Physiological secretion of chemokines in human breast milk

C.A. Michie; E. Tantscher; T. Schall; A. Rot

At birth, the neonatal mammal is exposed to a wide range of infectious agents and foreign antigens. Mammals are first fed with maternal secretions which have been demonstrated to deliver specific immune protection in the form of antibodies and a range of non-specific defense mechanisms such as iron-binding proteins. Mammalian milk also contains large numbers of viable leukocytes, including neutrophils, monocytes and lymphocytes. Mother's milk-derived phagocytes, upon ingestion, can contribute to the mucosal defenses of the newborn by virtue of their antimicrobial functions [1, 2]. Milk-derived lymphocytes, as indicated by animal models, can migrate from the gut into the circulation and gut-associated lymphoid tissues of the newborn, delivering more specific immune functions [3, 4].

In order to appear in milk, leukocytes have to leave the maternal circulation and traverse several membranes and cell barriers. Chemotactic cytokines have been implicated in the recruitment of inflammatory and immune cells to sites of inflammation across these barriers, as in appendicitis or a rejecting renal graft [5, 6]. Leukocyte traffic into milk differs from these pathological conditions as it is physiological cell trafficking induced by lactation. We hypothesized that physiological movement of leukocytes from the maternal circulation into milk also involves chemokines. We analyzed milk samples from 50 mothers in a cross-sectional study, and in 13 mothers longitudinally for interleukin-8 (IL-8) and in 3 mothers for RANTES. In addition, cultured, human mammary epithelial cells (HMEC) were tested for IL-8 and RANTES production after stimulation with TNF-alpha, INF-gamma, IL-1alpha, IL-1ß and prolactin. These chemokines were chosen, as IL-8 is a prototype C-X-C chemokine which attracts neutrophils and lymphocytes and RANTES is a C-C chemokine which has specific chemotactic activity for monocytes, eosinophils and CD4⁺ CD45RO⁺ lymphocytes, the predominant lymphocyte in milk [3, 7, 9].

Milk samples were collected according to the clinical protocol approved by an ethical committee, from three groups of informed, consenting patients, in the fourth group informed parental consent was obtained.

1. A group of 25 women donated samples in the first three days of lactation following delivery of term infants. Five of these mothers had had a Cesarean section and 3 mothers, a forceps-assisted delivery. All mothers were well nourished, apyrexial and none had inflammation of the breast.

2. A second group of 25 mothers expressing milk for their infants on a special care baby unit donated samples both in early lactation, and two weeks into lactation. Sixteen of these mothers had delivered infants of less than 32 weeks gestation. Ten of the mothers had had Cesarean sections. Twelve of the mothers received two doses of corticosteroid over the 24 hours prior to delivery of their infants. Several mothers gave a succession of samples in a longitudinal study which lasted, for three of them, the first three months of lactation. None of the mothers had pyrexia or mastitis during the study.

3. Four samples of milk were collected from women with galactorrhoea unrelated to pregnancy, receiving therapy at an endocrine clinic.

4. Three, 200 µl samples were collected with an Eppendorf pipette from healthy infants producing "witch's milk" a secretion from breast tissue in the newborn period. Samples were collected in the first week of life.

Samples (200 µl-2ml) of expressed breast milk were collected and divided into aliquots for cellular analysis, creamatocrit, biochemical analysis and chemokine assay. A cell count was performed using a counting chamber as described, and the creamatocrit was measured following centrifugation in a capillary tube [9]. Aliquots for biochemical and chemokine analysis were centrifuged at 1,000 g for 3 min in order to separate the cream, milk and cells within 30 min of expression. The milk component was carefully separated, frozen at 20° C, then stored at 70° C. All assays were carried out within two months of storage in this manner. The sodium content of the milk samples was determined by routine electrochemical analysis.

Cells were collected from milk samples in the first three days of lactation. Samples were centrifuged at 250 g for 8 min, the supernatant discarded and the cell pellet resuspended in sterile RPMI medium. Cells were washed twice, then cultured at a density of one million cells per ml in Nunc vials in RPMI with 5% human AB serum, glutamine and antibiotics at 37° C in 5% CO₂. Samples of the supernatant were collected at regular intervals for assay.

Normal, human mammary epithelial cells (HMEC) were purchased from Clonetics Corporation (Walkersville, MD) and cultured under serum-free conditions as specified by the supplier. HMEC were delivered cryopreserved in the seventh passage and used for experiments within the following three passages. For stimulation, HMEC were seeded in NUNC 96-well plates at a density of 3,000-5,000 cells per well and cultured until they reached 80% confluence. Cells were stimulated in triplicate for 24 hours with 200-250 U/ml TNF-alpha, 50-1,280 U/ml IL-1alpha, 50 U/ml IL-1ß, 50-500 U/ml IFN-gamma (all Genzyme, Cambridge, MA), 1-100 ng/ml Prolactin (Peninsula Laboratories, St. Helens, UK). After stimulation supernatants were stored at 20° C until IL-8 and RANTES concentrations were determined using ELISA. Cells were fixed and protein content was measured by sulforhodamine B-assay.

Cells were fixed for 10 min in 4% paraformaldehyde, washed in distilled water several times and then incubated with a SRB (Sulforhodamine B) solution 0.4% in 1% acetic acid for 30 min. After washing the cells 4-5 times with 1% acetic acid, a 10 mM unbuffered Tris solution was added. Plates were agitated for 5 min and extinction was determined at 550 nm (reference 690 nm).

Enzyme-linked immuno-sorbent assay (ELISA) was carried out in triplicate for IL-8 using own antibodies (Novartis Forschungsinstitut, Vienna) and established methodology. RANTES concentration of cell culture supernatants was determined using a commercial available cytokine ELISA (R&D Systems, Abingdon, UK).

Human breast tissue was removed for diagnostic purposes and stored as archival material in liquid nitrogen. Frozen sections 5-8 µm thick were prepared from the blocks which were judged to contain normal breast tissue. The sections were stained with mouse monoclonal antibodies against human recombinant IL-8 (Novartis Forschungsinstitut) and RANTES (R&D). The binding was visualized by a APAAP detection kit (Dako) and the slides counterstained by hemalaun. Equimolar IgG (Dako) was used as a negative control.

Chemokine concentrations measured in the milk from four patient groups in the cross-sectional study are shown in Figures 1 and 2. IL-8 and RANTES concentrations varied from 0.5 to 6.1 and from 0 to 10.1 ng per ml of collected milk, respectively. No significant differences in chemokine expression were observed comparing term deliveries to pre-term deliveries, or by comparing those two groups to galactorrhoea cases and "witch's milk". Correcting for the sodium concentration of the milk did not change the observed pattern. Chemokine concentrations correlated with sodium concentrations (Spearman's r = 0.71 for IL-8, Figure 3; r = 0.75 for RANTES, data not shown), but not with the creamatocrit (r = 0.23) or cell counts (r = 0.46).

Data collected in the longitudinal study are demonstrated in Figures 4A and 4B. Chemokine concentrations were the highest in colostrum and subsequently decreased, showing a rather irregular pattern. We observed that both chemokines were stable if milk samples were frozen to 20° C, but were destroyed by the routine pasteurization used in the milk bank (not shown).

Immunohistochemical staining of breast tissue by monoclonal antibodies against IL-8 and RANTES revealed immunoreactivity for both chemokines (Figures 5A and 5B), which was observed in the cytoplasm of the epithelial cell lining of the acini and ducts. Chemokine staining was the strongest at the luminal side of the cells.

In vitro cultured, primary, human mammary epithelial cells (HMEC) secreted IL-8 upon stimulation with TNF-alpha, IL-1alpha, IL-1ß and prolactin. TNF-alpha and prolactin only slightly enhanced IL-8 production; there was no prolactin dose-dependency (Figure 6). IL-1alpha and IL-1ß (50 U/ml) increased IL-8 secretion over ten and six fold, respectively; in the case of IL-1alpha a dose-dependency of the response was shown (Figures 6 and 7). RANTES production by HMEC was induced by IFN-gamma in a dose-dependent manner and was not further enhanced by TNF-alpha, which, on its own, had no effect (Figure 8). Similarly, IL-1alpha had only a marginal effect on the induction of RANTES (Figure 8). No difference in chemokine expression was detected when cells were grown on laminin-coated surfaces (data not shown). Protein content per well of cultured cells was detected by sulforhodamine B-assay and did not show any major variations, therefore all the results are presented as the absolute supernatant concentrations found.

Movement of immune cells from the maternal circulation into secreted milk involves crossing several membranes and cell barriers. Leukocytes initially traverse the endothelial cell layer and the basal membrane of maternal vessel walls. This migration is preceded by sequential adhesion steps involving several classes of leukocyte and endothelial cell adhesion molecules; similar features characterize the diapedesis of the neutrophils, monocytes and lymphocytes [10]. Leukocyte integrins are activated upon stimulation of leukocytes by several factors including chemotactic cytokines (chemokines) [11, 12]. Endothelial cell-bound gradients of chemokines induce the direct movement of leukocytes across the endothelium [12]. Once in the breast tissue leukocytes are known to migrate through the interstitium and across either acinary or ductal epithelial lining of the mammary gland into the lumen. Our data suggest that chemoattractants may be the driving force in the trans-epithelial step of leukocyte migration in the mammary tissue, as described in other epithelial tissues [13].

The chemotactic cytokine IL-8, which was recently also found in human sweat [14], and RANTES were observed in human milk by us and others [15] in concentrations sufficiently high to account for the leukocyte recruitment [16]. Our data do not show a correlation between the chemokine concentrations and the leukocyte counts. However this may be due to the fact that it was expressed milk that was sampled rather than secretions from within the secretory ducts.

In addition, the chemokines in milk may be involved in the trafficking of ingested leukocytes across the gut epithelium of a newborn, a phenomenon observed in a number of animal models, predicted by Mohr [17] and Ogra et al. [18], and implied to take place in humans [19] but has not yet been formally proven. High concentrations of IL-8 and RANTES in breast milk, both stable even in a low pH environment (unpublished data), could bind to enterocyte proteoglycans [20], inducing adherence and diapedesis of maternal leukocytes into the gut tissue. IL-8 and RANTES can induce chemotaxis of T-lymphocytes, and primarily CD45RO positive T-lymphocytes, respectively [7, 21]. Transplantation of relatively long-lived, maternal memory T lymphocytes could account for medium and long-lasting, non-inherited maternal effects observed in human, including protection against necrotising enterocolitis, modified lymphocyte response to antigens and the outcome of solid organ transplants [22, 27]. Aside from conferring immunity to the infant, the cellular transplantation from the mother might provide for tolerance to the vast number of food antigens passing through the infant bowel in the early days of feeding.

In multiple pathologies investigated so far, IL-8 and RANTES act as inflammatory mediators [16, 28]. This potential of milk-borne chemokines to induce inflammatory responses is not usually observed in either mother or infant, but does manifest in the mother if the flow of breast milk is slowed for any reason. The rapid onset of mastitis, its hypersymptomatic appearance and the lack of obvious infection in the majority of cases, may relate to the milk's cytokine or, in particular, chemokine content.

Incubation of breast milk leukocytes in culture was not accompanied by the production of either IL-8 or RANTES (data not shown), but cultured human mammary epithelial cells showed IL-8 and RANTES production upon stimulation with different cytokines. Palkowetz and coworkers [29] demonstrated IL-8 production by a spontaneously immortalized human mammary gland epithelial cell line MCF-10 (a result that we were also able to reproduce, data not shown). In these cells, in contrast to primary cells (Figure 6), prolactin does not alter the production of cytokines/interleukins [29] (and own unpublished observation).

Chemokine genes are expressed under different experimental and clinical conditions by the epithelial cells in the lung, liver and sweat glands [14, 30, 31]. Therefore, our observations in human milk may relate to other epithelial cells bordering the external environment [30]. Our data suggest that the source of milk-borne chemokines is the mammary epithelium and the secretion of chemokines is related to the rate of ionic transport and water secretion by these cells. Regulation of this secretion and its role in non-lactating tissue will be important in the understanding of the relationship of immune responses to breast malignancy.

The signals which lead to chemokine production by mammary epithelium are not yet clearly delineated. In contrast to previous reports which used mammary epithelial cell line MCF10 [29], we have shown, using primary cells, that prolactin, a hormone responsible for induction of lactation, stimulates the production of IL-8 in mammary epithelial cells. A prominent role may also be played by the inflammatory cytokines, e.g. IL-1 which, as we show, induces a massive IL-8 secretion by normal, mammary epithelial cells. High levels of IL-1 were found in normal human milk [32, 33]. Therefore, it is possible that IL-1 and maybe other inflammatory cytokines, are responsible for triggering chemokine production in the lactating mammary gland. Additional mediators affecting chemokine production may include prostaglandins. Recently, it was demonstrated that the expression of chemokines can be induced by prostaglandin F_2alpha [34] whereas, several prostaglandins of the F series have been identified in normal human, bovine and murine milk [35].

In vitro cultured, mammary epithelial cells produce only very low levels of chemokines without stimulation by either cytokines or hormones. This is in apparent disagreement with the immunohistochemical evidence of IL-8 and RANTES being present in vivo in normal, non-lactating mammary epithelial cells (Figure 5). This difference may be explained by the fact that the in vivo environment may provide constant stimulation by a variety of factors produced by the adjacent heterologous cells and in distant organs, for example hormones and cytokines. Naturally, the lactating mammary gland should receive higher level of hormonal stimulation whereas background levels of circulating hormones or cytokines may be enough to induce the chemokine production. Additionally, in vivo, the cells are surrounded by complex matrices which may play a co-stimulatory role in the production of chemokines. Also, it is possible that in vivo, the non-lactating mammary epithelial cells produce and store chemokines, while their secretion may be induced by additional stimuli. The complexity of production, storage and secretion of chemokines in lactating and non-lactating mammary gland clearly requires additional in-depths studies.

Chemotactic molecules in human milk may play a significant role in the movement of maternal neutrophils, monocytes and lymphocytes into milk, and perhaps subsequently across the neonatal bowel wall. Such trafficking contributes to mucosal defense and the development of the immune system of the newborn.

We should like to thank Dr. Jozsef Timár for the breast tissue samples and Kamillo Thierer for skillful technical assistance. Dr. Ann Prentice gave valuable advice in her critical review of this paper.

1. Cunningham A S, Jelliffe D B, Jelliffe E F. 1991. Breast feeding and health in the 1980s: a global epidemiologic review.
J. Pediatr. 118: 659.

2. Jain N, Mathur N B, Sharma V K, Dwarkadas A M. 1991. Cellular composition including lymphocyte subsets in preterm and full term human colostrum and milk. Acta Paediatr. Scand. 80: 395.

3. Wirt D P, Adkins L T, Palkowetz K H, Schmalstieg F C, Goldman A S. 1992. Activated and memory T lymphocytes in human milk. Cytometry 13: 282.

4. Seelig L L, Billingham R E. 1981. Concerning the natural transplantation of maternal lymphocytes via milk. Transplant. Proc. 13: 124.

5. Oppenheim J J, Zachariae C O, Mukaida N, Matsushima K. 1991. Properties of the novel proinflammatory supergene "intercrine" cytokine family. (Review) Annu. Rev. Immunol. 9: 617.

6. Pattison J, Nelson P J, Huie P, von Leuttichau I, Farshid G, Sibley R K, Krensky A M. 1994. RANTES chemokine expression in cell-mediated transplant rejection of the kidney. Lancet 343: 209.

7. Schall T J, Bacon K, Toy K J, Goeddel D V. 1990. Selective attraction of monocytes and T lymphocytes of the memory phenotype by cytokine RANTES. Nature 347: 669.

8. Rot A, Krieger M, Brunner T, Bischoff SC, Schall TJ, Dahinden CA. 1992. RANTES and macrophage inflammatory
protein-1 alpha induce the migration and activation of normal human eosinophil granulocytes. J. Exp. Med. 176: 1489.

9. Paape MJ, Keller M. 1985. Determination of the numbers and functions of cells in human milk. In: Jensen RG and Neville MC. Eds. Human Lactation, Milk components and Methodologies, New York: Plenum Press, 53.

10. Butcher EC. 1991. Leukocyte-endothelial cell recognition: three (or more) steps to specificity and diversity. (Review) Cell 67: 1033.

11. Rot A. 1992. Endothelial cell binding NAP-1/IL-8: role in neutrophil emigration (Review). Immunol. Today 13: 291.

12. Middleton J, Neil S, Wintle J, Clark-Lewis I, Moore H, Lam C, Auer M, Hub E, Rot A. 1997. Transcytosis and surface presentation of IL-8 by venular endothelial cells. Cell 91: 385.

13. Madara JL. 1994. Migration of neutrophils through epithelial monolayers. Trends Cell. Biol. 4: 4.

14. Jones A P, Webb L M, Anderson A O, Leonard E J, Rot A. 1995. Normal human sweat contains interleukin-8. J. Leukoc. Biol. 57: 434.

15. Rudloff H E, Schmalstieg F C Jr, Palkowetz K H, Paszkiewicz E J, Goldman A S. 1993. Interleukin-8 in human milk. J. Reprod. Immunol. 23: 13.

16. Strieter R M, Koch A E, Antony V B, Fick R B Jr, Standiford T J, Kunkel S L. 1994. The immunopathology of chemotactic cytokines: the role of interleukin-8 and monocyte chemoattractant protein-1. (Review) J. Lab. Clin. Med. 123: 183.

17. Mohr JA. 1973. The possible introduction and/or acquisition of cellular hypersensitivity associated with the ingestion of colostrum. J. Pediatr. 82: 1062.

18. Ogra S S, Weintraub D, Ogra P L. 1977. Immunologic aspects of human colostrum and milk. III Fate and absorption of cellular and soluble components in the gastrointestinal tract of the newborn. J. Immunol. 119: 245.

19. Michie C A, Harvey D. 1995. Maternal milk lymphocytes engraft the fetal gut. Ped. Res. 37: 129A.

20. Webb L M, Ehrengruber M U, Clark-Lewis I, Baggiolini M, Rot A. 1993. Binding to heparan sulfate or heparin enhances neutrophil responses to interleukin-8. Proc. Natl. Acad. Sci. 90: 7158.

21. Leonard E J, Skeel A, Yoshimura T, Noer K, Kutvirt S, Van Epps D. 1990. Leukocyte specificity and binding of human neutrophil attractant/activation protein-1. J. Immunol. 144: 1323.

22. Michie CA, McLean A, Alcock C, Beverley P C. 1992. Lifespan of human lymphocyte subsets defined by CD45 isoforms. Nature 360: 264.

23. Ten Wolde S, Breedveld F C, de Vries R R, D'Amaro J, Rubenstein P, Schreuder G M, Claas F H, van Rood J J. 1993. Influence of non-inherited maternal HLA antigens on occurrence of rheumatoid arthritis. Lancet 341: 200.

24. Duncan B, Ey J, Holberg C J, Wright A L, Martinez F D, Taussig L M. 1993. Exclusive breast feeding for at least 4 months against otitis media. Pediatrics 91: 867.

25. Beeby P J, Jeffery H. 1992. Risk factors for necrotising enterocolitis: the influence of gestational age. Arch. Dis. Child. 67: 432.

26. Zhang L, van Bree S, van Rood JJ, Claas F H. 1991. Influence of breast feeding on the cytotoxic T cell allorepertoire in man. Transplantation 52: 914.

27. Campbell D A, Lorber M I, Sweeton J C, Turcotte JG, Niederhuber J E, Beer A E. 1984. Breast feeding and maternal-donor renal allografts. Transplantation 37: 340.

28. Rot A. 1996. Inflammatory and physiological roles of chemokines. Path. Oncol. Res. 2: 16.

29. Palkowetz K H, Royer C L, Garofalo R, Rudloff H E, Schmalstieg F C Jr, Goldman A S. 1993. Production of IL-6 and IL-8 by human gland epithelial cells. J. Reprod. Immunol. 26: 57.

30. Standiford T J, Kunkel S L, Basha M A, Chensue S W, Lynch J P 3d, Toews G B, Strieter R M. 1990. Interleukin-8 gene expression by a pulmonary epithelial cell line. A model for cytokine networks in the lung. J. Clin. Invest. 86: 1945.

31. Thornton A J, Strieter R M, Lindley I, Baggiolini M, Kunkel S L. 1989. Cytokine induced gene expression of a neutrophil chemotactic factor/IL-8 in human hepatocytes. J. Immunol. 144: 2609.

32. Söder O. 1987. Isolation of interleukin-1 from human milk. Int. Archs Allergy appl. Immunol. 83: 19.

33. Munoz C, Endres S, van der Meer J, Schlesinger L, Arevalo M, Dinarello C. 1990. Interleukin-1ß in human colostrum. Res. Immunol. 141: 505.

34. Haworth J D, Rollyson M K, Silva P, McIntush E W, Niswender G D. 1998. Messenger ribonucleic acid encoding monocyte chemoattractant protein-1 is expressed by ovine corpus luteum in response to prostaglandin F_2alpha. Biol. Reprod. 58: 169.

35. Grosvenor C E, Picciano M F, Baumrucker C R. 1992. Hormones and growth factors in milk. (Review) Endocr. Rev. 14: 710.

36. Wilkinson L, Blank G, Gruber C. 1996. In: Desktop Data Analysis with Systat. New Jersey, Upper Saddle River, Prentice Hall.