Main

Breast-feeding is associated with a reduction in the incidence of gastrointestinal illness(1–3), respiratory illness(3), and otitis media(1,3). Long-term epidemiologic studies suggest a link between breast-feeding and a lower incidence of some diseases appearing later in life such as type I diabetes(4), allergies(2,3), and Crohn's disease(5). This supports the concept that breast milk contains immunomodulating agents that not only may act directly to neutralize the detrimental effects of pathogens encountered by the infant but also may influence the maturation of the infant's own developing immune system.

Among the many potentially active immunologic components identified in human breast milk are the polypeptide cytokines that often act in femtomolar concentrations to regulate the behavior or properties of cells associated with the immune system. A range of both proinflammatory and antiinflammatory cytokines have been detected by a variety of methodologies in breast milk collected at various stages of lactation. These include IL-1β(6–9), IL-6(8–15), TNF-α(6,8,9,14–18), and TGF-β(8,19–22). Many of these studies had small sample sizes and most looked for the presence of cytokines in colostrum or "early" milk only.

Another family of substances present in human milk that has the capacity to influence physiologic functions of the newborn infant are the lipid eicosanoids such as PGE2(23–25). Prostaglandins are derived from essential fatty acids supplied in the diet and are involved in the modulation of gastric acid and mucous secretion, homeostasis, smooth muscle contraction, cytoprotection, and leukocyte function. More recently, PGE2 has been shown to be associated with the regulation of proinflammatory cytokine production by monocytes and macrophages(26,27).

In this longitudinal study, we used a sensitive and specific ELISA method to determine the concentration of IL-1β, IL-6, TNF-α, TGF-β1, and TGF-β2 in samples obtained sequentially over 12 wk from a large cohort of healthy mothers delivering at term. In addition, we determined the concentration of PGE2 by RIA during this period.

METHODS

Subjects. Forty-nine mothers who planned to breast-feed for at least 12 wk and who had delivered healthy term infants were recruited from the postnatal ward at Flinders Medical Centre. All women were healthy, aged 18 y or more, and were not taking antiinflammatory medication at the time of enrollment. The present study was approved by the Flinders Medical Centre Committee on Clinical Investigation (Ethics) and written informed consent was obtained from all participants.

Sample collection. In this longitudinal study, milk samples (2-5 mL) were collected from each mother by manual expression into sterile polypropylene containers on d 1-6 and at 2, 4, 6, 8 and 12 wk postpartum either at home or at the hospital. All samples were stored at 4°C and processed within 2 h of collection. After centrifugation at 690 × g for 20 min, the aqueous phase was collected and stored at -80°C until analyzed for cytokine/eicosanoid content. All samples were analyzed for IL-1β, IL-6, and TNF-α within 16 wk of collection and for TGF-β1, TGF-β2, and PGE2 after storage for 20-30 wk. Information regarding recent illnesses or infections and the use of any medications was collected from each mother at each collection time.

Immunoassays for IL-1β, IL-6, TNF-α. ELISA for IL-1β, IL-6, and TNF-α were established using matched antibody pairs (IL-1β, Endogen, Inc., Cambridge, MA; IL-6 and TNF-α, R&D Systems, Minneapolis, MN). Briefly, 96-well flat-bottomed plates (Nalge Nunc International, Naperville, IL) were incubated with 2 µg/mL mouse anticytokine MAb in carbonate/bicarbonate buffer (50 mM, pH 9.6) overnight at room temperature. Each well was washed 3 times with wash buffer (PBS/0.05% Tween 20) and the blocked with PBS/2.5% BSA for 1 h at 37°C. Test samples or cytokine standards (in duplicate) diluted in assay buffer (PBS/0.05% Tween 20/1% BSA) were added to washed wells and incubated for 2 h at 37°C. After washing, biotinylated mouse anti-IL-1β MAb (500 ng/mL) or biotinylated goat anticytokine antibody (IL-6, 25 ng/mL and TNF-α, 200 ng/mL) diluted in assay buffer was added to each well for 2 h at 37°C. Washed wells were incubated with ExtrAvidin alkaline phosphatase (Sigma Chemical Co. BioSciences, St. Louis, MO) at 1:5000 dilution in assay buffer for 1 h at 37°C. After a further three washes, the substrate solution (p-nitrophenyl phosphate in diethanolamine buffer) was added to each well, and the color reaction was allowed to proceed until read on an automatic plate reader at OD 405 nm. The limit of detection for each ELISA was 15 pg/mL. Complete recovery of added recombinant cytokine standards was obtained from expressed breast milk aqueous-phase samples, indicating a lack of interfering substances in these assays.

Immunoassays for TGF-β1 and TGF-β2. ELISA for TGF-β1 and TGF-β2 were established using matched antibody pairs (R&D Systems, Minneapolis, MN) as described above with the following modifications. After coating with MAb, the wells were blocked with PBS/0.05% Tween 20/5% sucrose/0.05% sodium azide for 1 h. Samples (100 µL) were acidified by the addition of 1N HCl (20 µL) for 10 min at room temperature followed by neutralization with 20 µL 1.2N NaOH/0.5 M HEPES. Standards and acidified samples were diluted in 100 mM Tris-HCl/150 mM NaCl/0.05% Tween 20/0.1% essential fatty acid-free BSA, pH 9.6 as required followed by a 1:2 dilution in PBS/0.05% Tween 20/1.4% essential fatty acid-free BSA assay buffer and were incubated for 30 min on a shaker at room temperature followed by 90 min at 37°C. Biotinylated goat anti-TGF-β antibody was used at a concentration of 100 ng/mL. The limit of detection for each ELISA was 30 pg/mL. The recovery of recombinant TGF-β1 and TGF-β2 added to appropriately diluted breast milk samples ranged from 75-85%. According to the manufacturer, there is <1% cross-reactivity between the two isoforms of TGF-β when the appropriate antibody combinations are used.

Immunoassay for PGE2. Aqueous-phase samples were diluted 1:2 in Tris/NaCl buffer and PGE2 concentrations were measured in triplicate by RIA as described previously(28). The limit of detection for the assay was 10 pg/mL.

Data analysis. Cytokine/eicosanoid data are presented as mean ± SEM. For calculations of mean concentrations of groups of all data, those samples below the level of detection were assigned a value of half the limit of detection, i.e. 7.5 pg/mL for IL-1β, IL-6, and TNF-α and 5 pg/mL for PGE2. Because the point of left truncation is very small when compared with the mean, this manipulation of the data introduces only a small error, i.e. there is little change in the calculated mean whether those samples below the level of detection are assigned a value of 15, 7.5, or 0 pg/mL. The relationships between immune factors at each time were determined by Pearson correlation using SPSS for Windows 6.0 (SPSS Inc., Chicago, IL), as were the associations between subject characteristics and cytokine/eicosanoid concentration.

RESULTS

Subject characteristics. Forty-nine women between 18-38 y of age and parity 1-5 were recruited into the study. Twelve (24%) of the participants smoked cigarettes, 78% had delivered vaginally, and, at each collection time, a maternal illness or infection was reported by 11-27% of mothers who provided a sample. Each mother provided a colostral sample 1-6 d after delivery. Additional milk samples were obtained from 48 of these mothers at wk 2 (d 11-17), 45 at wk 4 (d 24-33), 42 at wk 6 (d 39-47), 37 at wk 8 (d 53-60), and 36 at wk 12 (d 80-88). Ninety-seven percent (250/257) of samples were collected from mothers who reported that they were exclusively breast-feeding their infants.

Postpartum changes in cytokine/eicosanoids. Cytokine/eicosanoid levels in human milk samples collected at each time postpartum are presented in Figures 1–3 (individual data points) and Tables 1 and 2 (mean ± SEM, range). At each time, the mean concentration of each cytokine/eicosanoid was compared with all other cytokines/eicosanoid. No associations were found. There was no correlation between smoking, birth type, or maternal illness/infection and cytokine or PGE2 concentration.

Figure 1
figure 1

A-C, Concentrations (pg/mL) of IL-1β, IL-6, and TNF-α in 257 colostrum and milk samples obtained from 49 lactating women. Numbers in parentheses correspond to the number of samples below the level of detection at each time.

Full size image
Figure 3
figure 3

Concentrations (pg/mL) of PGE2 in 257 colostrum and milk samples obtained from 49 lactating women. Numbers in parentheses correspond to the number of samples below the level of detection at each time.

Full size image
Table 1 The concentrations (mean ± SEM, pg/mL) of IL-1β, IL-6, and TNF-α in samples collected at each time throughout the study. The total number of samples tested at each time and the number of samples with detectable levels of cytokine are indicated
Full size table
Table 2 The concentrations (mean ± SEM, pg/mL) of TGF-β1, TGF-β2 and PGE2 in samples collected at each time throughout the study. The total number of samples tested at each time and the number of samples with detectable levels of cytokine are indicated
Full size table

IL-1β, IL-6, and TNF-α (Table 1 and Fig. 1,A-C) were detected in 5-51% of samples collected at each time in the study. Overall, mean levels seemed to be higher in early milk and then decreased as lactation continued. With each of these cytokines, there was a wide range of concentrations detected at each time in the study (Fig. 1,A-C).

After acidification, TGF-β1 and TGF-β2 (Fig. 2 and Table 2) were detected in all samples tested throughout the study. TGF-β2 was present in approximately 10-fold higher amounts than TGF-β1, and there was a wide range of values at each time of lactation. There was little change in the mean concentration of either isoform of TGF-β present in milk samples during the first 12 wk of lactation.

Figure 2
figure 2

A and B, Concentrations (pg/mL) of TGF-β1 and TGF-β2 in colostrum and milk samples (TGF-β1, n = 252; TGF-β2, n = 257) obtained from 49 lactating women.

Full size image

PGE2 (Fig. 3 and Table 2) was detected in the aqueous phase of 47/49 (96%) colostral samples, 41/42 (98%) milk samples collected at wk 6, and all remaining samples tested. PGE2 content increased from early lactation and then remained steady for the period of the study.

Cytokine profile of milk from individual mothers over 12 wk lactation. The longitudinal design of the study enabled us to investigate the relationship between different breast milk cytokines of individuals from delivery to 12 wk postpartum. Of 36 mothers who provided a sample at every time, seven did not have detectable levels of any of the proinflammatory cytokines (IL-1β, IL-6, TNF-α) at any stage of lactation. In some other individuals, these proinflammatory cytokines seemed to track together over time; however, overall no consistent pattern of association between the cytokines measured was observed. There was no relationship between the appearance of the proinflammatory cytokines and maternal reports of an infection or illness immediately preceding or at the time of milk collection. The postpartum changes in TGF-β1 and TGF-β2 seemed to follow each other in some individuals, although there was no statistically significant association between the two. High levels of proinflammatory cytokines were not statistically associated with low levels of immunosuppressive cytokines in any individual.

DISCUSSION

The composition of human milk is subject to change during lactation; however, there is little known about postpartum changes in its immunologic constituents. This is the first comprehensive study investigating a range of cytokines in colostrum and milk samples obtained from delivery to 3 mo postpartum from a single large cohort of women. Our results clearly show the extensive heterogeneity in the concentration of all cytokines measured as well as PGE2 in milk among subjects at each collection time.

There is diversity in the reported values for the proinflammatory cytokines in human breast milk (Table 3). Our finding of IL-1β present in a small proportion of samples is in contrast with the data described by both Srivastava et al.(8) and Basolo et al.(15) who were unable to detect any IL-1β and Munoz et al.(6) who reported IL-1β at high concentrations in all colostral samples tested (Table 3). Similarly, our data for IL-6 contrast with reports of low amounts of colostral and mature milk IL-6 (maximum 34 and 42 pg/mL, respectively) by one group(8) but are comparable to data for colostrum from other authors(9,13) (Table 3). Although IL-6 shares many proinflammatory functions with IL-1β and TNF-α, it has also been associated with the local production of IgA in the breast(10), which may be an important function in the breast-fed infant.

Table 3 Overview of studies investigating the presence of IL-1β, IL-6, TNF-α in human colostrum and milk. The method of detection, stage of lactation (in days postpartum), number of samples in each study, mean concentration (pg/mL ± SEM or SD as indicated), and range of concentrations detected are described where possible
Full size table

The presence of substantial amounts of two soluble TNF receptors (sTNFRI and sTNFRII) in colostrum and milk has been described(18). These receptors can bind TNF-α and thus reduce the amount of "free" antigen available for detection by an ELISA that is unable to recognize both free and bound antigen. However, the matched antibody pairs purchased for the establishment of the TNF-α ELISA used in the present study have been shown to have no significant cross-reactivity or interference with either soluble receptor and, hence, measure total TNF-α (R&D Systems, Technical Service, personal communication). Indeed, our results are in the ranges reported by Buescher and Malinowska(18) who used an enzyme-amplified sensitivity immunoassay based on multiple MAb that enabled recognition of total (free plus bound) antigen and Na et al.(14), rather than the extremely low concentrations (<2-26 pg/mL) that had also been reported(8) (Table 3).

TGF-β is synthesized and secreted as a biologically inactive complex associated with a precursor molecule that is designated the latency-associated peptide. Active TGF-β can be released from the latency-associated peptide by a variety of treatments, the most common being acidification, which is a possible mechanism of activation in the infant gut. In the present study, free TGF-β was not detected in any samples; however, after acidification, all colostrum/milk samples contained considerable amounts of TGF-β1 and TGF-β2. In general, both isoforms of TGF-β seem to have overlapping biologic activities in vitro; however, TGF-β2 was present at approximately 10-fold greater concentrations than TGF-β1 throughout our study. It is likely that each isoform has specific roles, as suggested by studies with TGF-β1 null mice in which the TGF-β1 "knockout" offspring develop an excessive inflammatory response leading to cellular infiltration of vital organs and death at 3-5 wk of age(29). The null newborn pups survive initially, apparently due to maternal sources of TGF-β1 via both placental transfer and then milk while suckling(30). These studies indicate that the immunosuppressive properties of TGF-β are both conserved after ingestion by the neonatal pup and are vital for immune regulation. In addition to the immunosuppressive properties of the TGF-β family, other functions likely to be of significance to the breast-fed infant are involvement in the induction of oral tolerance(21,31,32), stimulation of IgA isotype switching in B cells(33), and maintenance of intestinal epithelium barrier function(34).

The role of PGE2 in breast milk is unclear. Although synthesis of IL-1β and TNF-α by monocytes/macrophages can be inhibited by PGE2 in vitro(26,27), which may account for some of its down-regulating effects on lymphocyte proliferation, there was no statistical association between cytokine and PGE2 concentrations in our study.

The broad range of concentrations of IL-1β, IL-6, TNF-α, TGF-β1, TGF-β2, and PGE2 in milk samples collected during the first 12 wk of lactation and the lack of correlation among them makes interpretation difficult. Analysis of our data indicates that factors such as smoking, birth type, and maternal illness/infection did not contribute to the wide variation in values. Time since last feed and whether the sample was fore milk, hind milk, or a complete expression may, however, be confounding variables. Samples from individual mothers did not contain consistently high (or low) levels of either proinflammatory or antiinflammatory cytokines, although there was a trend for both isoforms of TGF-β to follow a parallel pattern of change throughout lactation. It is apparent from the existing literature that whereas TGF-β and PGE2 are consistently present in human milk samples in readily detectable quantities(8,25), there are huge discrepancies in reported values (and incidence) of cytokines such as IL-1β, IL-6, and TNF-α (Table 3). These differences are not readily explained by differences in methodology, stage of lactation, or maternal characteristics.

Our data support the accumulating evidence that immunomodulating agents are normally present throughout lactation in quantities that have the potential to exert in vivo effects on the recipient infant. For most in vitro applications, cytokines exert their biologic activity in the concentration range of 1 pg/mL to 10 ng/mL. A newborn infant consumes 400-500 mL of breast milk per day, and consumption increases to approximately 800 mL/d by 3 mo of age(35). Therefore, although the mean concentration of some of the cytokines measured decreased as lactation continued, many infants would be consuming nanogram or microgram amounts per day throughout the first 3 mo of life. The specific immunologic effects of cytokines in human breast milk are likely to be determined by the balance between a number of bioactive factors with complementary and competing activities. Breast milk clearly contains components with immunosuppressive properties that provide a mechanism for evolving immune tolerance and thus protect against the development of overt inflammation and immunologically mediated diseases. On the other hand, the presence of proinflammatory cytokines in breast milk may be important in activating the developing neonatal immune system and in providing immunologic protection in the infant with an immature cytokine network.