The Human Milk Microbiome – Possible Implications for Infant Health

INSERT FOR PAEDIATRICIANS

CONTENT Human milk: what is normal? Modulation by diet Maternal and perinatal factors Shaping the infant gastrointestinal microbiome © Ramona Heim / Fotolia

Human milk research group The Human Milk Microbiome –

Possible Implications for Infant Health

The human body contains and carries a large number of microorganisms and, surprisingly, the mammary gland and mother’s milk also contain their own microbiota. Very little is known, however, about the function of the human milk microbiome or the factors responsible for the marked variation observed within and among populations all over the world. It is hypothesized that the human milk microbiome plays an important role in the development of an infant‘s gastrointestinal microbiota and thus for its health in the future. Scientific exploration of the human milk microbiome and possible strategies for modification has just began but appears to be very promising.

The first studies investigating the quantitative, and only 30 % of bacte- Entero-mammary pathway bacterial content of human milk were un- ria can be cultured. On the other hand, The presence of obligate anaerobic bacteria dertaken in the 1970s and 1980s because culture-independent methods such as was not compatible with the traditional view of concerns that bacteria might affect pre- 16S rRNA gene sequencing are ex- that human milk is sterile by nature. Bacteria term babies fed with human milk. At that pensive but may provide in-depth detected in milk were thought to stem exclu- time, the only bacteria that had been de- information about the microbi- sively from the breast skin or from the baby’s tected in human milk were Staphylococci al composition in human milk. mouth colonizing the mammary gland, said and Streptococci, reported Dr. Esther Jimé- 16S rRNA sequencing data indicatedthat Jiménez. nez, Madrid (Spain). It was not until 2003 the bacterial communities of human milk In another study the same lactic acid that lactic acid bacteria were isolated from are complex and often stable over time bacteria were found in locally fermen- human milk [1]; these were regarded as within an individual [3]. In addition, sub- ted food, maternal milk, and fecal sam- beneficial because they were able to sup- stantial inter-individual variability was ob- ples of breastfeeding mothers and their press the growth of Staphylococcus aureus served. “Each woman’s milk has its own babies [5]. This observation supports the hy- in the laboratory [2]. bacterial profile,“ concluded Jiménez. pothesis that bacteria are transferred from In particular, 16S rRNA sequencing the mother’s gastrointestinal tract to her milk Culture-dependent and culture- has been very useful in identifying bacte- and through the milk to the infant’s gastroin- independent techniques ria that cannot be cultured, such as those testinal tract. According to Jiménez, about 50 gene- that cannot survive in the presence of oxy- As Jiménez reported, evidence has in- ra and 200 species have been detected gen (anaerobes), in different body fluids. creased that a mother’s gastrointestinal bac- in human milk using culture techniques. For instance, a Swiss group detected such teria are able to reach the mammary gland The normal concentration of bacteria in obligate anaerobic genera like Bifidobac- via an endogenous route. For this bacterial milk from healthy women was shown to terium, Veillonella, and Bacteroides in the be about 103 colony-forming units (cfu) feces of healthy breastfed neonates using a per milliliter. Culture techniques, how- comprehensive approach with culture and ever, have limitations. They are semi- molecular methods [4]. INSERT FOR PAEDIATRICIANS

translocation, dendritic cells appear to play those of non-secretor mothers [10]. Factors se diets contained more Bacteroides. Micro- an important role. These specialized cells influencing these variations are widely un- nu-trient intake correlated inversely with the were shown to open the tight junctions be- known. In McGuire’s opinion, “we need to re- relative abundance of Firmicutes, while diets tween epithelial cells in the digestive tract. think the idea of there being a normal range rich in eicosapentaenoic acid (EPA) or doco- By sending projections into the intestine, of milk compounds.“ sahexaenoic acid (DHA) were associated with dendritic cells were able to sample bacteria more Firmicutes. Diets rich in amino acids residing there and take them to the lamina Marked variations positively correlated with higher counts of propria [6]. From there, bacteria likely tra- Furthermore, the INSPIRE project revealed Proteobacteria and Fusobacteria. vel with immune cells through the body; in- clear population differences and variation In McGuire’s view, “the only way to verify if deed, DNA signatures of intestinal bacteria within populations in the maternal milk mi- we can modulate the milk microbiome with were found in immune cells in both human crobiome [11]. For instance, in a Hispanic diet is to do controlled intervention studies.“ milk and blood of lactating mothers [7]. population from California (USA) more Sta- McGuire differentiated between three phylococcus and Escherichia/Shigella were potential routes whereby maternal diet could Human milk: what is normal? found as compared with a population from influence maternal milk microbiota: Little is known about what is normal in terms Washington/Idaho (USA), in which Lactoba- — Bacteria consumed with probiotic food of human milk composition, including its mi- cillus didn’t even belong to the 30 most fre- translocate from the gastrointestinal crobiome. Therefore, the INSPIRE project was quently encountered genera. In populations tract to the mammary gland. from Peru or Spain, Streptococcus was pre- — Ingestion of micronutrients influences dominant; in Sweden, however, Staphylococ- the growth of bacteria present in the cus was most frequently encountered. While gut, from where these bacteria get to milk microbiomes in urban and rural popu- the mammary gland. lations in Gambia were similar, a remarkable — Nutrients reach the mammary gland (via dissimilarity was observed between an urban circulation) where they affect local bac- and a rural population from Ethiopia. teria differently. McGuire reported that Staphylococcus But things may not be that simple, because and Streptococcus (present in 99 % and 98 % bacteria could contribute to milk nutrients of the samples) were universal core bacteria and thereby impact the milk microbiome. [11]. The next-frequent genera were Propio- “So, to look at the influence of diet, we need nibacterium (in 78 % of the samples) and Co- to assess diet, the human milk, and fecal mi- rynebacterium (in 74 % of the samples). The crobes all at the same time,“ said McGuire. data showed, however, that there are differ- The INSPIRE Project also found a large va- ent core bacteria in different regions. riability in healthy infant fecal microbiomes “At the time being we don’t understand within and among populations [11]. In or-

© Anatomy Insider / stock.adobe.com © Anatomy what is a normal or healthy human milk mi- der to explain the variability, it is not enough crobiome,“ said McGuire. What is regarded as to differentiate between “breastfed“ and “not normal in one population may be subopti- breastfed,“ said McGuire. In her opinion, it is started to determine (using standardized data mal for another. In McGuire‘s opinion, how- necessary to account for the variability in the collection methods) how human milk compo- ever, “understanding this is a must before we human milk composition. sition and microbiome vary in different parts can understand what is unhealthy or inter- of the world. As Dr. Michelle McGuire, Pull- vene.“ And “normal“ or “healthy,“ she said, is Maternal and perinatal factors man (USA), reported, 413 healthy breastfeed- clearly not a one-size-fits-all construct. In addition to geographic location and diet, ing women with healthy babies belonging to several other maternal and perinatal factors populations from Africa, Europe, South Amer- Modulation by diet have been shown to be related to the human ica, and the USA were included in this study. In another 21 healthy breastfeeding women, milk microbiome, reported Dr. Maria Carmen Substantial population differences were the McGuire research group performed de- Collado, Valencia (Spain). found regarding the content of milk prote- tailed dietary assessment over a period of 6 Cabrera-Rubio et al. observed human in, immune factors, human milk oligosaccha- months postpartum to investigate the in- milk microbiome changes during the course rides (HMOs) such as 2‘-fucosyllactose (2’FL), fluence of diet on the human milk microbi- of lactation [13]. Weissella, Leuconostoc, and even lactose, which is regarded as ome [12]. As McGuire reported, multiple re- Staphylococcus, Streptococcus and Lac- beeing the least variable component of hu- lationships were found. For instance, high- tococcus were predominant in colostrum, man milk [8–10]. HMO profiles in milk pro- calorie diets were associated with a higher whereas in mature human milk lactic acid duced by mothers categorized as secretors abundance of Firmicutes in milk, whereas bacteria prevailed. Furthermore, in mature (2’FL producers) differed substantially from milk of mothers consuming less-energy-den- milk an increased abundance of such oral

2 Springer Medizin | Insert for Paediatricians bacteria as Veillonella, Leptotrichia and relationship was evident. According to Mc- of life, breastfed infants obtained 27.7 % of Prevotella was observed. Guire, there is a clear pattern “indicating that their intestinal bacteria from human milk and Maternal health may also be an impor- the secretor status is somehow interacting 10.4 % from areolar skin. Changes in the in- tant factor. For instance, milk Bifidobacteri- with many relationships of interest.“ fant fecal microbiome were associated with um levels were significantly lower in aller- The association between maternal se- the proportion of breastfeeding in a dose-de- gic as compared with non-allergic mothers cretor status and the infant fecal microbio- pendent manner. (p=0.003) [14]. Collado and co-workers de- ta composition has been confirmed in other The infant intestinal microbiome under- tected lower Bifidobacterium and higher studies. For example, infants fed by non-sec- lies a dynamic process. Initially the diversi- Staphylococcus levels in milk produced by retor mothers were delayed in the establish- ty of the microbiome is low, but it increases obese mothers than in normal-weight moth- ment of Bifidobacterium-laden microbiota over the first 2 years of life (Fig. 2) [22]. For ers [15]. In milk of normal-weight mothers a (Fig. 1) [18]. Another study suggested an in- specific bacteria specific patterns of relative higher bacterial diversity was observed [13]. fluence of maternal and infant secretor status abundance are observed. For instance, lev- The milk microbiomes from mothers with va- on the infant microbiota composition at the els of Bifidobacterium and Enterobacteria- ginal delivery or Cesarean section differed in age of 2 or 3 years [19]. ceae steadily decrease during maturation of bacterial composition and diversity [13, 16]. the gastrointestinal microbiome, while the Interestingly, differences were observed in Shaping the infant abundance of Bacteroidaceae increases [23]. milk microbiotas of mothers with emergency gut microbiome As Collado pointed out: “Any disturbance of or programmed Cesarean section [13], repor- Maternal microbiomes and diet (e. g. human the microbiota development during this cri- ted Collado. In her opinion, these differences milk versus formula feeding, introduction of tical phase of life would affect the microbial may be due to differential stress levels impac- solid food) have been shown to be associa- composition.“ For example, antibiotic expo- ting intestinal permeability. In a recent study, ted with variation in the infant’s gastrointes- sure or Cesarean section may disrupt the es- lower levels of Staphylococcus and Pseudo- tinal microbiome composition [20]. tablishment of the infant’s intestinal micro- monas spp. were found in the colostrum of “Human milk is a constant source of mi- biome [22]. mothers after vaginal delivery than after Ce- crobes,“ said Collado. The importance of the sarean section [17]. In addition, the mode of milk microbiome for the establishment of the Infant gastrointestinal delivery had a strong influence on the micro- infant gastrointestinal microbiome has been microbiota and health biota network. confirmed in a study in which the infant gut A mother’s microbiome is now recognized microbial communities were shown to be as one of the essential factors determining HMOs  the role more closely related to the mother’s milk and neonatal health outcomes. Early exposure to of the secretor status skin as compared with the milk and skin of a mother’s milk microbiota is thought to im- In non-secretor mothers (as opposed to the random mother [21]. During the first month pact the immune system development and whole study population) McGuire and co- workers found a strong relationship between Figure 1 HMO profiles in milk and the milk microbi- Relative levels of gut microbiota in infants fed by secretor mothers (S) as compa- ome and similarly (in contrast to secretor red with non-secretor mothers (N) mothers) between immune factors in maternal milk and the milk microbiome. Secretor Status: NS Maternal and perinatal factors appear to 100 p<0.05 p<0.05 affect not only the human milk microbiota, 90 but also biologically active milk compounds 80

(such as HMOs, fatty acids, cytokines, poly- 70 amines, proteins), concluded Collado, and 60 “we need to understand the interactions between the human milk microbial commu- 50 nities/milk-born compounds and the infant 40 gut development.“ 30 Taxa relative abundance (%) The relationship between milk and the 20 infant fecal microbiomes seemed to be 10 modulated by the maternal secretor sta- 0 tus as well. In the whole study population, Bacteroides Biﬁdo- Clostridium Entero- Escherichia/ Streptococcus bacterium bacteriaceae Shigella McGuire found no association between milk (Other) microbiomes and the infant fecal microbiomes, but in non-secretor mothers a clear modified after [18]

Mother’s milk for preterm infants Preterm infants often are delivered by Cesa- rean section and receive antibiotics; they differ from term babies in nutrition, have delayed hospital stays and often an impaired gastric barrier and an aberrant gastrointestinal microbiota, characterized mainly by a low number of Bifidobacterium and Lactobacil- lus and a high prevalence of Enterobacteria. Therefore, feeding human milk may be even more important for preterm infants than for term infants. According to Jiménez, the best option is the infant’s own mother’s milk, fol- lowed by donor human milk and formula milk. Jiménez pointed out that distortions in the development of intestinal microbiota may have a strong impact on preterm infants, for instance causing an increased risk of de- veloping necrotizing enterocolitis or sepsis. 01234567891011 12 14 16 18 20 22 24 Month of life Breast cancer linked with bacterial dysbiosis modified after [22] A mother’s microbiome may be important not only for the health of the infant but also for the mother’s health. dances of M. radiotolerans and S. yanoikuyae increase in the abundance of Acinetobacter A link between breast cancer and dysbi- correlated inversely in normal breast tissue and Xanthomonadaceae and a marked de- osis in breast tissue was suggested by Xuan but not in tumor tissue. In addition, the bac- crease in bacteria believed to confer bene- et al., who analyzed microbiota in tumor tis- terial load in tumor tissue was reduced com- ficial effects for infants (Bifidobacterium and sue and paired adjacent breast tissue from pared with that in adjacent normal breast tis- Eubacterium, but not Lactobacillus). the same patients [24]. As Jiménez pointed sue or breast tissue of healthy women. out, bacteria of the Methylobacteriaceae fa- In a lactating woman diagnosed with Mastitis  response mily occurred in much higher abundance in Hodgkin’s lymphoma in the 2nd trimester to milk lactobacilli tumor tissue (p=0.0237), whereas bacteria of pregnancy, chemotherapy affected both Mastitis is a painful condition and regarded of the Sphingomonadaceae family were sig- bacterial diversity and profiles of her milk as the principal reason to stop breastfee- nificantly enriched in adjacent breast tissue [25]. According to Jiménez, bacterial compo- ding. According to Jiménez, mastitis is cha- (p=0.0079) (Fig. 3). Among these families, sition changed completely between baseline racterized by bacterial dysbiosis and high M. radiotolerans and S. yanoikuyae were the and weeks 4–16 of therapy. In particular, che- counts of one or two predominating spe- most abundant species. The relative abun- motherapy was accompanied by a striking cies. In a study by Mediano et al., the num-

Springer Medizin | Insert for Paediatricians 4 INSERT FOR PAEDIATRICIANS

Figure 3 — Would it be beneficial to enrich the ma- Abundance of the Sphingomonadaceae and the Methylobacteriaceae families in ternal diet or infant formula diet with paired breast cancer tissue and adjacent normal breast tissue from 20 patients prebiotic supplements? with estrogen receptor-positive breast cancer — Should it be an aim to influence the ma- *p=0.0079 *p=0.0237 ternal microbiota and thereby possibly 7000 17500 the long-term outcome of the child? 6000 15000 — And finally: “Do we have sufficient scien- 5000 12500 tific data to convince the regulator and 4000 10000 get approval?“ 3000 7500 Convincing the regulator, however, has pro- 2000 5000 family abundance family abundance ven difficult in many cases. One reason is that Methylobacteriaceae Sphingomonadaceae 1000 2500 “science is always ahead of regulation, which 0 0 Paired normal Tumor Paired normal Tumor only very gradually adapts,“ reported Salmi- nen. For instance, in microbiology the use of modified after [24] culture techniques has been standard. How- ever, new methods have been introduced, such as PCR (polymerase chain reaction) or ber of bacterial species in 1849 milk samples by their healthcare providers. After 3 weeks, 16S rRNA DNA sequencing, and the regula- from women suffering from mastitis probiotic treatment was associated with a tor asks: “What is the generally accepted sci- ranged between 1 and 7 [26]. The milk sam- clear reduction of bacterial counts and less ence?” Furthermore, the regulator often re- ples clustered in two groups: One (about breast pain [28]. gards the relation between cause and conse- 60 % of samples) contained predominantly quence as not being clear enough to support Staphylococcus epidermidis, often in com- Where science a health claim for a specific pre- or probiotic. bination with Streptococcus (mitis or sali- and regulation meet In Salminen’s opinion, the decision pro- varius). The remaining 40 % of samples con- Knowledge regarding the influence of mater- cess is especially complicated in the Euro- tained a mixture of mainly different Strep- nal diet on human milk, specifically the hu- pean Union (EU). The European Food Safe- tococcus species (especially mitis) together man milk microbiome, and the possible im- ty Authority (EFSA), responsible for assessing with Rothia spp., Staphylococcus aureus, or pact of nutrition on the infant’s outcome is submitted health claims for scientific sub- S. epidermidis. steadily increasing. Consequently, new ques- stantiation, is an independent authority ad- In clinical studies performed by Jimé- tions are arising, said Prof. Dr. Seppo Salmi- vising the European commission and parlia- nez‘ group, lactating women with staphy- nen, Turku (Finland). For instance: ment. The decision about approval, however, lococcal mastitis derived benefits from oral — Do we need to introduce “protective fac- is always made by the European commis- treatment with lactic acid bacteria isolated tors originating from human milk” in ad- sion, often in consultation with the European from milk of healthy mothers [27, 28]. dition to breastfeeding? parliament or the EU member states. In spe- In a pilot study, women with mastitis who had previously failed to respond to antibiotics randomly received Lactobacillus salivarius and Lactobacillus gasseri or placebo HiPP initiative: Human milk research group for 4 weeks [27]. Mean staphylococcal counts in milk samples of both groups were simi- Human milk has always been considered the natural model for the production of lar at baseline. Following treatment, howe- infant milk formula, as human milk optimally supports infants‘ natural development. ver, they were substantially lower in the pro- Therefore, the ”Human milk research group“ – initiated by HiPP – has been intensively biotic group as compared with controls. At exploring the composition of human milk and its positive effects on human health for the end of treatment, lactic acid bacteria years. were not detectable in the milk of the control The research group has met for thematic workshops on a regular basis. Representa- group but were found in substantial numbers tives of HiPP‘s nutritional science and product development divisions discussed the in the milk of most patients in the probiotic issue of ”The microbiome of human milk“ together with Professor Michelle McGuire, group, reported Jiménez. School of Biological Sciences at Washington State University in Pullman (WA, USA), These results were confirmed in a study Dr. Esther Jiménez Quintana, Departamento de Nutrición, Bromatologia y Tecnologia of 352 women with infectious mastitis who de los Alimentos, Universidad Complutense, Madrid (Spain), Dr. Maria Carmen Collado, were randomized into three groups to be Instituto de Agroquimica y Tecnologia de Alimentos  Consejo Superior Investigaci- treated with Lactobacillus fermentum, Lacto- ones Cientificas (IATA-CSIC), Valencia (Spain), and Professor Seppo Salminen, Functio- bacillus salivarius, or antibiotics as prescribed nal Foods Forum, Faculty of Medicine, University of Turku (Finland).

5 Springer Medizin | Insert for Paediatricians © fotoliaxrender / Fotolia 6 Ulrike Hafner(responsible) Head ofCorporate Publishing: Andrea Krahnert Editor: Dr. Günther Springer, Darmstadt Coverage: Prof. Dr. SeppoSalminen, Turku, Finnland Dr. Maria Carmen Collado, Valencia, Spanien Dr. Esther Jiménez Quintana,Madrid, Spanien Referents: Prof. Dr. McGuire, Michelle Washington, USA Chair: Erding, 9 Microbiome Hipp Research Workshop ontheHumanMilk Imprint significance of thisrelationship, however, has fants’ microbiomes are related. The functional There isgoodevidencethatmothers’ andin- Summary tion andactivity,“ concludedSalminen. health claimsrelated to microbiota composi- positive evaluationorapositive decisionon ent time, “it difficultto geta would bevery generally accepted. Therefore, at the pres- dysbiosisnortheconsequencesarebacterial potentiallyneither theriskfactors leadingto quences. Butwithregard to microbiota, causes,on between andconse- riskfactors sion, inSalminen’s opinion, istherelati- est standards.for thedeci- Mostimportant en, has to be done according to the high- groups, suchaschildren orpregnant wom- especially iftheyare related to vulnerable cific casestheassessmentofhealthclaims, Springer Medizin | Insert for Paediatricians Springer Medizin |Insert INSERT FOR PAEDIATRICIANS th October 2017 October Fachverlagsgruppe SpringerNature. Springer Medizin Verlag of GmbHispart CEO: JoachimKrieger, Fabian Kaufmann 14197 Berlin Heidelberger Platz 3 Springer Medizin Verlag GmbH Vertrieb KG, Pfaffenhofen/Ilm This by HippGmbH&Co. publication issupported Volume 166,Issue7,July2018 in Insert breast cancerormastitis. as indicated by links with diseases such as ome apparently influences maternal health, babies: variation in the human milk microbi- to infanthealth.However, itisnotallabout the variationinmaternal milkmicrobiomes to relate addition, it is necessary by diet. In the humanmilkmicrobiome, for instance tion studiesto seeifitispossibleto influence step interven- The willbeto conduct next ulation andbythematernal secretor status. the infantfecal microbiome variesbypop- thehumanmilkmicrobiomebetween and diet, andmaternal health. The relationship asenvironmentalfactors exposure, genet-ics, globe whichappeared to bedriven bysuch stantial variationhasbeenfound around the mal” or milkmicrobiome.“healthy“ Asub- has beenonunderstandingwhatisa “nor- yet to bedetermined. Upto nowthefocus Monatsschrift Kinderheilkunde“ ”Monatsschrift 8 Arroyo Retal.,Dis2010,50:1551 ClinInfect 28. JiménezEetal., ApplEnviron Microbiol 2008, 27. 2017,33:309318 MedianoPetal., JHumLact 26. UrbaniakCetal., Microbiome 2014,2:2434 25. Xuan Cetal., PLoS ONE2014,9:e83744 24. Yassour Metal., Sci Transl23. Med2016, BokulichNAetal., Sci Transl Med2016, 22. Pannaraj PSetal., JAMA Pediatr 2017, 21. Tanaka M,NakayamaJ, Allergol 2017, Int 20. Smith-Brown Petal., PLoS One2016, 19. Lewis ZTetal., Microbiome 2015,3:13 18. Toscano Metal., BMCMicrobiol 2017,17:205 17. Cabrera-Rubio Retal., JDevOrigHealthDis 16. Collado MCetal., Pediatr Res2012,72:7785 15. Grönlund MMetal., ClinExpAllergy 2007, 14. Cabrera-Rubio Retal., AmJClinNutr2012, 13. Williams JEetal., JNutr2017,147:17391748 12. McGuire etal., unpublishedresults 11. McGuire MKetal., AmJClinNutr2017, 10. RuizLetal., Front 2017,8:696 Immunol 9. McBrideetal., unpublishedresults 8. Perez PFetal., Pediatrics 2007,119:e724e732 7. Rescigno Metal., 2001,2:361 Nat Immunol 6. Albesharat Retal., Syst AppMicrobiol 2011, 5. Jost T etal., BrJNutr2013,110:12531262 4. HuntKMetal., PLOS One2011,6:e21313 3. MP, Heikkilä JApplMicrobiol 2003, SarisPEJ, 2. Retal., JPediatr Martin 2003,143:754758 1. Literature of reference. ineachcasebythereader,for accuracy usingothersources application. Informationofthisnature should bereviewed be heldliablefprinformation ondosageandmethodsof use. Product liability: The printers andthepublisherscannot copyright andtrademarklaws orthattheyare for general these namesshouldberegarded inaccordance withthe the presence norabsenceofsuchdesignation meansthat areused inthistext designated assuch.However, neither Some commercial names, tradenamesandtrademarksetc. Printers: DruckpressGmbH,Leimen © Springer Medizin Verlag GmbH 1558 74:46504655 8:343ra81 8:343ra82 171:647654 66:515522 11:e0161211 2016, 7:5460 37:17641772 96:544551 105:10861100 367 34:148155 95:471478

48268