to speculate that the small amount of intact oligosaccharides in the urine could offer some protection against urinary tract infections (32). The increased concentration of oligosaccharides at the end of the alimentary canal demonstrates that they are present throughout the intestine in sufficient concentrations to protect against pathogens. These results also favor the hypothesis that the function of the oligosaccharides in milk is primarily protective rather than nutritional. HUMAN MILK GLYCOCONJUGATES PROTECT INFANTS FROM DISEASE The three examples of protective oligosaccharides and glycoconjugates in human milk that have been discussed are representative of many classes of glycoconjugates found in human milk, including oligosaccharides, glycoproteins, glycopeptides, mucins, glycosaminoglycans, and glycolipids (Table 3). These inhibitors all have in common a complex carbohydrate moiety as part of their structure and the ability to inhibit binding of pathogen to its host cell surface receptor. The growing list of glycoconjugates in human milk that protect nursing infants from specific pathogens is strong evidence that the glycoconjugates in the aggregate may represent a major mechanism of human milk in protecting the infant from disease.TABLE 3: Human milk inhibitory glycoconjugates The function of human milk components, delineated in Table 4, includes three tiers of defense in addition to nutrients and immunomodulatory agents. These three tiers include the immunoglobulins (i.e., sIgA), produced by the mother against epitopes of pathogens to which she has a history of exposure and secreted into the milk. A second line of defense is the multifunctional agents, which may serve as nutrients but also have properties that can inhibit a broad range of pathogens. Glycoconjugates constitute the third line of defense.TABLE 4: Biological functions of human milk components These seemingly redundant tiers of defense are probably all important components of a synergistic system. For example, the glycoconjugates inhibit specific pathogens, whereas the multifunctional agents tend to have a broader activity. Both these constitutive elements of human milk could be seen as a preliminary line of defense during the period that elapses between the mother's exposure to a pathogen and the production of secretory antibodies against it in her milk. On the other hand, glycoconjugates could be the primary line of defense against enteric pathogens. The data relating human milk lactadherin levels and differential susceptibility to symptoms of rotavirus support this view. Antibodies could serve a secondary role, providing essential protection from disease if the genetic background of the mother precludes the synthesis of a functional enzyme needed to produce a specific milk component that inhibits a specific pathogen. It is also possible that the relative importance of these three tiers of protection differs for different pathogens. This multiplicity of agents in human milk that inhibit common diseases in infants may justify a shift in emphasis in the study of human milk. Rather than thinking of human milk as a food with some antibodies, perhaps it is more accurate to think of it as a collection of biologically active protective agents that also provides nutritional support. Thus, milk may protect the infant from harmful pathogens during the first year of life, before the immune system is fully mature and before a stable microflora has been established, while providing an environment that stimulates the growth of commensal bacteria. These considerations provide justification for public health policy geared toward promotion of breast-feeding in both developing and developed countries. Furthermore, the stability of the milk oligosaccharides in storage and during intestinal transit suggest that oral supplementation of the native oligosaccharides or their homologues may be feasible. Acknowledgments: Supported by grant HD13021 from the National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland, U.S.A.