Introduction
Human milk oligosaccharides (HMO) vary in structure, size and composition and arise from the sequential addition of monosaccharides, such as L-fucose (Fuc), D-N- Acetylglucosamine, (GlcNAc), D-Galactose (Gal) and N-Acetyl- Neuraminic Acid (NeuAc) to lactose (Galβ1-4Glc). The smallest oligosaccharides are generated either when fucose is added to lactose, thus generating the trisaccharides termed fucosyllactoses (2′FL; Fucα1-2 Galβ1-4Glc and 3FL;Galβ1-4[Fucα1-3]Glc), or when the acidic monosaccharide NeuAc is added to lactose, generating the sialyllactoses (3′SL; NeuAcα1-3Galβ1-4Glc and 6′SL; NeuAcα1-6Galβ1-4Glc). The relatively high concentration of lactose in milk obscured until recently the presence of the oligosaccharide fraction, which was initially described with the name “gynolactose” and was later analysed and characterised by Grimmonprez and MontreuilReference Grimmonprez and Nontreuil1. In the same account these researchers hypothesised that lactose induces the synthesis of oligosaccharides; we now know that milk from other mammals may have high concentrations of lactose but contain a limited repertoire of oligosaccharidesReference Urashima, Kobayashi, Asakuma and Uemura2, Reference Uemura, Asakumna and Yon3, Reference Lara-Villoslada, Debras and Nieto4, Reference Gopal and Gill5. Oligosaccharides in milk from marsupials and monotrems are also differentReference Urashima, Messer and Bubb6, Reference Messer, Gadiel and Ralston7. With time, it became apparent that milk possessed structures that were not found elsewhere in free, soluble forms. Researchers then began to isolate and characterise HMO to establish catalogues of possible glycoforms and elucidate metabolic pathways. Examples are two of the first comprehensive and systematic accounts of both analytical methodologies and chemical structures published by Kobata and coworkersReference Kobata, Yamashita and Tachibana8, Reference Kobata, Ginsburg and Tsuda9. During the last two decades, explorations of both structure and function increased in number, depth and scope after the discovery of Lewis and Secretor blood group determinants in HMOReference Kobata10. Later on, structures were identified and characterised by biochemical and physical-chemical methods, in some cases corroborating the existence of molecular speciesReference Prieto and Smith11, Reference Egge, Dell and Von-Nicolai12. As reagents such as polyclonal antibodies raised against immunogenic neo-glycoconjugatesReference Zopf, Smith, Drzeniek and Tsai13, Reference Smith, Prieto and Torres14, monoclonal antibodies raised against cell surface carbohydratesReference Magnani, Nilsson and Brockhaus15, and lectins became available, it was discovered that HMO structures can also be found as components of glycolipids and glycoproteinsReference Prieto and Smith16, Reference Newburg17. Some of these cell surface oligosaccharides have structural motifs that are only expressed at certain stages of tissue development and occasionally reappear in tissues upon malignant transformationReference Hakomori18. In fact, the use of a monoclonal antibody raised against pancreatic adenocarcinoma with specificity for an oligosaccharide epitope was proposed as a diagnostic tool for cancerReference Magnani, Nilsson and Brockhaus15. Interestingly, from the discovery of the oligosaccharide fraction of breast milk to the discovery of antigenic determinants in particular structures, the study of HMO follows the development of the science of glycobiology, while emphasis on the nutritional roles of these carbohydrates has only been the focus of intense research in the last decade or so. Discussions on the roles of HMO in paediatric nutrition may be arbitrarily grouped into four areas of inquiry: 1) HMO as analogues of cell surface oligosaccharides used by pathogens and toxins for attachment to mucosa, thus acting as decoys that inhibit bacterial, viral or toxin attachment; 2) HMO as soluble fibres that could act as prebiotics promoting colonisation with beneficial bacteria, facilitating transit and preventing or controlling colonisation by pathogenic or putrefactive bacteria; 3) as sources of particular monosaccharides, and 4) HMO as substances that interact with the immune system and modulate its response. All of these areas represent actual or putative biological activities that are supported by varying amounts and quality of evidence. The main obstacle to discerning the biological roles of HMO has been the lack of availability of these structures in sufficient quantities and purity to clinically test hypotheses on their biological functions. Several clinical studies have been conducted to compare the physiology and clinical behaviour of breast-fed versus formula-fed infants and in some cases the results of these studies have been attributed to HMO; however, in these instances it has been impossible to ascribe the observed effects exclusively to oligosaccharides since other breast-milk components may elicit or boost particular responses. In the past decade, methods for the synthesis of HMO have yielded kilogram amounts of specific structures that in turn have been used to supplement infant formulae for clinical trialsReference Prieto19. In the absence of HMO, several groups used non-human oligosaccharides to supplement formulae while seeking to emulate the physiological effects of human oligosaccharides. Two major classes of oligosaccharides have been assessed in clinical experiments in adults and infants and are now being used to supplement commercially available infant formulae: fructooligosaccharides (FOS) and galactooligosaccharides (GOS). FOS can be produced either by taking advantage of the reverse reaction of fructanases or sucrasesReference Su, Sheu and Chien20 or through the enzymatic hydrolysis of inulinReference Lopez, Coudray and Levrat-Verny21, a polymer of fructose. Chemically, both types of FOS differ and commercial literature cites competitive advantages with each type. In terms of composition, FOS synthesised using glycosidases, frequently lack a reducing end and contain one glucose residue and two or more fructose moieties. These are well characterised, the structures have their own namesReference Zdunczy, Król and Juskiewicz22, and are sometimes termed short chain FOS (scFOS). The molecules generated from inulin have free anomeric carbons and generally contain only fructose. On the other hand galactooligosaccharides (GOS) are mostly produced from the catalysis of reverse hydrolysis using galactosidasesReference Sowimol, Vasileios and Keshavan23. Carbohydrates that cannot be hydrolysed and absorbed during their transit through the human gastrointestinal system but are hydrolysed and metabolised by beneficial bacteria residing in the human colon are considered to be “prebiotics”. Both FOS and GOS fall within this definition and HMO behave partially in such a manner since some are also found intact in the faeces of breast-fed infantsReference Sabharwal, Sjöblad and Lundblad24. Therefore, both FOS and GOS have been studied for their abilities to emulate the physiological and biochemical effects of HMO. Although the rest of the present review is focused on experiences with HMO and breast milk, it is important to refer the reader to experiences with non-human oligosaccharides because, as stated above, formulae supplemented with these structures have already reached the market place.
Human Milk and HMO
Since several HMO contain blood/tissue specific antigens, and lactating mothers belong to different blood/tissue antigen groups, milk samples are varied in their glycosylation. This is due to the fact that the HMO repertoire in a sample constitutes a phenotype of secondary gene products (oligosaccharides) that result from the expression and activity of primary gene products (the enzymes that synthesise oligosaccharides). Erney et al. and Thurl et al. described HMO profiles that agree with the predicted variability of phenotypes according to the Lewis Blood Group and Secretor status of the donorReference Erney, Malon and Skelding25, Reference Erney, Pickering, Ruiz-Palacios, Prieto and David26, Reference Thurl, Heneker, Siege, Tovar and Sawatzki27. More recently, Chaturvedi et al. Reference Chaturvedi, Warren and Altaye28 corroborated the findings of Enrney et al. in terms of HMO profile variability amongst samples and through lactation. Enrney and Prieto have assembled evidence of the existence of profiles that are not consistent with predictions based on known genotypes; the existence of these profiles was preliminary published in an abstractReference Prieto, Hilty, Erney, Pickering and Palacios29. This evidence indicates that milk from different mothers may be qualitatively and quantitatively different in regard to its oligosaccharide content. This may also apply to other glycoconjugates, for example, glycoproteins that are decorated or remodelled by the same glycosyltransferases that synthesise HMO. Samples from Lewis positive mothers contain both the expected oligosaccharides resulting from the fucosyltransferase encoded by the Le gene and glycoproteins that react with monoclonal antibodies against antigens of the Lewis system. Similarly, transgenic animals expressing glycosyltransferases under a lactogenic promoter produce both neooligosaccharides and neoglycoconjugatesReference Prieto, Mukerji and Kelder30, Reference Kelder, Enreny, Kopchick and David S31. The unavoidable conclusion is that breast-milk is different from one mother to the other. Some mothers synthesise fucosylated sugars such as Fucα1-2 Galβ1-4Glc, others do not, and these oligosaccharides can be present at concentrations up to 2 grams per litre according to our studies. If this oligosaccharide or, in general, Fucα1-2 Gal residues are important, some breast-fed infants are not being exposed to these structures through their mother's milk. With regard to glycosylation, breast-milk is heterogeneous.
Clinical evidence that supports biological roles for HMO
Table 1 is a summary of selected results from reports that explored the biological roles of HMO. This table includes studies other than clinical experiences because its purpose is to illustrate prevalent research lines. Also, there is a limited amount of clinical studies that specifically shed light on HMO functions in the paediatric nutritional context. When studying the effects of HMO on breast-fed infants, two confounding factors arise: putative biological effects of HMO have also been ascribed to glycoproteins and glycolipidsReference Bessler, de Oliveira and Giugliano32, Reference Rueda, Sabatel and Maldonado33 and supra-molecular constituents of milkReference Keenan34, Reference Keenan and Newburg35, and glycoconjugates may share structural features with HMO. These two factors make it difficult to ascertain the particular roles of HMO in clinical studies comparing breast-fed versus formula-fed infants and, to our knowledge, only one clinical study with a formula supplemented with a human milk oligosaccharide has been reported in the literatureReference Prieto19, Additional factors may confound interpretation of experiments aimed at probing the ability of particular oligosaccharides to function as prebiotics; one of these is the characterisation of species and varieties of components of the intestinal flora. A review of pertinent literature shows that increased faecal counts of lactobacillus and bifidobacterium species are considered to be beneficial outcomes of the ingestion of HMO or other oligosaccharidesReference Kunz, Rudloff S Baier and Klein36. One publication suggests the importance of a particular species and variety of bifidobacterium and mentions that the oligosaccharide Lacto-N-neotetraose (Galβ1-4GlcNAcβ1-3 Galβ1-4Glc) is a particularly suitable prebiotic for such a microbeReference Kunz and Rudloff37. While this may be true for in vitro experiences there is no clinical evidence to support this assertion. Species and strains are sometimes confused because of the lack of standards to define them. More than ten years ago Dubey and MistryReference Dubey and Mistry38 reported another confounding factor: species of bifidobacterium grow at different rates under the same conditions. These include degree of anaerobiosis, pH, particular nutrients and temperature. Conducting in vitro studies to determine the prebiotic effect of oligosaccharides is relatively easy and results of significant difference can be detectedReference Prieto19. This is true when comparing bacterial responses to different oligosaccharides using either growth or selected metabolic or enzymatic activities as benchmarks. In contrast, similar experiences in the presence of relatively large concentrations of lactose (such as in milk) generate results that are difficult to interpret. When saccharide-supplemented infant formula is used to assess its prebiotic effect as a whole, other factors such as the concentration of lactulose (generated during the thermal process) become critical because its prebiotic effect may mask or enhance the effect of other saccharide structuresReference Vanhoutte, De Preter and De Brandt39. In lactose-free formulations the concentration of sucrose becomes as important specially when assessing the effects of fructooligosaccharides. As indicated above, only one clinical study has been published in which a pure HMO was used to supplement infant formula. The oligosaccharide (LNnT) was present in formula at the average concentration in which it is found in breast milk (200 mg/l). Perhaps such a study should only be considered a proof of the concept that demonstrates the feasibility of conducting large-scale clinical studies with HMO-supplemented formulae. A detailed analysis of Table I reveals that a systematic effort has been conducted by Ruiz-Palacios and coworkers to probe the anti-pathogenic effects of fucosylated oligosaccharides, specifically those that contain the Fucα1-2 structural motifReference Ruiz-Palacios, Cervantes, Ramos, Chavez-Munguia and Newburg40, Reference Newburg, Ruiz-Palacios and Altaye41. This group used CHO cells transfected with a human fucosyltransferase that catalyses the synthesis of Fucα1-2 residues. These cells express fucosylated glycoproteins in their cell surfaces which allowed them to compare the ability of strains of enteropathogenic Campylobacter jejuni to bind to these cells and to the wild typeReference Prieto, Larsen, Cho and Rivera42. In addition they also took advantage of a transgenic animal model in which mice expressing the same fucosyltransferase express 2′FL in their milkReference Prieto, Mukerji and Kelder30. The investigators inoculated newborn mouse pups with C. jejuni and compared the persistence of colonisation in pups fed by transgenic mothers and those fed by wild type mice. Pups fed with milk containing transgenically expressed Fucα1-2 oligosaccharides and glycoconjugates shed the pathogenic bacteria faster than pups fed with control milk. Furthermore, this group also conducted a cohort study to determine the effect of Fucα1-2 on diarrhoeaReference Newburg, Ruiz-Palacios and Altaye41. Because some mothers do not synthesise this structure, they were able to ascertain that infants fed milk containing Fucα1-2 were less susceptible to diarrhoea. In this particular case there is a clear progression from in vitro to clinical relevance. Other studies have compared the ability of breast-milk to promote colonisation with bifidobacteria or lactobacillus, or to decrease counts of pathogenic bacteria, or have focused on quantifying HMO in faecesReference Sabharwal, Nilsson, Grönberg and Chester43, Reference Sabharwal, Nilsson, Chester and Lindh44, Reference Iseki45. Ex vivo digestibility studies complement these observationsReference Gnoth, Kunz, Kinne-Saffran and Rudloff46. The evidence suggests that the least controversial role for HMO is as primordial fibres and prebiotics as elucidated by Coppa and coworkersReference Coppa, Bruni, Morelli, Soldi and Orazio47. Additional work has been conducted to determine if HMO can be used as nutritional anti-infectivesReference Andersson, Porras, Hanson, Lagergård and Svanborg-Edén48, Reference Idänpään-Heikkilä, Simon, Zopf and Vullo49. In this regard, the work by Idänpään and coworkers was predicated on the ability of a particular structural feature of oligosaccharides to inhibit colonisation of epithelial cells of animals with S. pneumoniae. This work also evolved in demonstrations of efficacy in animal models and progress into the referred clinical trialReference Prieto19.
Conclusions
HMO present unique problems for the assessment of their biological functions and the conducting of tests in the context of paediatric nutrition. The work of Ruiz-Palacios and coworkers suggests that infants fed with milk lacking or having low concentrations of 2′FL may be more susceptible to diarrhoea than their counterparts fed with human milk containing 2′FLReference Ruiz-Palacios, Cervantes, Ramos, Chavez-Munguia and Newburg40. This simple observation implies that emulating human milk and HMO is not as simple as previously thought. HMO may provide nutritional advantages to infants but may also have other applications. If it is true that fucosylated glycoconjugates that contain Fucα1-2 residues provide protection from diarrhoea in a paediatric nutritional context, then these saccharides could be excellent candidates to develop pharmaceutical applications for adults as well as infants. It is proposed here that it is no longer sufficient to reiterate that “more research is needed in the field of HMO”; now that these structures can be synthesised in large quantities it is feasible to pursue at least two classical paths of inquiry; 1) traditional in vitro, in vivo safety and pharmacokinetic studiesReference Leach, Garber, Marcon and Prieto50 and 2) well-controlled clinical trials. The first approach may or may not lead to nutritional applications but may branch out as a specific type of pharmaceutical pool of drug candidatesReference Kobata51. Recognition of subtleties and perhaps limitations imposed by nature on breast-milk and its components is the first step to understanding applications of HMO that extend beyond the realm of nutrition.
Conflict of interest statement
PAP worked for Abbott Laboratories until March, 2007 and still holds stocks of this company in his retirement fund. Abbott Laboratories produces infant formula. RME and MT have no conflict of interest to declare. The article was co-written by all authors.