Microarray analysis of human milk cells: persistent high expression of osteopontin during the lactation period

Abstract

To continue the search for immunological roles of breast milk, cDNA microarray analysis on cytokines and growth factors was performed for human milk cells. Among the 240 cytokine-related genes, osteopontin (OPN) gene ranked top of the expression. Real-time PCR revealed that the OPN mRNA levels in colostrum cells were approximately 100 times higher than those in PHA-stimulated peripheral blood mononuclear cells (PBMNCs), and 10 000 times higher than those in PB CD14⁺ cells. The median levels of OPN mRNA in early milk or mature milk cells were more than three times higher than those in colostrum cells. Western blot analysis of human milk showed appreciable expression of full-length and short form proteins of OPN. The concentrations of full-length OPN in early milk or mature milk whey continued to be higher than those in colostrum whey and plasma as assessed by ELISA. The early milk (3–7 days postpartum) contained the highest concentrations of OPN protein, while the late mature milk cells (1 years postpartum) had the highest expression of OPN mRNA of all the lactating periods. The results of immunohistochemical and immunocytochemical staining indicated that OPN-producing epithelial cells and macrophages are found in actively lactating mammary glands. These results suggest that the persistently and extraordinarily high expression of OPN in human milk cells plays a potential role in the immunological development of breast-fed infants.

INTRODUCTION

Human milk is specifically adapted to the needs of infants. The major milk proteins (secretory IgA, lactoferrin, α-lactalbumin and casein) serve as multifunctional agents [1]. Human milk also contains a variety of cytokines and growth factors including IL-1β [2], IL-6 [3], IL-8 [4], IL-12 [5], IL-18 [6], tumour necrosis factor (TNF)-α [7], interferon (IFN)-γ [8], monocyte chemotactic protein 1 (MCP-1) [9], macrophage colony-stimulating factor (M-CSF) [10], granulocyte colony-stimulating factor (G-CSF) [11] and vascular endothelial growth factor (VEGF) [12]. Various immunological substances are identified in human milk, implicating a role for protective immunity against enteric and respiratory pathogens [13,14]. To determine the quantitative profile of cellular cytokines in human milk, the cytokine gene expression of human milk cells was comprehensively analysed by using cDNA microarray. The osteopontin (OPN) mRNA expression showed the extremely high level among the genes examined in milk cells in comparison with phytohemagglutinin (PHA)-stimulated peripheral blood mononuclear cells (PBMNCs) of healthy adults. OPN, one of the multifunctional proteins secreted by macrophages, T cells and epithelial cells, induces cell-mediated immune responses [15], chemotaxis of inflammatory cells [16,17], and anti-inflammatory responses exerted by inhibiting nitric oxide (NO) production [18]. Although OPN production in human milk was reported [19,20], the precise expression levels and concentrations in human milk have not been clarified. Therefore, we focused on the kinetics and source of OPN in human milk during the lactation period to refer to the potential roles of OPN in the development of host defense and mucosal immunity throughout the infancy. The biological role of OPN in human milk, especially in late lactating stage, is discussed.

MATERIALS AND METHODS

Milk sample collection

Colostrum (within the first 72 h postpartum), early milk (from 72 h to 7 days postpartum), and mature milk (more than 28 days after delivery) were collected from healthy mothers in Fukuoka City Hospital and Maternity Centre (Hirata Breastfeeding Consultant), after informed consent was obtained. All mothers delivered full-term healthy neonates and had no mastitis or other inflammations. Mature milk samples were further classified into three groups, mature milk (a) (approximately 1 month postpartum) (b) (approximately 4–7 months postpartum) and (c) (approximately 11–14 months postpartum). All sample specimens were collected into sterile polypropylene tubes by manual breast pumps, and were immediately centrifuged at 1000 × g for 5 min. After removal of lipids, the whey was stored at –80°C. The pellets of cellular components were washed with PBS by repeated centrifugation, and were then stored at –80°C until use.

Preparation of mononuclear cells (MNCs) and CD14⁺ monocytes from human peripheral blood

Peripheral blood (PB) samples were obtained from healthy nonlactating adults or lactating donors. PBMNCs were obtained by density gradient centrifugation using Lymphocyte Separation Medium (1·077 g/ml) (Cappel, Aurora, OH, USA) as previously described [21]. Some PBMNCs from nonlactating donors were purified into CD14⁺ subpopulation by using anti-CD14 MicroBeads and Mini Macs with positive selection column (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacture's instruction. Briefly, PBMNCs were incubated with saturating concentrations of the MicroBeads for 15 min at 6°C and washed with PBS containing 2 m m EDTA and 0·5% bovine serum albumin (BSA). Labelled and positively enriched cells were eluted after removal of the columns from the magnetic device. The purity of CD14⁺ monocytes was confirmed to be >98·4% by flow cytometry using EPICS Altra XL (Beckman Coulter, Hialeah, FL, USA). PBMNCs (2 × 10⁶ cells/ml) from five nonlactating donors were independently stimulated with 10 µg/ml of PHA for 48 h.

cDNA microarray analysis

Total RNA was extracted from the pellets of human milk whole cells (collected within 7 days postpartum) and PHA-stimulated PBMNCs by RNA extraction kit, Isogen (Nippon Gene, Osaka, Japan), according to the manufacture's instruction. High-purity of RNA was confirmed by agarose gel electrophoresis and absorbance (A) ratio (A₂₆₀/A₂₈₀). cDNA microarray chip (Human Cytokine CHIP version 2·0, TaKaRa, Otsu, Japan) has 240 cDNA fragments of human cytokine-related genes. To make cDNA probes, approximately 10–15 µg of total RNA was labelled with Cy-3 or Cy-5 by reverse transcription using Labeling core kit (TaKaRa) with oligo-dT primer. Hybridization was performed at 65°C for 12–16 h. Microarray was washed with the buffer containing 2xSSC, 0·2% SDS and rinsed 0·05xSSC as previously described [22]. Scanning was performed with FLA-8000 (Fuji film, Tokyo, Japan). Image files were quantified by ArrayVision Software (Amersham Biosciences, Piscataway, NJ). All data were analysed by Microsoft Excel (Microsoft Corporation, Redmond, WA, USA). The β-actin was used as the internal control for normalization of each data. The microarray analysis on pooled cDNA of milk cells from seven donors was carried out in comparison with pooled cDNA of PHA-stimulated PBMNCs from five independent donors. The experiments were repeated three times independently to validate the reproducibility.

Quantitative PCR for OPN gene expression by TaqMan method

cDNA was synthesized from the total RNA samples by using a First-Strand cDNA Synthesis Kit (Amersham Pharmasia Biotech, Tokyo, Japan) with random hexamers. Nucleotide sequences of PCR primers and TaqMan probe for the quantitative PCR were as follows; forward primer: 5′-GCAGACCTGACATCCAGTAC CCT-3′, reverse primer: 5′-GCCTTGTATGCACCAT TCAACTC-3′, and TaqMan probe: 5′-AGACGAGCACAT CACCTCACACATGGA-3′. TaqMan probe was labelled at the 5′ end with the reporter dye molecule, FAM (6-carboxyfluorescein; emission I, 538 nm), and at the 3′ end with the quencher fluor TAMRA (6-carboxytetramethyl rhodamine; emission I, 582 nm) via a linker arm nucleotide (LAN). Standard curve for quantitative analysis of target gene was obtained by cDNA generated from milk cells. PCR primers and target probe for β-actin were purchased from ABI (Applied Biosystems, Foster City, CA) as a kit of TaqMan β-actin Control Reagent. TaqMan assay was performed as described previously [23]. Gene dosages of OPN and β-actin mRNA were analysed by ABI PRISM 7700 Sequence Detector (ABI). To calculate the relative amount of OPN gene in cells, each value of OPN gene expressions was divided by that of the internal control, β-actin, and then the final expression level was described as a ratio to that of milk cells. All analyses were carried out in duplicate samples.

Western blot analysis

SDS-PAGE and immunoblotting to detect the human milk OPN protein were performed as previously described [24]. Briefly, one microgram of whey protein was subjected to 10% SDS-PAGE and was transferred to polyvinylene difluoride (PVDF) membrane. The membrane was then blocked for 30 min in blocking buffer (5% skim milk in Tris-buffered saline) and probed with 1 µg/ml rabbit anti-human OPN polyclonal antibody (clone O-17) (IBL: Immuno-Biological Laboratories, Gunma, Japan) which reacts with N-terminal region of OPN protein, or 1 µg/ml mouse anti-human OPN monoclonal antibody (clone 10A16) (IBL) which reacts with C-terminal region of the protein. The possible recognition sites of these two antibodies were shown in Fig. 2c. The membranes were rinsed with Tris-buffered saline. The protein content was visualized using horseradish peroxidase-conjugated secondary antibodies (anti-rabbit IgG to O-17, or anti-mouse IgG to 10A16) followed by enhanced chemiluminescence (Amersham Pharmasia Biotech). The purified human milk OPN (Hokudo, Hokkaido, Japan) and thrombin-cleaved OPN performed as previously described [19] were used as the positive controls.

Fig. 2.

Western blot analysis for OPN in the whey of human milk. (a, b) According to the Materials and Methods, SDS-PAGE and immunoblotting were performed using the rabbit anti-human OPN polyclonal antibody (a: O-17) or the mouse monoclonal antibody (b: 10A16). Full-length OPN is detected as the 75 kD band, and thrombin-cleaved OPN is detected as the 35 kD band. (c) Structure of OPN protein and possible recognition sites of antibodies (O-17, 10A16) used. Gray shaded area indicates the RGD domain. A closed triangle indicates the position of thrombin-cleavage site. lane a: purified human milk OPN, lane b: thrombin-cleaved OPN, lane c: colostrum (48 h postpartum), lane d: early milk (5 days), lane e: transitional milk (17 days), lane f: mature milk (1 month), lane g: mature milk (2 months), lane h: mature milk (5 months), lane i: mature milk (1 years)

Enzyme-linked immunosorbent assay (ELISA)

The concentrations of OPN protein in human milk whey and plasma were measured by the human OPN ELISA kit with the coated antibody (O-17) and the secondary antibody (10A16) (IBL), which detects the full-length form, but not the short forms of the protein. All samples were diluted from 10 to 100 000-fold to adjust within the detectable range (5–320 ng/ml). The measurements were repeated with different dilutions to confirm the validity of analyses. To verify the specificity of this system, recovery tests were performed by adding specific amounts of recombinant human OPN to whey or plasma.

Immunohistochemical and immunocytochemical stainings

Lactating mammary gland tissue specimens were donated from a patient with benign breast tumour, after informed consent was obtained. The tissue specimens were fixed in 10% buffered formalin and were embedded in paraffin. After deparaffinization, each specimen was incubated for 5 min with 3% hydrogen peroxide, followed by a 5-min incubation with anti-OPN polyclonal antibody (O-17), monoclonal antibody (10A16) or control antibody for 10 min. The specimens were incubated with biotinylated second antibodies for 10 min, followed by treatment with peroxidase-labelled streptavidin for 10 min. Staining of tissue specimens was completed after a 10-min incubation with 3% 3-amino-9-ethylcarbazole. Cellular components in human milk were separated and were subjected to Cytospin 3 (Shandon, Frankfurt, Germany), followed by the immunocytochemical staining with the polyclonal antibody O-17. Staining of milk cells was performed by the same protocol as for immunohistochemistry.

Statistical analysis

Statistical significance of the differences in the expression levels of mRNA or protein between two groups was determined by Mann–Whitney U-test. The correlation between two groups was analysed applying Spearman's correlation coefficient (CC) by rank test. P < 0·05 was considered to indicate statistical significance. Microsoft Excel was used for computation.

Ethics

This study was approved by the Regional Ethics of Committee for Human Research at the Faculty of Medicine of Kyushu University.

RESULTS

Highly expressed genes in human milk cells by cDNA microarray analysis

To first determine the profile of highly expressed genes of cytokines and growth factors, the gene expression levels were compared by cDNA microarray between human milk cells and PHA-stimulated PBMNCs, which were expected to express many cytokine genes at increased levels. As shown in Table 1, 16 genes were expressed in milk cells (collected within 7 days postpartum) more than two times higher than in PHA-stimulated PBMNCs. OPN [19,20], MCP-1 [9], macrophage inflammatory protein 1-alpha (MIP-1α) [25], and IL-8 [4,26] were identified as the highly expressed genes in milk cells compared to PHA-stimulated PBMNCs of healthy adults. OPN showed the highest expression of all 240 genes studied. Reproducibility was confirmed by repeated experiments and dye swap according to the MIAME (minimum information about microarray experiment) document.

Table 1.

Cytokine and growth factor genes highly expressed in human milk cells

Ratio	Gene
621·65	Osteopontin (secreted phosphoprotein 1)
58·95	Small inducible cytokine subfamily A (Cys-Cys), member 18
38·03	Vascular endothelial growth factor
25·00	Small inducible cytokine subfamily A (Cys-Cys), member 20
20·38	Growth regulated oncogene 2
17·21	Small inducible cytokine A3-like 1
16·87	Small inducible cytokine A2 (monocyte chemotactic protein 1)
15·66	Small inducible cytokine A3 (homologous to mouse Mip-1α)
12·88	Tumour necrosis factor (ligand) superfamily, member 13
10·37	Small inducible cytokine A4 (homologous to mouse Mip-1β)
8·11	Colony stimulating factor 1 (macrophage)
7·34	Interleukin 8
4·70	Growth regulated oncogene 1
4·67	Interferon regulatory factor 1
3·36	Tumour necrosis factor (TNF superfamily, member 2)
2·96	Interleukin 1, β

The microarray analyses were repeated three times independently. Expression levels of each cytokine or growth factor genes in milk cells are shown as a ratio to that in PHA-stimulated PBMNCs.

Quantification of OPN mRNA in human milk and peripheral blood cells

The amounts of OPN mRNA in human milk cells and PBMNCs were determined by real-time PCR to confirm the results of cDNA microarray analysis (Figs 1a–c). The expression levels of OPN in colostrum cells (median 0·04) were higher than those in PHA-stimulated PBMNCs (median 4·6 × 10⁻⁴) (P = 0·008), unstimulated PBMNCs (median 1·6 × 10⁻⁷) (P < 0·001) or PB CD14⁺ cells (median 3·4 × 10⁻⁶) (P = 0·002) (Fig. 1a). The OPN mRNA levels in human milk cells were compared between the lactating periods (Fig. 1b). The OPN mRNA levels in early milk cells (median 0·27) and mature milk cells (median, a: 0·31, b: 0·17, c: 0·76) were all higher than those in colostrum (median 0·04) (P < 0·001). There was no difference in the expression levels between early milk cells and mature milk cells (a and b). On the other hand, the late mature milk cells around the weaning of breastfeeding (1 years postpartum) (c) transcribed the highest OPN genes of all lactating periods. The OPN mRNA levels in mature milk cells (a) were much higher than those in PBMNCs of the same lactating mother donors (median 2·3 × 10⁻⁶) (P < 0·001) (Fig. 1c).

Fig. 1.

OPN mRNA levels in human milk cells and peripheral blood cells. (a) The levels in colostrum cells, PHA-stimulated or unstimulated PBMNCs, and PB CD14⁺ cells. (b) The levels in human milk cells during the lactation period. (c) The levels in mature milk cells and PBMNCs sampled from the same donors. Each bar represents the median value.

OPN protein in the whey of human milk

OPN protein in the whey was examined by Western blotting with two antibodies. As shown in Fig. 2a, the full-length forms of OPN (75 kD) were detected by polyclonal antibody (O-17) in early milk (lane d) and mature milk (lanes e∼h), but only faintly in colostrum (lane c). By using monoclonal antibody (10A16), thrombin-cleaved forms (35 kD) were detected in concert with the expression of full-length forms (75 kD) in early milk (lane d) and mature milk (lanes e–h), but not in colostrum (Fig. 2b).

Concentrations of OPN protein determined by ELISA

The OPN concentrations in the whey and plasma were then measured by ELISA system, which detects the full-length form of human OPN. The results of recovery test: the ratios of the measured and calculated concentrations were 100·1 ± 7·9% (mean ± SD) in colostrum, 96·7 ± 7·1% in early milk, 98·8 ± 3·7% in mature milk (a), 104·5 ± 3·7% in mature milk (b), and 103·0 ± 7·4% in mature milk (c). This assay system demonstrated that the milk components have no significant effects on the quantification of OPN. The production level of OPN in early milk (median 1493·4 µg/ml) or mature milk (median, a: 896·3, b: 550·8, c: 412·7 µg/ml) was significantly higher than that in colostrum (median 2·7 µg/ml) (P < 0·001) (Fig. 3a). The early milk contained the highest concentrations of OPN of all lactating periods. Mature milk (a) contained much higher levels of OPN protein than the maternal plasma (median 339·0 ng/ml) collected at the same sampling time of breast milk (P < 0·001) (Fig. 3b). There was a positive correlation between the OPN concentrations in the milk whey and those in plasma for individual mothers (CC = 0·627, p = 0·047) (data not shown). There was no significant correlation between OPN mRNA levels in milk cells and OPN protein concentrations in the milk whey (data not shown).

Fig. 3.

OPN concentrations in the whey of human milk and plasma measured by ELISA (a) The concentrations in the whey of human milk during the lactation period. (b) The concentrations in the whey of human milk and plasma sampled from the same mother donors. Each bar represents the median value.

Identification of OPN-producing cells

To further address the source of OPN in breast milk, immunohistochemical analysis was carried out using the tissue of lactating mammary glands. OPN positive cells were found in the epithelial cells of ductal tissue with active secretion (O-17) (Fig. 4a, arrow), but not in those of nonductal tissue with inactive secretion (Fig. 4a, arrow head). The immunocytochemical staining of milk cells further revealed that OPN-producing cells in human milk are macrophages (Fig. 4b, arrow).

Fig. 4.

Immunohistochemical and immunocytochemical staining of OPN in mammary glands and human milk cells (a) OPN was detected in the epithelial cells of lactating mammary glands with active secretion (arrow) but not in those with inactive secretion (arrowhead). (b) OPN was positive for macrophages (b, arrow) in human milk.

DISCUSSION

The present study demonstrated the extremely high expressions of OPN mRNA and protein in human milk. Notably, the transcription level was at the top among the cytokine and growth factor genes analysed by the cDNA microarray. Both mRNA and protein levels increased during the secreting phase of early milk, and were maintained in mature milk throughout the lactation. The kinetics of expression and production of OPN was different from that of other cytokines, the concentrations of which are consistently elevated in colostrum [3,5,6,9,26]. Furthermore, the OPN mRNA levels in late mature milk cells (1 years postpartum) were the highest of all lactating periods, in contrast to the highest protein levels in the whey of early milk (3–7 days postpartum). The major sources of OPN were mammary gland epithelial cells and milk macrophages/monocytes. These results suggest that OPN continues to be vigorously secreted in human milk, and may play a potential role in the immunological development of breast-fed infants.

Human milk during the early secreting phase contains a large amount of protein (15·8 g/l), and the concentrations gradually decrease to 8·0–9·0 g/l with the establishment of lactation [27]. The OPN concentration in mature human milk accounted for approximately 5–10% of the total milk proteins, which nearly corresponded to the lactoferrin concentration. The quantitative significance is emphasized by the fact that OPN was the highest expressing gene of all 240 cytokine-related genes studied by the microarray analysis (Table 1).

Colostrum contains approximately 5 × 10⁶/ml of leucocytes, which decrease to 10⁵/ml in mature milk. However, the number of cells supplied to the infants might be almost constant throughout the lactating period because of the increase in feeding volume. As for the cell types, colostrum includes 40–50% of macrophages, 40–50% of neutrophils, and 5–10% of lymphocytes. On the other hand, human mature milk consists of 85% macrophages and 10% lymphocytes [28]. The milk samples in this study showed similar constituents of the cell types (data not shown). The increasing fraction of macrophages in cellular components might affect the sustained high expression levels of OPN in mature milk cells. In addition, estradiol (E2) inhibits the endogenous OPN gene expression [29]; the postpartum reduction of maternal oestrogen levels could be associated with the raised expression of OPN in mammary glands. These hormonal changes will affect not only local but also systemic OPN levels; this is consistent with the positive correlation of OPN concentrations between milk and plasma for individual mothers. OPN proteins were exclusively detected in the whey, but negligible in the plasma of lactating mothers. There may be a local effect of OPN on the self-differentiation of mammary glands because of a lack of mammary alveolar structures in pregnant mice treated with the anti-sense OPN [30]. The high concentration of OPN in early milk may account for the proliferation and differentiation of mammary glands. On the contrary, OPN mRNA levels increased rather in late mature milk cells. Considering that milk cell counts are reduced in late lactating period, and that there is no correlation of the milk OPN levels between mRNA and protein, the main source of milk OPN should be mammary glands rather than milk macrophages.

OPN has been regarded as a key molecule inducing the T helper type 1 (Th1)-type immunity. The OPN-null mice have a defective Th1 response, and are more susceptible to infection by Herpes simplex or Listeria monocytogenes than the wild-type mice [15]. Mycobacterium bovis bacillus Calmette-Guérin (BCG) grew more rapidly in macrophages derived from OPN-null mice than in those from wild-type mice [31]. Nau et al. [32] found an inverse correlation between the tissue OPN expression levels and disease progression after inoculation of Mycobacterium bovis BCG in humans, and corroborated the findings in the cases with the disseminated infection of Mycobacterium tuberculosis. Pabst et al. [33] reported that breast-fed infants showed an increased proliferative response of T cells compared with bottle-fed infants when both were vaccinated with BCG at birth, and suggested the effects of breastfeeding on the Th1 cell stimulation [34]. In this line, the present results indicated a potential contribution of milk OPN to the host resistance of infants via the augmentation of Th1 response. The OPN-deficient mice were considered to be susceptible to the infection by rotavirus [35]. Furthermore, there were many studies that demonstrated the efficacy of breastfeeding in the prevention of rotavirus infections [36,37]. It is possible that human milk OPN may be useful in preventing rotavirus infections during lactating period. OPN can bind with high affinity to several different integrins via the RGD (Arg-Gly-Asp) integrin-binding motif. The RGD motif located in the centre of the OPN molecule (Fig. 2c) may competitively inhibit the attachment of the adenoviral RGD motif to the αv integrin expressed on the targeted cells [38]. As shown in Fig. 2, both fragmented and full-length OPN proteins continued to be produced in the mature milk. N-terminal fragment of OPN showed a broaden smear (Fig. 2a), suggesting post-translation modifications including phosphorylation. A proteolytic fragment from N-terminal region of OPN, which contains the integrin binding site, is sufficient to induce macrophage IL-12 expression [15]. Taken together, both full-length and fragmented OPN in early and mature milk may augment the host-resistance to various pathogens via the mechanisms of the Th1 polarization.

The prolonged secretion of OPN may raise other biological roles in the infant life. The Th1 shift in the gastrointestinal milieu promoted by milk OPN may prevent the development of allergic diseases although it is difficult to show the clear evidence that breastfeeding exerts an antiallergic effect because of the environmental interference [39,40]. OPN is one of the noncollagenous proteins of bone matrix [41], and is involved in bone resorption [42]. The bone resorption is mediated by parathyroid hormone in OPN-dependent manner [43]. Human milk OPN might contribute to the bone formation in infants as suggested previously [20]. OPN is also detected in bovine milk. Bayless et al. [44] reported that bovine milk OPN yielded ∼8 mg from 1 litre of raw milk, which appeared to be less than 50fold as small as human milk OPN. Further analysis of milk OPN may shed some lights on the biological significance of the multifunctional protein in the mammalian physiology.