PLoS ONE. Dec/31/2010; 6(4)

Published online Apr/26/2011

PMID: 21556375

PMC: 3083434

doi: 10.1371/journal.pone.0019433

The Host Defense Proteome of Human and Bovine Milk

Vladimir N. Uversky

Abstract

Introduction

Milk is the single source of nutrients for the newborn mammal. The composition ofmilk of different mammals has been adapted during evolution of the species tofulfill the needs of the offspring. Milk not only provides nutrients, but it alsoserves as a medium for transfer of host defense components to the offspring. Thehost defense proteins in the milk of different mammalian species is expected toreveal signatures of evolution. Proteins are a major contributor to host defensecomponents in milk [1], [2]. In humans, a positive relation between breastfeeding andhealth of babies has been noted from the time of the first recorded use ofhuman-milk substitutes, going back thousands of years [3].

Because bovine milk is used as a substitute for human milk, it is important to knowthe differences in host defense proteins between human and bovine milk. Despite thedescription of several differences between human and bovine milk, there is limitedknowledge on differences in the host defense proteome. A recent overview comparedthe human and bovine milk proteome [4]. Data were collected,however, from studies using various types of samples and analytical techniques. Dataon the presence of cytokines and hormones, for example, were available only forhuman milk and not for bovine milk. As a result, we now only have limited knowledgeon differences in host defense proteome between human and bovine milk.

To study the milk proteome, milk is usually separated into three protein fractions:caseins, serum, and milk fat globule membrane (MFGM) [5], [6]. As a start, the whole milk isseparated in cream and skim milk. The cream contains the milk fat, which is presentin globules. These globules consist of a triglyceride core surrounded by the MFGM,derived from the apical membrane of the milk-producing epithelial cells [7]. The proteincomponent of the MFGM (about 1–4% of total milk protein) can beisolated from the cream. The skim milk can be centrifuged to obtain a casein pelletand a supernatant containing the serum proteins. The MFGM and serum proteinfractions, which contain the low-abundance proteins from milk, can then be used forproteomic analyses.

In this study, we compared the proteomes of serum and MFGM from human and bovinemilk, with the aim to determine differences in host defense proteomes. The overlapas well as the difference we found in the host defense proteomes increases ourunderstanding of human and bovine milk. This knowledge will help to identify theproteins responsible for immunity-promoting properties of milk for theoffspring.

Results and Discussion

We identified a total of 268 proteins in human milk and 269 proteins in bovine milk,of which 147 proteins were found in both species (Table 1). We identified a larger number ofproteins in milk then has been published previously. Most studies used excision ofspots on 2D-gels, followed by mass-spectrometry e.g. [5], [8], [9]. With this 2D-gel method onlyexcised spots are analyzed. With our 1D-gel method, however, we analyzed the wholegel lane and did, thus, not rely on visible protein staining. In addition, our1D-gel method is more suitable for analyzing membrane proteins, which are ubiquitousin MFGM [10].The same 1D-gel method, was recently used for studying the proteome of bovine milkserum [11] andbovine MFGM [12].

10.1371/journal.pone.0019433.t001

Table 1

Number of total, serum, and milk fat globule membrane (MFGM) proteins inhuman and bovine milk.

Proteins	Human	Bovine	Common
Total	268	269	147
Serum	222	192	105
MFGM	234	232	118

In bovine serum, we identified a total of 192 proteins. Previously, 148 proteins wereidentified in bovine milk serum [11]; 132 of these were also identified here. In bovine MFGM,we identified 232 proteins while in a previous study only 116 proteins wereidentified [12]; 95 of these were also identified here. Both comparisons showthat our approach enabled us not only to identify about 90% of the alreadyreported proteins but also to nearly double the number of identified proteins. Manyof the newly identified proteins in our study were enzymes, that usually occur atlow concentration. This suggest that the increase in number of identified proteinscan be explained by the higher sensitivity of our method compared with previousmethods.

The identified proteins were categorized according to their GO annotation (Table 2). Of all the proteinsannotated, 44 proteins in human milk and 51 proteins in bovine milk were related toa host defense function. Although the total number of host defense proteins wassimilar in both milk samples, the predicted function of the individual proteinsdiffered between species. Bovine milk, for example, contained a wider range ofantibacterial proteins, whereas human milk contained a wider range ofimmunoglobulins.

10.1371/journal.pone.0019433.t002

Table 2

Number of protein functions according to GO annotation in human andbovine milk.

Function	Human	Bovine	Common
Cell wall/cell adhesion	21	17	8
Coagulation	3	7	3
Cytoskeleton	12	8	7
Enzymes	70	50	25
Host defense	44	51	33
Other	18	13	9
Protease inhibitor	12	15	8
Protein synthesis/chaperone	11	9	4
Signaling	15	19	7
Transport	48	64	39
Unknown	14	16	4

So far, we have reported qualitative differences in the proteome of human and bovinemilk. For a better understanding of the biological differences between milk of thesespecies, we also performed a quantitative analysis of the host defense proteome. Forquantification, a filter-based sample preparation method was used, as this allows amore reproducible quantification compared to gel-based methods. The relative proteinconcentrations of host defense proteins in human and in bovine milk is shown inTable 3. Some host defenseproteins were detected only with the qualitative (gel-based) method and not with thequantitative (filter-based) method (Table 3). The failure to detect certain proteins with the quantitativemethod is caused by its lower sensitivity compared with the qualitative method.

10.1371/journal.pone.0019433.t003

Table 3

Presence and relative concentration of host defense proteins in human andbovine milk serum and in human and bovine milk fat globule membrane(MFGM).

Gene code	Protein	Human serum	Bovine Serum	Human MFGM	Bovine MFGM
A1BG	Alpha-1B-glycoprotein	<1	<1	<1	<1
AGP/ORM1	Alpha-1-acid glycoprotein	<1	<1	<1	9*
B2M	Beta-2-microglobulin	94	61	<1	<1
C3	Complement component C3	65	121	12	26
C4A	Complement component C4A	21	<1	12*	<1
C4BPA	C4b-binding protein alpha chain	ND	<1	ND	<1
C6	Complement component C6	ND	<1	ND	ND
C7	Complement component C7	<1	<1	ND	<1
C9	Complement component C9	ND	<1	ND	<1
CAPG	Macrophage-capping protein	<1	ND	ND	ND
CATHL1	Cathelicidin-1	ND	<1	ND	189*
CATHL2	Cathelicidin-2	ND	ND	ND	122*
CATHL4	Cathelicidin-4	ND	ND	ND	13*
CATHL6	Cathelicidin-6	ND	ND	ND	88*
CATHL7	Cathelicidin-7	ND	ND	ND	<1
CD14	Monocyte differentiation antigen CD14	262*	5	146*	31
CD46	Membrane cofactor protein precursor	ND	<1	ND	15*
CD59	MAC-inhibitory protein	<1	ND	229	133
CD81	CD81 antigen	<1	<1	<1	<1
CD5L	CD5 antigen-like	ND	<1	ND	<1
CFB	Complement factor B (Fragment)	<1	<1	<1	<1
CFI	Complement factor I	<1	<1	<1	<1
CLU	Clusterin	151*	<1	672*	<1
CRISP3	Cysteine-rich secretory protein 3	ND	19*	ND	<1
CTSS	Cathepsin S	<1	ND	<1	ND
DCD	Dermicidin	102	61	151*	<1
DEFA3	Neutrophil defensin 3	ND	ND	<1	ND
ERAP1	Endoplasmic reticulum aminopeptidase 1	ND	<1	ND	<1
GLYCAM1	Glycosylation-dependent cell adhesion molecule 1	<1	3294*	11	2565*
HF1	Complement factor H	ND	<1	ND	<1
IGHA	Immunoglobulin alpha chain C region	4566*	<1	493*	<1
IGHG	Immunoglobulin gamma chain C region	127*	<1	112*	<1
IGJ	Immunoglobulin J chain	616*	<1	<1	<1
IGK	Immunoglobulin kappa chain C region	1285*	<1	<1	<1
IGKV	Immunoglobulin kappa chain C region	<1	ND	<1	ND
IGLC	Immunoglobulin lambda chain C region	115*	ND	<1	ND
IGLV	Immunoglobulin lambda chain V region	<1	ND	<1	ND
IGHM	Immunoglobulin mu chain C region	<1	220*	<1	214*
LBP	Lipopolysaccharide-binding protein precursor	ND	<1	ND	<1
LPO	Lactoperoxidase	20	161*	<1	10*
LTF	Lactoferrin	11182*	181	7045*	59
LYZ	Lysozyme C	3274*	<1	674*	<1
MFGE8	Milk fat globule-EGF factor 8	31	57	326	2663*
MUC1	Mucin-1	<1	<1	72	181
MUC4	Mucin-4	<1	ND	70*	ND
MUC15	Mucin-15	ND	<1	ND	213*
MUC16	Mucin-16	ND	<1	ND	<1
IPI00712983	Mucin-20-like	ND	<1	ND	<1
PIGR	Polymeric immunoglobulin receptor	2745*	422	215	799*
PSME2	Proteasome activator complex subunit 2	ND	ND	<1	ND
S100A8	S100 calcium-binding protein A8 (Calgranulin-A)	<1	ND	<1	<1
S100A9	S100 calcium-binding protein A9 (Calgranulin-B)	ND	ND	<1	<1
S100A12	S100 calcium-binding protein A12 (Calgranulin-C)	ND	ND	ND	<1
SAA3	Serum amyloid A protein	ND	ND	ND	<1
SCFV	Single-chain Fv	<1	ND	<1	ND
SERPINA1	Alpha-1-antitrypsin	31	21	<1	<1
SERPINA3	Alpha-1-antichymotrypsin	250*	<1	<1	<1
SERPING1	Plasma protease C1 inhibitor	<1	<1	<1	<1
SPP1	Osteopontin	762	451	42	78
TLR2	Toll-like receptor 2	ND	<1	27	31
VTN	Vibronectin	ND	ND	<1	ND
XDH	Xanthine dehydrogenase/oxidase	282	243	1084	1457

Numbers are averaged peak heights of the three most abundant peptides(arbitrary units).

* significantly higher (p<0.05).

<1: Detected with the qualitative method, but not the quantitativemethod

ND: Not detected using either qualitative or quantitative method

Immunoglobulins are the most abundant group of host defense proteins in human milkserum. A wider range as well as a larger amount of immunoglobulins was identified inthe serum fraction of human milk compared with bovine milk (Table 3). Bovine colostrum was found to containsimilar amounts of immunoglobulins as human colostrum [13]. The concentration ofimmunoglobulins in bovine milk declines faster after the first days of lactationthan human milk [13], [14]. Our analysis showed that immunoglobulin A (IgA) was themost abundant immunoglobulin in human milk (Table 3; gene code: IGHA). In other studies, IgAwas also found to be the most prominent immunoglobulin in milk [15], [16]. This relatively high IgAconcentration in human milk has been linked to the absence of this immunoglobulin inthe intestine of the newborn baby [16]. It is also known that already at the age of 4 days, acalf is able to produce IgA in its intestine [17], which probably explains therelatively low IgA concentration in mature bovine milk. The high concentration ofpolymeric immunoglobulin receptor (PIGR) found in human milk serum (Table 3) can be related to thehigh IgA concentration, because PIGR is used for the transcytosis of IgA from thebasolateral to the apical side of epithelial cells [18].

The newborn human is also known be deficient in CD14, which is part of the Toll-likereceptor (TLR)-4 complex [16]. The TLR-4 complex can detect lipopolysaccharides ongram-negative bacteria and subsequently activate the innate immune system. CD14 is,therefore, important for protection against pathogen invasion [16], [19]. CD14 has been shown to bepresent in human milk, with the highest concentration being found in colostrum [19]. Bovinecolostrum contains similar amounts of CD14 as human colostrum [19]. Although CD14 was not detectedby them in commercial bovine milk [19], we detected CD14 in unprocessed bovine milk serum andMFGM (Table 3). Absence ofCD14 in the previous study may be related to heating of their milk, a treatmentwhich we did not apply to our samples.

IgA and CD14 are important proteins for the mucosal immune system [20], [21]. Alsolactoferrin (LTF) and lysozyme (LYZ) play an important role in the mucosal immunesystem [20], [21]. We foundthat the concentration of these two antibacterial proteins is much higher in humanmilk than in bovine milk (Table3), which is consistent with literature [22]. LTF was relatively abundant inthe MFGM fraction of human milk (Table 3), which may seem remarkable for a secreted protein. A previousstudy, however, found that part of the LTF in human milk was strongly bound to theMFGM membrane [23].This finding may be related to the defense of the epithelial membrane of the mammarygland, as MFGM originates from the epithelial membrane. Additionally, themembrane-bound LTF may have a host defense function in the newborn. LTF and LYZ havebeen shown to be more abundant in colostrum than in mature milk for humans andbovines. The differences in their concentration in colostrum of humans and bovinesis smaller than between the mature milks [22], [24]. The four proteins (IgA, CD14,LTF, and LYZ) described above are all part of the mucosal immune system. The newbornhuman is deficient in two of them (IgA and CD14) during infancy [16], whereas thecalf is not [17].Although the concentration of these two proteins is similar in bovine and humancolostrum [14],[19], our datashow a higher concentration of these components in mature human milk compared withmature bovine milk. This higher concentration in human milk may be related todifferences in maturation of the immune system between babies and calves.

Clusterin is another protein that is more abundant in human milk than in bovine milk.Clusterin, a highly glycosylated protein that is also known as apolipoprotein J, isone of the most abundant proteins in the human MFGM fraction (Table 3). Although its function is not completelyclear, clusterin has been linked to cell damage and apoptosis and has been shown tobe overexpressed at damaged or stressed tissues and to provide a chaperone-likeactivity to protect other proteins against damage [8]. Milk fat globule-EGF factor8 (MFGE8) is a protein that has a similar function as clusterin [25]. Our data showsthat MFGE8 is more abundant in bovine milk than in human milk is (Table 3). MFGE8, known also aslactadherin and PAS-6/PAS-7, is a glycoprotein, like clusterin, but its function isnot completely clear; however, it has been linked to cell damage and apoptosis [25], [26]. It was shownthat MFGE8 plays an important role in the maintenance of intestinal epithelialhomeostasis and the promotion of mucosal healing [25]. It may be an important milkprotein, therefore, for protecting the intestinal tract of the newborn. Thisprotective effect may be related to the finding that MFGE8 is a protein that linksto apoptotic cells so they can be recognized by phagocytes for engulfment [26]. This effecton apoptotic cells corresponds to the finding that MFGE8 was upregulated ininvoluting mammary glands, where they undergo a substantial increase in the rate ofepithelial cell apoptosis [27]. The presence of a high concentration of clusterin inhuman milk and of MFGE8 in bovine milk may thus coincide, because these proteinshave a similar function.

Our results also show that bovine milk contains a large amount ofglycosylation-dependent cell adhesion molecule 1 (GlyCAM1). This proteins is themost abundant host defense protein in bovine milk serum (Table 3). GlyCAM1, known also as lactophorin andPP3, consists of a diverse group of glycoproteins/glycopeptides. GlyCAM1 is amucin-like antibacterial component expressed at the membrane of epithelial cells ofthe mammary gland. The active site of this membrane-bound GlyCAM1, however, isabsent in the secreted form of the protein, as found in milk serum or whey [28]. It ispossible, therefore, that secreted GlyCAM1 has a different function in milk comparedwith its function on the epithelial cell membrane [28], [29]. The soluble form of MFGE8 hasbeen hypothesized to be involved in lubrication and protection of the intestinaltract and may have an antibacterial function in the intestinal tract [28].

The concentration of antibacterial proteins, mainly of LTF and LYZ, was shown to behigher in human milk [22]. Our analyses revealed, however, that bovine milkcontained a wider range of antibacterial proteins (Table 3). The difference in the number ofantibacterial proteins was caused by 5 cathelicidins and 3 mucins, which werepresent only in bovine milk (Table3). Cathelicidins are antimicrobial proteins found in different tissuesof many mammals. The cathelicidin gene (gene code CAMP) has been shown to beexpressed in the human mammary gland, and the polypeptide itself has been detectedin ducts of the human mammary gland [30], [31]; we did, however, not detectthe protein in our human milk sample. Cathelicidins have an N-terminal cathelin-likedomain, which is conserved between mammals, and a diverse C-terminal antimicrobialdomain (Figure 1). Thisantimicrobial domain differs in both length (12 to 80 residues) and structurebetween the different cathelicidins [32]. Most of the peptides weidentified (Figure 1) were fromthe cathelin-like domain. Although this domain of the protein is conserved in thedifferent cathelicidins, there are enough differences in the amino acid sequence todiscriminate between the cathelicidins. This cathelin-like domain is separated fromthe antimicrobial domain during the maturation, which is caused by neutrophilelastase [32].This elastase and cathelicidins are present in polymorphonuclear leukocytes, but indifferent granules [33], [34]. The mature forms of these antimicrobial peptides arefound at mucosal surfaces and within bodily secretions [35]. The bovine genome containsat least 10 cathelicidin copies, whereas the human genome contains only one [32], [36]. The expansionsin the cathelicidin gene family in the bovine genome has been hypothesized to berelated to increased exposure to bacteria at the epithelial surface of the bovinemammary gland [36].

10.1371/journal.pone.0019433.g001

Figure 1

Overview of the amino acid sequence of the 5 bovine cathelicidinsfound.

The red-colored amino acids designates the signaling peptide, thegreen-colored amino acids the cathelin-like domain and the black-coloredamino acids the antimicrobial peptide. Bold amino acids are identical in>50% of the sequences, normal capitals show amino acids occurringin multiple sequences ad lower case amino acids occur in only one sequence.The peptides which were identified are underlined, and the yellow markingshows the peptides used for quantification. For comparison, also the aminoacid sequence of the human cathelicidin is shown.

In addition to these antibacterial proteins, another antibacterial protein islactoperoxidase (LPO), which is present in higher concentrations in both serum andMFGM of bovine milk than of human milk (Table 3). The primary function of this protein isto catalyze oxidation of certain molecules, using hydrogen peroxide, to generatereactive products with a wide antimicrobial activity [36], [37]. LPO is excreted mainlyin milk and saliva [36]. The concentration of LPO in bovine milk has been shownto increase significantly in the first 5 days of lactation, reaching a plateau levelafter 2 weeks [37]. In milk and saliva, the main component known to beoxidized is the thiocyanate ion (SCN-) [36], [37]. The diet of the cowconsists mainly of plant materials and is a good source of SCN- [38]. This SCN- canbe converted by LPO into hypothiocyanate (SCNO-), which is a potent inhibitor ofbacterial growth [37], [38]. In human milk, however, the limiting factor for LPOactivity is its low SCN- concentration [39]. The higher concentration ofLPO in bovine milk, compared with human milk, may be related, therefore, todifferences in SCN- availability in the diet of the cow compared with the diet ofhuman.

In summary, results demonstrate our ability to detect a wide range of proteins,including those from the host defense system, in human and bovine milk. Qualitativeand quantitative differences were found in the milk of these two mammals. A numberof antimicrobial proteins (cathelicidins, lactoperoxidase) were more abundant inbovine milk. The high concentration of lactoperoxidase is probably linked to thehigh amount of thiocyanate in the plant-based diet of cows. Higher concentrations offour proteins involved in the mucosal defense system (IgA, CD14, LTF, and LYZ) werefound in human milk than in bovine milk. It is known that the newborn baby isdeficient for two of these proteins, i.e. IgA and CD14. The concentrations of thesefour proteins, which are relatively similar in human and bovine colostrum, arehigher in mature human milk compared to mature bovine milk. These differences inconcentration between species may be related to differences in the development ofthe immune system of babies and calves. These results may, therefore, indicate aslower maturation of the immune system in babies than in calves. This first detailedanalysis of host defense proteins in human and bovine milk is an important step inunderstanding the function of milk in these two mammals.

Materials and Methods

The different steps involved in our analysis are described in this section. Figure 2 gives an overview of theexperimental procedure. Milk samples were donated anonymously for this study andpooled before use, so IRB approval was not required. The regulations on which theexemption is based are 1. The “Law on medical-scientific research/Wetmedisch-wetenschappelijk onderzoek” and 2. the “Code Good Practice/CodeGoed Gebruik” of the “Dutch federation of Biomedical ScientificSocieties”.

10.1371/journal.pone.0019433.g002

Figure 2

Overview of the experimental procedure.

Pooled milk samples

Human milk was collected from 10 healthy mothers between 3 and 10 months inlactation. Samples of 10 mL were collected and frozen for later analysis. Milksamples were donated anonymously for this study and pooled before use, so IRBapproval was not required. After thawing, the 10 samples were pooled and proteinfractions were separated (see below). One bovine tank milk sample was collectedfrom the university farm “De Ossekampen” in Wageningen, TheNetherlands, which was milk from 30 clinically healthy cows which were between 3weeks and 10 months in lactation.

Separation of milk serum and MFGM protein fractions

The separation of the serum and MFGM proteins was done as described by [5]. Milksamples were centrifuged at 1500 g for 20 min at 4°C. The cream was used forMFGM protein isolation. 5 mL of the skimmed milk was centrifuged for 90 min at100,000 g to pellet the casein; the resulting supernatant, containing the serumproteins, was frozen at −45°C. The cream (about 10 mL) was washed 4times by careful shaking with 30 mL phosphate-buffered saline followed bycentrifugation. The washed cream was mixed 1∶1 (vol) with Milli-Q water,sonificated for 2 min, and centrifuged to remove fat. The watery subnatant,containing the MFGM proteins, was frozen at −45°C.

Protein quantification

The protein content of all samples was quantified using a BSA Protein Assay kit(Thermo, San Jose, CA, USA). The results from these analyses were used to loadthe same amount of protein per fraction on the SDS-PAGE gel or centrifugalfilter device.

SDS-PAGE

Pre-cast 12% Precise Protein Gels were used with HEPES buffer (Thermo, SanJose, CA, USA). The thawed protein samples were mixed 1∶1 (vol) with 2xsample buffer (125 mM Tris-HCl (pH 6.8), 4% SDS, 20% glycerol, and0.01% bromophenol blue in Milli-Q water; just before use, 5%β-mercaptoetanol was added) and heated for 5 min at 95°C. The proteinload on the gel was about 30 µg of protein per well. The gel was run for45 min at 130 Volt. The proteins were stained for 4 h using the Colloidal BlueStaining Kit (LC6025, Invitrogen, Carlsbad, CA, USA), and destained overnight inMilli-Q water.

Qualitative proteomics

Except when stated otherwise, all solutions were prepared in 50 mM NH4HCO3 (pH8). After each step, samples were sonicated for 1 min followed by spin downusing a centrifuge. For each sample put on the SDS-PAGE gel, the gel lane wascut in 8 slices of about equal size. Each slice was cut into <1 mm3 piecesand transferred to low-binding microcentrifuge tubes (0030 108.094, Eppendorf,Hamburg, Germany). The gel pieces were washed twice with water. The proteinswere reduced by incubation in 50 mM dithiotreitol for 1 h at 60°C followedby incubation in 100 mM iodoacetamide for 1 h at room temperature in the dark.After reduction, the gel pieces were washed 3 times with 50 mM NH4HCO3. Samplewere then frozen and thawed 3 times to increase the accessibility for trypsin.20 µL of freshly prepared trypsin solution (10 ng/µL) was added tothe gel pieces. Extra 50 mM NH4HCO3 was added to cover the gel piecescompletely. The gel pieces were incubated overnight at room temperature. Aftertrypsin digestion, the supernatant was transferred to a clean low-bindingmicrocentrifuge tube (Eppendorf). 10 µL 5% trifluoroacetic acid(TFA) in water was added to the gel pieces, and after sonication the acidicsupernatant was added to the same microcentrifuge tube. 10 µL 10%acetonitrile/1% TFA was then added to the gel pieces, and aftersonication the supernatant was added to the same microcentrifuge tube. The pH ofthe final peptide mixture was verified to be about 2, using pH paper.

Quantitative proteomics

The previously prepared milk serum and MFGM protein fractions were analyzed infivefold using an adapted version of [40]. 20 µL of proteinsolution, containing about 25 µg of protein, was solubilized in 180µL of Solution A (100 mM Tris/HCl (pH 7.6) containing 4% SDS and0.1 M DTT). Samples were heated for 5 min at 95°C. After cooling each sampleto room temperature, 10 µL was loaded on a filter-containing centrifugaldevice (10–20 kDa cutoff, OD003C34; Pall, Washington, NY, USA) andcentrifuged at 20,000 g for 1 min. 100 µL of Solution B (8 M Urea in 100mM Tris/HCl pH 8) was added and the device was centrifuged for 30 min at 20,000g. 100 µL of Solution C (0.05 M iodoacetamide in Solution B) was added.The device was mixed for 1 min, followed by incubation for 10 min. The devicewas then centrifuged for 30 minutes at 20,000 g. Three wash steps, with 110, 120and 130 µL respectively, of Solution B were performed with centrifugationfor 30 min at 20,000 g after each wash step. 140 µL of solution D (0.05 MNH₄HCO₃) was added followed by centrifugation at20,000 g for 30 min. The filter unit was then transferred to a low-bindingmicrocentrifuge tube (Eppendorf) and 1 µL sequencing-grade trypsin (Roche,Penzburg, Germany) in Solution D (total volume 100 µL) was added to thefilter. The filters were incubated overnight at room temperature. Filters werethen centrifuged for 30 min at 20,000 g. Finally, 3.5 µL 10% TFA inwater was added. The pH of the final peptide mixture was verified to be about 2,using pH paper.

LC-MS/MS

Samples were analyzed by injecting 18 µL of sample over a 0.1032 mmProntosil 300-3-C18H (Bischoff, Germany) pre-concentration column (prepared inhouse) at a maximum pressure of 270 bar. Peptides were eluted from thepre-concentration column onto a 0.10200 mm Prontosil 300-3-C18H analyticalcolumn with an acetonitrile gradient at a flow of 0.5 µL/min. The gradientconsisted of an increase from 9% to 34% acetonitrile in water with1 mL/L formic acid in 50 min, followed by an increase in the percentageacetonitrile to 80% (with 20% water and 1 mL/L formic acid in theacetonitrile and the water) in 3 min, as a column-cleaning step. Between thepre-concentration and analytical columns, an electrospray potential of 3.5 kVwas applied directly to the eluent via a solid 0.5 mm platina electrode fittedinto a P777 Upchurch microcross. Full scan positive mode FTMS spectra weremeasured between m/z 380 and 1400 on a LTQ-Orbitrap (Thermo electron, San Jose,CA, USA). MSMS scans of the four most abundant doubly- and triply-charged peaksin the FTMS scan were recorded in data-dependent mode in the linear trap (MSMSthreshold = 5.000).

Peptide and protein identification

Each run with all MSMS spectra obtained was analyzed with Bioworks 3.3.1 (Thermoelectron, San Jose, CA, USA). A maximum of totally 1 differential modificationper peptide was set for oxidation of methionines and de-amidation of N and Q.Carboxamidomethylation of cysteines was set as a fixed modification(enzyme = trypsin, maximally 2 missed cleavages, peptidetolerance 10 ppm, fragment ions tolerance 0.5 amu).

A combined protein database was constructed from the human and bovine IPIdatabases (downloaded as fasta files from ftp://ftp.ebi.ac.uk/pub/databases/IPI/current/ accessed August 2009). Aset of 31 protein sequences of common contaminants was added including Trypsin(P00760, bovin), Trypsin (P00761, porcin), Keratin K22E (P35908, human), KeratinK1C9 (P35527, human), Keratin K2C1 (P04264, human), and Keratin K1CI (P35527,human). A decoy database was created by adding the reversed sequences usingSequenceReverser from the MaxQuant package [41]. These steps gave a databasecontaining 242906 proteins in total.

The peptide identifications obtained were filtered in Bioworks with four filtercriteria: ΔCn >0.08, Xcorr >1.5 for charge state 2+, Xcorr>3.3 for charge state 3+, and Xcorr >3.5 for charge state 4+[42].Finally, proteins were displayed based on minimally 2 distinct peptides, an Sfscore >1, and a probability <0.5. The false discovery rate (the number ofhits against the inverted decoy proteins within filter settings divided by thetotal number of proteins within filter settings times 100%) was0%. The function of the identified proteins was checked in the UniProtKBdatabase (http://www.uniprot.org/accessed November 2009).

Protein quantification

Peak height of peptides belonging to an identified protein was determined usingBioworks. For the host defense proteins, the 3 most abundant peptides perprotein were summed [43]. The same 3 peptides were chosen for the fivereplicates. The summed peptide heights were compared between the human andbovine samples using an independent two-sample t-test, using PASW statistics 17(SPSS Inc, Chicago, IL, USA). If a protein was not detected in a specificsample, the value for the peak height was set to 10⁴ (minimumdetection level) for statistical calculations and “<1” in Table 3.

Footnotes

Competing Interests: The authors have read the journal's policy and have the following conflictto declare: TvH is an employee of FrieslandCampina, a dairy company thatdevelops and markets dairy products. This did not alter the authors'adherence to all the PLoS ONE policies on sharing data and materials.

Funding: Funding support was provided by the Dutch Dairy Association. The funders had norole in study design, data collection and analysis, decision to publish, orpreparation of the manuscript.

Acknowledgments

We would like to thank the mothers who donated milk for this research as well as thehospital “Gelderse Vallei Ede” for their help in obtaining the humanmilk samples. We would also like to thank Michael Grossman for his editorialcomments on the manuscript.

References

1. GermanJBDillardCJWardRE2002Bioactive components in milk.Curr Opin Clin Nutr Metab Care5653658[PubMed][Google Scholar]
2. SeverinSWenshuiX2005Milk biologically active components as nutraceuticals:review.Crit Rev Food Sci Nutr45645656[PubMed][Google Scholar]
3. NewburgDS2001Bioactive components of human milk: evolution, efficiency, andprotection.Adv Exp Med Biol501310[PubMed][Google Scholar]
4. D'AllesandroAScaloniAZollaL2010Human milk proteins: and interactomics and updated functionaloverview.J proteome Res933393373[PubMed][Google Scholar]
5. SmolenskiGHainesSKwanFYBondJFarrV2007Characterisation of host defence proteins in milk using aproteomic approach.J Proteome Res6207215[PubMed][Google Scholar]
6. AffolterMGrassLVanrobaeysFCasadoBKussmannM2010Qualitative and quantitative profiling of the bovine milk fatglobule membrane proteome.J Proteomics7310791088[PubMed][Google Scholar]
7. HeidHWKeenanTW2005Intracellular origin and secretion of milk fatglobules.Eur J Cell Biol84245258[PubMed][Google Scholar]
8. CharlwoodJHanrahanSTyldesleyRLangridgeJDwekM2002Use of Proteomic Methodology for the Characterization of HumanMilk Fat Globular Membrane Proteins.Anal Biochem301314324[PubMed][Google Scholar]
9. FortunatoDGiuffridaMGCavalettoMGaroffoLPDellavalleG2003Structural proteome of human colostral fat globule membraneproteins.Proteomics3897905[PubMed][Google Scholar]
10. ReinhardtTALippolisJD2006Bovine milk fat globule membrane proteome.J Dairy Res73406416[PubMed][Google Scholar]
11. D'AmatoABachiAFasoliEBoschettiEPeltreG2009In-depth exploration of cow's whey proteome viacombinatorial peptide ligand libraries.J Proteome Res839253936[PubMed][Google Scholar]
12. ReinhardtTALippolisJD2008Developmental changes in the milk fat globule membrane proteomeduring the transition from colostrum to milk.J Dairy Sci9123072318[PubMed][Google Scholar]
13. KorhonenHMarnilaPGillHS2000Milk immunoglobulins and complement factors.Br J Nutr84S75S80[PubMed][Google Scholar]
14. PorterP1972Immunoglobulins in Bovine Mammary Secretions - Quantiativechanges in early lactation and absorption by the neonatalcalf.Immunology23225238[PubMed][Google Scholar]
15. ShahNP2000Effects of milk-derived bioactives: an overview.Br J Nutr84S3S10[PubMed][Google Scholar]
16. WalkerA2010Breast milk as the gold standard for protectivenutrients.J Pediatr156S3S7[PubMed][Google Scholar]
17. AllenWDPorterP1975Localization of immunoglobulins in intestinal mucosa and theproduction of secretory antibodies in response to intraluminaladministration of bacterial antigens in the preruminantcalf.Clin Exp Immunol21407418[PubMed][Google Scholar]
18. KaetzelCS2005The polymeric immunoglobulin receptor: bridging innate andadaptive immune responses at mucosal surfaces.Immunol Rev2068399[PubMed][Google Scholar]
19. LabetaMOVidalKMorsJE2002Innate recognition of bacteria in human milk is mediated by amilk-derived highly expressed pattern recognition receptor, solubleCD14.J Exp Med19118071812[Google Scholar]
20. KovarMGSerdulaMKMarksJSFraserDW1984Review of the epidemiological evidence for an association betweeninfant feeding and infant health.Pediatric74615638[Google Scholar]
21. SandersonIRWalkerWA1999Development of the gastrointestinal tract.Chapter 14. Blackwell ScienceOxford, UK[Google Scholar]
22. FoxPFMcSweeneyPLH2003Advanced dairy chemistry. Volume 1: proteins, 3^rdedition. Chapter 12.4.Kluwer academic plenum publishersNew York, NY, USA[Google Scholar]
23. ChoJKAzumaNLeeCHYuJHKannoC2000Purification of membrane-bound lactoferrin from the human milkfat globule membrane.Biosci Biotechnol Biochem64633635[PubMed][Google Scholar]
24. WalstraPWoutersJTMGeurtsTJ2005Dairy Science and Technology. 2^nd edition. Chapters2.7 and 16.6.CRC PressBoca Raton, FL, USA[Google Scholar]
25. BuHFZuoXLWangXEnsslinMAKotiV2007Milk fat globule–EGF factor 8/lactadherin plays a crucialrole in maintenance and repair of murine intestinalepithelium.J Clin Invest11736733683[PubMed][Google Scholar]
26. HanayamaRTanakaMMiwaKShinoharaAIwamatsuA2002Identification of a factor that links apoptotic cells tophagocytes.Nature417182187[PubMed][Google Scholar]
27. NakataniHAokiNNakagawaYJin-NoSAoyamaK2006Weaning-induced expression of a milk-fat globule protein, MFG-E8,in mouse mammary glands, as demonstrated by the analyses of its mRNA,protein and phosphatidylserine-binding activity.Biochem J3952130[PubMed][Google Scholar]
28. DowbenkoDKikutaAFennieCGillettNLaskyLA1993Glycosylation-dependent cell adhesion molecule 1 (GlyCAM 1) mucinis expressed by lactating mammary gland epithelial cells and is present inmilk.J Clin Invest92952960[PubMed][Google Scholar]
29. HouZBaileyJPVomachkaAJMatsudaMLockefeerJA2000Glycosylation-Dependent Cell Adhesion Molecule 1 (GlyCAM 1) IsInduced by Prolactin and Suppressed by Progesterone in MammaryEpithelium.Endocrinol14142784283[Google Scholar]
30. ArmogidaSAYannarasNMMeltonALSrivastavaMD2004Identification and quantification of innate immune systemmediators in human breast milk.Allergy Asthma Proc25297304[PubMed][Google Scholar]
31. MurakamiMDorschnerRASternLJLinKHGalloRL2005Expression and secretion of cathelicidin antimicrobial peptidesin murine mammary glands and human milk.Pediatr Res571015[PubMed][Google Scholar]
32. OppenheimJJYangD2005Alarmins: chemotactic activators of immuneresponses.Curr Opin Immunol17359365[PubMed][Google Scholar]
33. ZanettiMGennaroRRomeoD1995Cathelicidins: a novel protein family with a common proregion anda variable C-terminal antimicrobial domain.FEBS Lett37415[PubMed][Google Scholar]
34. GillenwatersENSeaburyCMElliottJSWomackJE2009Sequence analysis and polymorphism discovery in 4 members of thebovine cathelicidin gene family.J Hered100241245[PubMed][Google Scholar]
35. MookherjeeNWilsonHLDoriaSPopowychYFalsafiR2006Bovine and human cathelicidin cationic host defense peptidessimilarly suppress transcriptional responses to bacteriallipopolysaccharide.J Leukoc Biol8015631574[PubMed][Google Scholar]
36. LemayDGLynnDJMartinWFNevilleMCCaseyTM2009The bovine lactation genome: insights into the evolution ofmammalian milk.Genome Biol10R43[PubMed][Google Scholar]
37. KussendragerKDvan HooijdonkACM2000Lactoperoxidase: physico-chemical properties, occurrence,mechanism of action and applications.Br J Nutr84S1925[PubMed][Google Scholar]
38. FontehFAGrandisonASLewisMJ2002Variations of lactoperoxidase activity and thiocyanate content incows' and goats' milk throughout lactation.J Dairy Res69401409[PubMed][Google Scholar]
39. RussellMWBobekLABrockJHHajishengallisGTenovuoJ2005Mucosal Immunology. Third Edition.MesteckyJLammMEMcGheeJRBienenstockJMayerLStroberWChapter 5: Innate humoral defense factorsLondon, UKAcademic Press[Google Scholar]
40. WiśniewskiJRZougmanANagarajNMannM2009Universal sample preparation method for proteomeanalysis.Nat Methods6359362[PubMed][Google Scholar]
41. CoxJMannM2008MaxQuant enables high peptide identification rates,individualized p.p.b.-range mass accuracies and proteome-wide proteinquantification.Nat Biotechnol2613671372[PubMed][Google Scholar]
42. PengJMEliasJEThoreenCCLickliderLJGygiSP2003Evaluation of multidimensional chromatography coupled with tandemmass spectrometry (LC/LC-MS/MS) for large-scale protein analysis: The yeastproteome.J Proteome Res24350[PubMed][Google Scholar]
43. SilvaJCGorensteinMVLiGZVissersJPCGeromanosSJ2006Absolute quantification of proteins by LCMSE - A virtue ofparallel MS acquisition.Mol Cell Proteomics5144156[PubMed][Google Scholar]