Are All Human Milks Created Equal? Variation in Human Milk Oligosaccharides

The original observation that breast-feeding may protect infants from diarrheal diseases, and perhaps from respiratory and middle ear infections (1), is now well accepted. Although this effect has long been attributed to the immunoglobulins in milk, components other than the antibodies are now thought to provide important protection. Most of these protective factors are free oligosaccharides (complex sugars that are found almost exclusively in milk) or glycoconjugates. Human milk is especially rich in oligosaccharides, particularly those containing the sugar fucose as part of their structure. In laboratory studies, human milk oligosaccharides strongly inhibit binding to host cells by Campylobacter jejuni and Vibrio cholerae, inhibiting their pathogenicity in vitro and in vivo. They inhibit cell surface binding and diarrheagenic effects of stable toxin of Escherichia coli (ST) and adherence of enteropathogenic E. coli and Streptococcus pneumoniae to host cells. Their ability to inhibit other pathogens is under investigation. The inhibition of ST and the inhibitory activity against C. jejuni and V. cholerae have been attributed to specific subsets of fucosyloligosaccharides. This protection often results from homology of an oligosaccharide moiety with the cell surface receptor for the pathogen. These soluble homologues compete with the cell surface receptors, thereby blocking the pathogen's ability to bind to and infect the cell. A 1994 study in which matrix-assisted laser desorption–ionization mass spectrometry (MALDI-MS) was used indicated that thousands of fucosyloligosaccharides are present in human milk, rather than the hundred or so that had been identified previously (2). Such a high number of oligosaccharides greatly increases the probability of fucosyloligosaccharide inhibitors for other pathogens. The synthesis of specific milk oligosaccharides is determined by the presence in the mammary acinar cells of enzymes whose expression is under genetic control and varies among different populations. These or closely related enzymes also function in erythrocytes in the production of the cell surface glycolipids that determine blood group type and in other tissues, where they are involved in the synthesis of cell surface glycoconjugates that are receptors for pathogens. Heterogeneity in the expression of these receptors has been shown to underlie differential susceptibility to certain diseases. Because the glycoconjugate receptors and blood group types depend on the same or related enzymes, the receptors' presence or absence in an individual can be indirectly ascertained by determining blood group type. Thus, cholera susceptibility is related to the ABO blood group type, with O-type individuals several times more likely to be admitted to hospitals with the disease than AB individuals (3). Epithelial cells from individuals of the recessive p blood group type do not express a receptor for uropathogenic E. coli(4,5) and are resistant to urinary tract infections. Moreover, individuals who express less of the P blood group glycolipids in their erythrocytes more frequently develop hemolytic uremic syndrome (6). Secretor status (i.e., secretion or absence of certain fucosylated blood group antigens) influences susceptibility to recurrent urinary tract infections (7) and to Candida albicans infections. The health of an infant exposed to a given pathogen sometimes depends on her or his ability to produce ample cell surface receptors for the pathogen and also depends on her or his mother's inherited ability to produce sufficient milk oligosaccharide homologues for that receptor. For other pathogens, specific milk glycoconjugate and oligosaccharide moieties are protective by other mechanisms. Thus, the identification and measurement of specific human milk oligosaccharides and glycoconjugates can contribute greatly to the understanding of diseases of infants. Were variations in milk oligosaccharides predictable by ethnicity, prophylactic interventions could help infants in populations in which breast milk could not be expected to protect against one or another endemic disease. This possibility was encouraged by reports that the expression of human milk fucosylated oligosaccharides varies according to the mother's blood group type, secretor status, and Lewis blood group type (8,9). Variation of the acidic oligosaccharides was similarly linked to the ABO blood group status of the mother (9). Seen in the light of these findings, it was not surprising when milks from 50 German mothers revealed fucosyloligosaccharide patterns that seemed to fall into four groups, each of which was considered an expression of the four permutations of dominant and recessive secretor and Lewis genotypes (10). The method used, high-pH anion-exchange chromatography (HPAEC) of the neutral milk oligosaccharides, permitted measurement of 11 major, individual, fucosylated oligosaccharides from each milk sample (11). However, a more sensitive high-performance liquid chromatography (HPLC) technique (12) applied to milks from a Mexican population indicated that each measured fucosyloligosaccharide varies among individuals, not only in its presence or absence, but also in amounts (13). If the absolute amounts of oligosaccharides are taken into account, there seem to be many more than four groups of oligosaccharide patterns. Complicating the picture further are reports that concentrations of total or broadly categorized oligosaccharides vary over the course of lactation, diurnally (14) and during the first week (15) and first few months (16). The report by Erney et al. in this issue is the first to describe variability of human milk oligosaccharides in different countries. Because the frequencies of the Lewis (FucT-III) and secretor (FucT-II) genes are known to differ among populations, differences in fucosylated milk oligosaccharides should also be apparent. Ten sugars (seven of which are fucosylated) were studied in 549 samples from 435 different donors living in 10 different countries. The length of time since parturition was also noted. The results revealed some surprising population differences. One hundred percent of the Mexican mothers produced 2´-fucosyllactose, which has been considered the dominant oligosaccharide in human milk (17). However, only 46% of the Philippine mothers had milk with measurable amounts of this compound, and the percentage in the other countries ranged between these two values. When specific oligosaccharides were present, their amounts also varied by region and by time after birth at which the samples were obtained. This variation may be the consequence of heterozygosity of the secretor and Lewis gene, or of complex interrelationships between these and other fucosyltransferases. The results of this exploratory study raise interesting questions. If the Lewis and secretor genetic variation is a dominant feature underlying the variation seen in milks from different countries, genetically similar populations would be expected to have similar milk. For example, because the U.S. population has a preponderance of individuals of European ancestry (the ethnicity of sample donors is not specified in this report), European patterns would be expected to resemble those of the US more closely than is apparent in this data set. Also, different oligosaccharides whose ultimate synthetic step is catalyzed by the same enzyme would be expected to covary more strongly than is apparent. Several factors may account for some of the high variation among and within groups. Because much of the data come from one-time donations of milk, samples from different individuals and different times after birth are being compared. These procedures could allow sampling bias to contribute to some of the oligosaccharide differences seen during different periods of lactation. Also, the data are unbalanced in size of groups representing different areas and durations of lactation represented within groups. A particularly interesting result comes from the study's comparisons of milks grouped according to stage of lactation. 2´-Fucosyllactose tended to decrease late in lactation with a reciprocal increase in the concentration of 3-fucosyllactose, a result consistent with findings obtained with an entirely different method of analysis in a smaller, ethnically homogeneous population (18). This pattern may reflect changes in the complement of enzymes expressed for the synthesis of fucosyloligosaccharides late in lactation and may have ramifications for the ability of milk to protect against different pathogens at different stages of lactation. In conclusion, the data reported by Erney et al. clearly show that the classification of milk oligosaccharide patterns into three or four distinct groups related to blood type may not hold up when quantitative methods that measure individual oligosaccharides are used for studying highly diverse populations. At the same time, this study supports the potential for variation in the protective activity of human milk among diverse populations.