Báo cáo khoa học: Human milk provides peptides highly stimulating the growth of bi®dobacteria pptx

Several bi®dogenic peptides were puri®ed chromatographically from pepsin-treated human milk and identi®ed as proteolytically generated fragments from the secretory component of the solub

Human milk provides peptides highly stimulating

the growth of bi®dobacteria

Cornelia Liepke, Knut Adermann, Manfred Raida, Hans-JuÈrgen MaÈgert, Wolf-Georg Forssmann

and Hans-Dieter Zucht

IPF PharmaCeuticals GmbH, Hannover, Germany

The large intestine of breast-fed infants is colonized

predominantly by bi®dobacteria, which have a protective

e€ect against acute diarrhea In this study we report for the

®rst time the identi®cation of human milk peptides that

selectively stimulate the growth of bi®dobacteria Several

bi®dogenic peptides were puri®ed chromatographically from

pepsin-treated human milk and identi®ed as proteolytically

generated fragments from the secretory component of the

soluble polyimmunoglobulin receptor and lactoferrin; both

of these proteins exhibit antimicrobial e€ects Hydrolysis of

the identi®ed peptides with the gastrointestinal proteases

pepsin, trypsin and chymotrypsin did not lead to the loss of

bi®dogenic activity, indicating their potential function

in vivo Sequential comparison revealed a similar structural motif within the identi®ed peptides A correspondingly designed small peptide (prebiotic lactoferrin-derived peptide-I, PRELP-I) was found to stimulate the growth of bi®dobacteria as e€ectively as the native peptides The combination of antimicrobial and bi®dobacterial growth stimulatory activity in human milk proteins leads to highly speci®c compounds capable of regulating the microbial composition of infants' large intestine

Keywords: bi®dobacteria; human milk; intestinal micro¯ora; lactoferrin; peptides

The colonization of the intestine with nonpathogenic

microorganisms is essential for the health and wellbeing of

the newborn Breast-feeding creates an intraintestinal

envi-ronment that favours a simple ¯ora of bi®dobacteria and

few other microorganisms [1,2] Several epidemiological

studies have shown that the incidence of gastrointestinal

infections is lower in infants who are exclusively breast-fed

than in infants who are fed on formula [3±5] The occurrence

of bi®dobacteria in the large bowel is bene®cial for the

infant because it prevents the proliferation of pathogens that

cause diarrhea, such as salmonella or rotaviruses [6±8] It

has been shown that breast milk contains speci®c bi®dus

growth factors, such as certain carbohydrates, supporting

the implantation of these bacteria (reviewed in [9]) It has

been assumed also that signi®cant bi®dogenic activity is

associated with milk protein [10±14], either by direct growth

stimulation or by antimicrobial effects However, the

molecules and mechanisms underlying the bi®dogenic

activity of human milk protein, especially the effect of

peptides resulting from the digestion of these proteins, have

hitherto not been well understood In the present work, we

report that the bi®dogenic activity of human milk protein is

due to small proteolytic fragments of major milk proteins

These peptides obtained from the digestion of milk protein

with the gastric protease pepsin were identi®ed as effective growth factors for bi®dobacteria

M A T E R I A L S A N D M E T H O D S

Puri®cation of peptides Human milk and colostrum (collected 2±3 months and

2 days after birth, respectively) were obtained after informed consent from healthy voluntary donors Pepsin (20 mgág)1milk protein) was added to human and bovine milk after acidi®cation to pH 3.5 with HCl, and this mixture was incubated for 2 h at 37 °C The proteolysis was stopped

by boiling for 5 min After centrifugation (20 min, 6000 g,

4 °C), the fat was skimmed off and the supernatant was collected Tri¯uoroacetic acid was added to the supernatant

to a ®nal concentration of 0.1% Solid particles were removed by centrifugation (20 min, 2500 g, 4 °C) and

®ltration For peptide puri®cation, three subsequent HPLC steps were carried out, tracking the maximum growth stimulatory activity of the resulting fractions on Bi®dobac-terium bi®dum DSM 20082 First HPLC: separation of

20 mL pepsin-treated human milk using a C18-reversed-phase column (Parcosil, 10 mm ´ 125 mm, 5 lm, 100 AÊ, Biotek, OÈstringen, Germany); eluent A, 0.1% tri¯uoroacetic acid; eluent B, 0.1% tri¯uoroacetic acid in acetonitrile; ¯ow rate, 2 mLámin)1; gradient, 0±60% B in 45 min; UV detection, 280 nm (Fig 1A) Second HPLC: rechromato-graphy of bi®dogenic fractions with the same conditions as above using a shallower gradient, 20±50% B in 45 min Third and ®nal puri®cation step: rechromatography on a cation exchange column (Parcosil Pepkat, 4 ´ 50 mm,

5 lm, 300 AÊ, Biotek); eluent A, 10 mM phosphate buffer (pH 4.5); eluent B, 10 mMphosphate buffer (pH 4.5) with

1MNaCl; ¯ow rate, 0.75 mLámin)1; gradient, 0±60% B in

Correspondence to C Liepke, IPF PharmaCeuticals GmbH,

Feodor-Lynen-Strasse 31, D-30625 Hannover, Germany.

Fax: + 49 511 5466 132, Tel.: + 49 511 5466 201,

E-mail: c.liepke@ipf-pharmaceuticals.de

Abbreviations: CZE, capillary zone electrophoresis; hLACFR, human

lactoferrin; hPIGR, human polyimmunoglobulin receptor; PRELP-I,

prebiotic lactoferrin-derived peptide-I.

(Received 6 September 2001, revised 26 November 2001, accepted 29

November 2001)

60 min; UV detection, 214 nm The resulting HPLC

fractions were desalted prior to testing for bi®dogenic

activity by use of an analytical C18 column (Vydac,

4.6 ´ 250 mm, 5 lm, 300 AÊ, The Separations Group,

Hesperia, CA); eluent A, 0.1% tri¯uoroacetic acid; eluent

B, 0.1% tri¯uoroacetic acid in acetonitrile; ¯ow rate,

0.75 mLámin)1; gradient, 20±40% B in 20 min; UV

detec-tion, 214 nm

Peptide analysis

Peptides were analysed by electrospray MS using a Sciex

API III (Perkin Elmer) mass spectrometer Alternatively,

MALDI MS was performed with a LaserTec RBT II

(PerSeptive Biosystems, Framingham, MA) Amino-acid

sequencing was carried out with a gas phase sequencer

(model 473 A, Applied Biosystems) The purity of peptides

was analysed by capillary zone electrophoresis (CZE) with a

P/ACE 2100 CZE system (Beckman)

Peptide synthesis

Peptides were synthesized using Fmoc solid-phase chemistry

on a preloaded TentaGel S Trt resin (Rapp Polymere,

TuÈbingen, Germany) Crude peptides were puri®ed by reversed-phase HPLC (Vydac C18, 10 lm, 300 AÊ; gradient, 10±70% B in 30 min; eluent A, 0.07% tri¯uoroacetic acid/water; eluent B, 0.05% tri¯uoroacetic acid in acetonit-rile/water 4 : 1; ¯ow rate, 0.8 mLámin)1; UV detection, 215/230 nm) The purity and identity of synthesized peptides were checked by analytical HPLC, MS and sequence analysis

Bacterial growth assays All bi®dobacteria strains (Table 1) were purchased from the Deutsche Sammlung fuÈr Zellkulturen und Mikroorganis-men (Braunschweig, Germany) All other microorganisms (Escherichia coli ATCC 11229, Enterococcus faecalis ATCC

29212, Clostridium dicile ATCC 9684 and Candida albicans ATCC 10231) were tested at the Hygiene Institute

of the University of Heidelberg, Germany For growth assays, peptides were added to 200 lL of 29 gáL)1Elliker broth (Difco), inoculated with 25 lL of bacterial culture (A50of 0.05)

1 The samples were incubated at 37 °C under anaerobic conditions for 16±48 h (according to the strain) Bacterial growth was monitored by photometric absorbance measurements at 550 nm and by microscopy Pepsin-treated

Fig 1 Puri®cation of bi®dogenic peptides from human milk by consecutive chromatographic steps (A) Reversed-phase HPLC separation of 20 mL pepsin-treated human milk The black bars indicate the increased cell density of B bi®dum DSM 20082 cultures supplemented with the peptides from HPLC fractions corresponding to a 0.5 mL aliquot of milk The fraction exhibiting highest bi®dogenic activity, indicated by an arrow, was selected for further puri®cation (B) Reversed-phase chromatography of the selected HPLC fraction from (A) The fraction stimulating most e€ectively the growth of B bi®dum is indicated by a black arrow and was selected for the next separation step (C) Cation-exchange chromato-graphy of the selected HPLC fraction from (B) resulted in the puri®cation to homogeneity of the two bi®dogenic peptides hLACFR-Ib and hLACFR-Ic (D) Purity of the isolated peptides was proved by CZE As an example the electropherogram of the peptide hLACFR-Ic is shown (E) MALDI-MS analysis revealed a molecular mass of 5801 Da for the peptide hLACFR-Ic, concurring with the calculated molecular mass of

5798 Da The variation of 3 Da is within the accuracy of mass measurement (0.5%).

human milk (10%, v/v) and 100 lMN-acetylglucosamine,

both of which stimulate the growth of bi®dobacteria, were

used as positive controls

R E S U L T S

Identi®cation of human milk peptides that stimulate

the growth ofB bi®dum

Peptides capable of stimulating the growth of bi®dobacteria

were puri®ed from pepsin-treated human milk and

colo-strum by means of reversed-phase and cation-exchange

HPLC A typical reversed-phase separation of

pepsin-treated human milk is shown in Fig 1A For detection of

growth promoting activity within HPLC fractions B

bi®-dum DSM 20082 was used as a test strain The HPLC

fraction exhibiting highest growth stimulatory activity was

selected for further puri®cation In a second reversed-phase separation step bi®dogenic peptides were enriched in one major peak (Fig 1B) Use of cation-exchange chromato-graphy resulted in the effective separation of two bi®dogenic peptides from inactive components (Fig 1C) Purity of the active peptides was proved by CZE (Fig 1D) MALDI MS analysis revealed a molecular mass of 5801 Da for one of the puri®ed peptides (Fig 1E) and 5584 Da for the other (data not shown) Edman degradation determination of the primary structure of the isolated peptides (Fig 2A) and subsequent database analysis (SwissProt) led to the identi®cation of two lactoferrin fragments, termed hLACFR-Ib and hLACFR-Ic

Using this puri®cation procedure, several peptides that exhibited bi®dobacterial growth stimulatory activity were isolated from mature human milk and from colostrum From mature human milk the highly bi®dogenic peptides

Table 1 Susceptibility of di€erent bi®dobacteria towards bi®dogenic compounds The spectrum of growth stimulation is di€erent for the bi®dogenic compounds tested, which were pepsin-treated human milk (10%, v/v), 100 l M N-acetylglucosamine, and 1 l M of the peptides hLACFR-Ia, hLACFR-Ic, hLACFR-IIa, and PRELP-I A Ô+Õ indicates that the application of a compound led to an increased cell density of at least + 50% after 16±48 h incubation in 29 gáL )1 Elliker broth.

Microorganism

Bi®dogenic compounds Human milk N-acetylglusamine hLACFR-Ia hLACFR-Ic hLACFR-IIa PRELP-I

Fig 2 Primary structure of isolated human bi®dogenic peptides (A) The human bi®dogenic peptides are derived from the precursor proteins lactoferrin (hLACFR) and the secretory component of the polyimmunoglobulin receptor (hPIGR) hLACFR-Ia, -Ib, and -Ic are from the N-terminus and hLACFR-IIa and -IIb are from equivalent positions of the C-terminus of lactoferrin hLACFR-Ia corresponds to the generally known lactoferrin sequence hLACFR-Ib is obtained from a lactoferrin variant with an arginine deletion at position 3 and hLACFR-Ic is derived from a lactoferrin variant with two amino-acid residues exchanged (B) Partial sequence alignment of the bi®dogenic peptides hPIGR-I,

hLACFR-Ia, and hLACFR-IIa Identical residues are hatched with dark grey The residues probably related to bi®dogenic activity and used for the design of PRELP-I are marked with arrows.

hLACFR-Ia, hLACFR-Ib, and hLACFR-Ic (Fig 2A)

were puri®ed These peptides are fragments derived from

the N-terminus of three different lactoferrin variants In the

milk from one donor, the peptides hLACFR-Ib and

hLACFR-Ic were found, and from a second donor, the

peptides hLACFR-Ia and hLACFR-Ic were isolated

Further puri®cation led to the identi®cation of the

lactofer-rin fragments hLACFR-IIa and hLACFR-IIb, which

are derived from the lactoferrin C-terminus (Fig 2A)

From pepsin-digested colostrum, the bi®dogenic peptides

hPIGR-Ia and hPIGR-Ib were puri®ed; these represent

fragments of the soluble polyimmunoglobulin receptor

(hPIGR) (Fig 2A) In mature human milk hPIGR-Ia and

hPIGR-Ib were not detectable corresponding to the tenfold

lower amount of hPIGR in mature milk

Pepsin-treated bovine milk also exhibits bi®dogenic

activity at about half the level of that of mature human

milk (data not shown) However, no bi®dogenic peptides

were identi®ed in bovine mild by reversed-phase HPLC of

pepsin-treated bovine milk

Bi®dogenic activity of human milk peptides

In dose±response studies using the pure human peptides

hPIGR-Ib, hLACFR-Ia, hLACFR-Ib, hLACFR-Ic, and

hLACFR-IIa, a dose-dependent growth stimulatory activity

on B bi®dum DSM 20082 was con®rmed Each of the

puri®ed peptides stimulates B bi®dum by 50±250% at a

concentration of 0.5±3 lM(Fig 3)

The puri®ed milk peptides promote the growth of several

strains of bi®dobacteria that usually colonize infants'

intestines (Table 1) The peptides hLACFR-Ia and

hLACFR-Ic stimulated ®ve out of the seven bi®dobacterial

strains tested signi®cantly and the peptide hLACFR-IIa

stimulated four out of the seven In contrast, the known

bi®dogenic carbohydrate N-acetylglucosamine exhibits

growth promoting activity only towards strains of B

bi®-dum The growth stimulatory activity of

N-acetylglucos-amine was detected at a minimum concentration of 50 lM, which is 100-fold higher than the minimum effective concentration of the puri®ed bi®dogenic peptides

Structural characteristics of bi®dogenic peptides All bi®dogenic peptides identi®ed share a similar primary structure containing a disul®de bond, and a small hydro-phobic domain located C-terminally next to the cysteines (Fig 2B) Based on these structural characteristics, a small peptide named prebiotic lactoferrin-derived peptide-I (PRELP-I, Fig 2A) was designed and chemically synthe-sized PRELP-I exhibits a growth stimulatory effect on

B bi®dum DSM 20082, B bi®dum DSM 20215, B infantis and B breve in the same order of magnitude as the peptides isolated previously (Fig 3) In contrast, the growth of other microorganisms capable of colonizing the human intestine, such as C dicile, E coli, E faecalis and C albicans, was not affected signi®cantly by the synthetic peptide using the same anaerobic growth conditions used for bi®dobacteria (Fig 4)

To con®rm that the growth stimulatory activity of bi®dogenic peptides is related to their peptidic character, control experiments were performed No bi®dogenic activity was observed when applying amino-acid mixtures representing the amino-acid composition of the puri®ed peptides, simple disul®de-containing compounds or bi®do-genic peptides after chemical reduction of disul®de bonds Hydrolysis with the nonspeci®c protease subtilisin led to inactivation of hLACFR-Ia, hLACFR-IIa, and PRELP-I (data not shown) In contrast, treatment of hLACFR-Ia, hLACFR-IIa, and PRELP-I with pepsin, trypsin or chymotrypsin did not lead to the loss of bioactivity, indicating the stability of the peptides towards gastroin-testinal proteases

D I S C U S S I O N

Previous reports have shown that proteolytic milk hydro-lysate is an effective stimulant for the growth of bi®dobac-teria [10,14] However, the molecular structures of the substances responsible for this effect have been hitherto unknown In the present work, we show that proteolytic fragments of major human milk proteins are effective growth factors for bi®dobacteria Furthermore, it is shown that the bi®dogenic effect of human milk is not based entirely on their carbohydrate content, as the components isolated contain no glycosyl moieties

The isolated fragments of the polymeric immunoglobulin receptor secretory component (hPIGR) and of human lactoferrin (hLACFR) stimulate the bi®dobacterial growth even in the presence of an extraordinarily rich growth medium These peptides were shown to be 100 times more effective on a molar basis than the carbohydrate N-acetylglusosamine, a typical currently known bi®dogenic carbohydrate [9] (data not shown) Moreover, they stimu-late a larger set of different bi®dobacterial strains (Table 1) All of these results together show that the peptides described

in this study act very effectively as prebiotic growth factors Interestingly, no other bacterial species was found to be stimulated in a similar manner, which indicates that the action of these peptides is highly speci®c to bi®dobacteria It can be assumed that the bi®dogenic activity of breast milk

Fig 3 Dose±response of bi®dogenic peptides The selected bi®dogenic

peptides exhibit a dose-dependent growth stimulatory e€ect on

B bi®dum The plots show the relative cell density of B bi®dum DSM

20082 cultivated for 16 h in 29 gáL )1 Elliker broth supplemented

with: (A) hLACFR-Ia, -Ib, -Ic; (B) hPIGR-Ib; (C) PRELP-I;

(D) hLACFR-IIa.

based on peptides exceeds the relevance of the bi®dogenic

activity of milk carbohydrates

There are several lines of evidence that the characterized

peptidic bi®dus factors are important for the

establish-ment of the bi®dobacterial ¯ora in the large bowel of

breast-fed infants One ®nding is that peptide bi®dus

factors selectively stimulate bi®dobacterial strains usually

found in infants' intestine Furthermore, the time course

of the synthesis of the precursor molecules of peptide

bi®dus factors is coupled to the bi®dogenic activity of

human milk; the bi®dogenic activity of colostrum is higher

than that of mature milk [15], which coincides with the

10-fold increased excretion rate of the precursor proteins

of peptidic bi®dus factors immediately after birth

(rev-iewed in [16]) As a result, the bi®dogenic activity is

increased during the critical initial microbial colonization

process of the infants' large bowel In cow's milk, a

signi®cantly lower bi®dogenic activity is present because it

lacks peptidic and carbohydrate bi®dus factors [16] As a

result, feeding with unsupplemented cow's milk-based

formulas does not induce the typical bi®d ¯ora observed

in breast-fed infants [17]

Sequence comparison of the bi®dogenic peptides

identi-®ed a common structural motif (Fig 2B) that contains a

pair of cysteine residues forming a disul®de bond and two

small hydrophobic domains located C-terminally to the two

cysteines From this observation, a simple peptide with the

arti®cial sequence CAVGGGCIAL was designed and

tested This peptide exhibits bi®dogenic activity comparable

to those of the isolated peptides The activity is related to the

peptidic character, as the nonspeci®c protease subtilisin

degraded bi®dogenic activity and amino-acid mixtures had

no profound activity In addition, presence of the disul®de

bridge is essential as its chemical reduction leads to the

complete loss of activity The importance of the integrity of

the disul®de bond for bi®dogenic acitivity is in accordance

with the ®ndings of Poch and Bezkorovainy [18] and

Ibrahim and Bezkorovainy [19] Poch and Bezkorovainy

showed that bovine milk j-casein and hog gastric mucin lost their bi®dobacterial growth-promoting activities following reduction-alkylation or oxidation of their disul®de bonds Consistent with our results they also showed that simple disul®de-containing compounds exhibit no bi®dobacterial growth stimulating activity This adds to the ®ndings of Ibrahim and Bezkorovainy who suggest that the diverse bi®dobacterial growth-promoting factors reported have a common component, namely sulfur containing peptides

In contrast with these ®ndings Etoh et al [20] identi®ed a decapeptide with growth promoting activity towards

B bi®dum, which contains neither cystine nor cysteine Therefore, they suggest that sulfhydryl groups are not indispensable to growth-stimulating activity However, they

do not indicate in which concentration the decapeptide is active and no experiments were performed evaluating its speci®city for bi®dobacteria or the importance of the peptidic structure for the bi®dogenic activity Therefore, it

is possible that the decapeptide is responsible for bacterial growth stimulation as a nutritional source, in contrast with the bi®dus factors reported previously [18,19] and the bi®dogenic milk peptides presented in this study

The precursor proteins of peptide bi®dus factors are known to have a protective value for the suckling infant The polymeric immunoglobulin receptor precursor consists

of the secretory component that is usually bound to the immunoglobulins present in milk [21] Immunoglobulins are highly protective due to their capability to inactivate intraintestinal pathogens Our ®nding, that a portion of the secretory immunoglobulins is a prebiotic factor for bene®cial bacteria, reveals a novel function for this complex

As a result, immunoglobulins act protectively by regulating the infants' intestinal micro¯ora A further precursor protein yielding peptidic bi®dus growth factors is lactofer-rin Interestingly, the potential protective role of lactoferrin regulating the intestinal micro¯ora has already been discussed extensively according to its antimicrobial activity [22±24] There are antimicrobial lactoferrin fragments called

Fig 4 Growth in¯uencing activity of the synthetic peptide PRELP-I towards intestinal bacteria PRELP-I stimulates the growth of bi®dobacteria but does not a€ect the growth of other microorganisms that colonize the human intestine The plot shows the bacterial growth of B infantis, Clostridium dicile, Escherichia coli, Enterococcus faecalis, and Candida albicans cultivated for 20 h (72 h Clostridium dicile) under anaerobic conditions in 29 gáL )1 Elliker broth supplemented with PRELP-I at di€erent concentrations.

lactoferricines [25,26], which are structurally related to the

bi®dogenic peptides hLACFR-Ia, -Ib, and -Ic The

anti-microbial activity of lactoferrin described previously and

our results provide strong evidence that this protein is also

important for the determination of the intraintestinal

bacterial ¯ora of the newborn by combining the same two

activities observed for the secretory immunoglobulins;

antibiosis and prebiosis

All of the isolated lactoferrin fragments stem from two

equivalent positions within the lactoferrin two-lobe

struc-ture, which is based on an ancient sequence duplication

within lactoferrin and, obviously, yields two domains with

similar functions: iron binding and in¯uencing bacterial

growth Our ®ndings reveal that human milk contains

three slightly different variants of lactoferrin, each

corre-sponding to one of the isolated peptides hLACFR-Ia,

hLACFR-Ib, and hLACFR-Ic No description of the

proteins corresponding to hLACFR-Ib and hLACFR-Ic

has yet been published, although the cDNAs of these

variants have already been cloned (GenBank accession

numbers M73700, M93150) The difference in these

lactoferrin variants may be of physiological relevance

because previous reports show that N-terminal variations

of lactoferrin in¯uence its antimicrobial properties,

lipopolysaccharide binding and modulate cellular

immuni-ty reactions [27] In this work, however, a signi®cant

differences in the bi®dogenic activity of the variant peptides

was not observed

Several lines of evidence may answer the important

question of whether the bi®dogenic peptides identi®ed can

reach the colon of the suckling infant The precursor

proteins of peptide bi®dus factors are major components of

human milk Therefore, high amounts of bioactive peptide

fragments can be liberated during digestion Furthermore,

there is evidence that a signi®cant amount of bioactive

peptides can successfully passage through the small

intes-tine, as the structure required for bioactivity is not degraded

by the gastrointestinal proteases pepsin, trypsin or

chym-otrypsin This is also supported by the observation that the

entire lactoferrin molecule and larger fragments are present

in the stools of infants [28]

Our results demonstrate a novel function of a major

human milk protein; a direct and selective growth

stimulation of bi®dobacteria coupled to antimicrobial

properties of the proteins Bovine milk, a major part of

formula diets, has a signi®cantly lower bi®dobacteria

growth-stimulatory activity Hence, an additional

expla-nation as to why breast-feeding is bene®cial for the

suckling infant is provided, re¯ecting the different

gastro-intestinal microbiology and physiology of humans and

ruminants The composition of formula diets should

re¯ect these requirements Simple bi®dogenic peptides,

such as the peptide PRELP-I might be useful as an

additive to infant formulas to improve their value by

supporting the implantation of bi®dobacteria, which are

bene®cial to the suckling infant by competing with

pathogenic microorganisms

A C K N O W L E D G E M E N T S

The authors thank Prof Dr Geiss (Hygiene Institute, University of

Heidelberg, Germany) for technical support This work was supported

by the Deutsche Forschungsgemeinschaft, DFG (ZU 98/2±2).

R E F E R E N C E S

1 Kleessen, B., Bunke, H., Tovar, K., Noack, J & Sawatzki, G (1995) In¯uence of two infant formulas and human milk on the development of the faecal ¯ora in newborn infants Acta Paediatr.

84, 1347±1356.

2 Stark, P.L & Lee, A (1982) The microbial ecology of the large bowel of breast-fed and formula-fed infants during the ®rst year of life J Med Microbiol 15, 189±203.

3 France, G.L., Marmer, D.J & Steele, R.W (1980) Breast-feeding and Salmonella infection Am J Dis Child 134, 147±152.

4 Larsen, S.A & Homer, D.R (1978) Relation of breast versus bottle feeding to hospitalization for gastroenteritis in a middle-class U.S population J Pediatr 92, 417±418.

5 Golding, J., Emmet, P.M & Rogers, I.S (1997) Gastroenteritis, diarrhoea and breast feeding Early Hum Dev 49, S83±S103.

6 Gibson, G.R & Wang, X (1994) Regulatory e€ects of bi®do-bacteria on the growth of other colonic bi®do-bacteria J Appl Bacteriol.

77, 412±420.

7 Bielecka, M., Biedrzycka, E., Biedrzycka, E., Smoragiewicz, W & Smieszek, M (1998) Interaction of Bi®dobacterium and Salmonella during associated growth Int J Food Microbiol 45, 151±155.

8 Saavedra, J.M., Bauman, N.A., Oung, I., Perman, J.A & Yolken, R.H (1994) Feeding of Bi®dobacterium bi®dum and Streptococcus thermophilus to infants in hospital for prevention of diarrhoea and shedding of rotavirus Lancet 344, 1046±1049.

9 Beerens, H., Romond, C & Neut, C (1980) In¯uence of breast-feeding on the bi®d ¯ora of the newborn intestine.

Am J Clin Nutr 33, 2434±2439.

10 Kehagias, C., Jao, Y.C., Mikolajcik, E.M & Hansen, P.M.T (1977) Growth response of Bi®dobacterium bi®dum to a hydro-lytic product isolated from bovine casein J Food Sci 42, 146±150.

11 Bezkorovainy, A & Topouzian, N (1981) Bi®dobacterium bi®dus var pennsylvanicus growth promoting activity of human milk casein and its derivatives Int J Biochem 13, 585±590.

12 Balmer, S.E., Scott, P.H & Wharton, B.A (1989) Diet and faecal

¯ora in the newborn: lactoferrin Arch Dis Child 64, 1678±1684.

13 Petschow, B.W & Talbott, R.D (1990) Growth promotion of Bi®dobacterium species by whey and casein fractions from human and bovine milk J Clin Microbiol 28, 287±292.

14 Raynaud, M & Bizzini, B (1971) Puri®cation et proprieÂteÂs du facteur bi®dus 2 Ann Nutr Alimen 25, 209±223.

15 Gyorgy, P (1953) A hitherto unrecognized di€erence between human milk and cow's milk Pediatrics 11, 98.

16 Blanc, B (1981) Biochemical aspects of human milk ) comparison with bovine milk World Rev Nutr Diet 36, 1±89.

17 Yoshioka, H., Iseki, K & Fujita, K (1983) Development and di€erences of intestinal ¯ora in the neonatal period in breast-fed and bottle-fed infants Pediatrics 72, 317±321.

18 Poch, M & Bezkorovainy, A (1991) Bovine milk j-casein trypsin digest is a growth enhancer for the genus Bi®dobacterium J Agric Food Chem 39, 73±77.

19 Ibrahim, S.A & Bezkorovainy, A (1994) Growth-promoting factors for Bi®dobacterium longum J Food Sci 59, 189±191.

20 Etoh, S., Asamura, K., Obu, A., Sonomoto, K & Ishizaki, A (2000) Puri®cation and identi®cation of a growth-stimulating peptide for Bi®dobacterium bi®dum from natural rubber serum powder Biosci Biotechnol Biochem 64, 2083±2088.

21 Kaetzel, C.S., Blanch, V.J., Hempen, P.M., Philips, K.M., Piskurich, J.F & Youngman, K.R (1997) The polymeric immu-noglobulin receptor: structure and synthesis Biochem Soc Trans.

25, 475±480.

22 Wharton, B.A., Balmer, S.E & Scott, P.H (1994) Faecal ¯ora in the newborn E€ect of lactoferrin and related nutrients In Lac-toferrin: Structure and Function (Hutchens, T.W., Rumball, S.V & LoÈnnerdal, B., eds), pp 91±98 Plenum Press, New York, London.

23 Heine, W (1996) Ein¯uûfaktoren verschiedener

Nahrungsfor-men auf die intestinale Mikro¯ora Monat Kinderhkd 144,

169±173.

24 Sanchez, L., Calvo, M & Brock, J.H (1992) Biological role of

lactoferrin Arch Dis Child 67, 657±661.

25 Bellamy, W., Takase, M., Yamauchi, K., Wakabayashi, H.,

Kawase, K & Tomita, M (1992) Identi®cation of the bactericidal

domain of lactoferrin Biochim Biophys Acta 1121, 130±136.

26 Tomita, M., Takase, M., Wakabayashi, H & Bellamy, W (1994)

Antimicrobial peptides of lactoferrin In Lactoferrin: Structure and

Function (Hutchens, T.W., Rumball, S.V & LoÈnnerdal, B., eds),

pp 209±218 Plenum Press, New York, London.

27 Berkel, V.P.H.C., Geerts, M.E.J., Veen, V.H.A., Mericskay, M., Boer, d H.A & Nuijens, J.H (1997) N-terminal stretch Arg 2 , Arg 3 , Arg 4 and Arg 5 of human lactoferrin is essential for binding

to heparin, bacterial lipopolysaccharide, human lysozyme and DNA Biochem J 328, 145±151.

28 Goldman, A.S., Garza, C., Schanler, R.J & Goldblum, R.M (1990) Molecular forms of lactoferrin in stool and urine from infants fed human milk Pediatr Res 27, 252±255.