Main

Several soluble anti-infective components are present in human milk such as specific secretory IgA antibodies and nonspecific components, including LF and lysozyme(1). These components may protect against pathogens without provoking clinically evident inflammation. Goldman et al.(2) pointed out that human milk is poor in initiators and mediators of inflammation, but rich in anti-inflammatory agents. The iron-binding LF is also present in specific granules of polymorphonuclear leukocytes and in exocrine secretions other than milk(35). LF is associated with host defense at mucosal surfaces through its antibacterial and iron-binding properties. Recently a pepsin-cleaved fragment of human and bovine LF, LFcin, was found to contain the structural domain responsible for the bactericidal properties of LF(6, 7).

LF receptors are found on many types of cells including monocytes and macrophages(8), lectin-stimulated human peripheral blood lymphocytes(9), brush-border cells(1013), and tumor cell lines(e.g. HT-29, HL-60, K562)(1416). In addition to the role of LF as an essential growth factor for both human B and T lymphocytic cell lines(17) and as an inducer of growth of HT-29 cells(18), LF is a negative regulator of myelopoiesisin vitro and in vivo(8, 19). This latter function is mediated through suppression of IL-1 and granulocyte/macrophage-colony-stimulating factor release from monocytes and macrophages(20, 21).

After binding of bacterial LPS to macrophages, T cells, and cultured human monocytes, these cells synthesize TNF-α, IL-1, IL-6, and colony-stimulating factor(2224). Cells participating in the inflammatory response carry several different LPS-binding receptors(25, 26). Such cells also have receptors for LF. Miyazawa et al.(15) observed an interaction between LPS and LF, the complex being bound to the cells also via LPS receptors. Recently, Crouch et al.(27) showed that LF exerted an inhibitory effect on the production of IL-1 and TNF in LPS-stimulated monocytes.

In the present investigation the inhibitory effect of human milk LF on the LPS-induced IL-6 response of human monocytic cells was studied. Also, bovine LF and its fragment, LFcin, were investigated.

METHODS

Endotoxin. Escherichia coli O6K13H1 or O18K1 were cultured on nutrient broth agar plates containing 1% glucose at 37°C over night. Bacteria were harvested and washed. The pellet was extracted with the hot phenol-water method for preparation of LPS(28). LPS from E. coli O127 was purchased from Sigma Chemical Co. (St. Louis, MO).

Shed cell wall fragments from E. coli O18K1 were obtained by culturing bacteria in minimal base medium. After centrifugation, the supernatant was filtrated through a 0.2-μm filter using Mediakap (Microgon, Laguna Hills, CA). The supernatant was concentrated by ultrafiltration(Ultrasette; membrane filter Omega 100K, Filtron, Northborough, MA) and then dialyzed against distilled pyrogen-free water. The pellet of bacterial cells was washed twice. Both the concentrated supernatant and bacterial cells were lyophilized.

Bacteria, shed cell wall fragments, and purified LPS were all analyzed by gas chromatography as described earlier for their content ofβ-hydroxymyristic acid, a marker of the LPS(29). Two nanograms of O18 LPS corresponded to 14 ng of E. coli O18K1 bacterial cells (dry weight), and to 95 ng of shed cell wall fragments (dry weight).

LF. Human milk LF was purchased from Sigma Chemical Co.(approximately 95% purity as measured by SDS-PAGE). Bovine LF and a characterized pepsin-derived LF peptide fragment thereof. LFcin (both demonstrating a purity over 95% as analyzed by HPLC), were kindly provided by Dr. W. Bellamy at the Morinaga Industry C.O., Ltd., Japan(6, 7). The bovine LF was prepared by the procedure of Law and Reiter(30).

To obtain a low content of endotoxin in the LF preparations, they were dissolved in pyrogen-free water and mixed with an endotoxin-absorbing gel(KuttsuClean-D, Maruha Co., Ibaraki, Japan). A peptide of the Tachyplesin family constituted the endotoxin-binding ligand of the gel. After incubation overnight the gel was removed by centrifugation. The endotoxin content of such purified LF, which was tested by the Limulus lysate assay(Chromogenix, Mölndal, Sweden) ranged between 12 and 36 EU/mg(corresponding to approximately 1-3 ng of LPS/mg of LF). The bovine LF contained 72 EU/mg and LFcin less than 2.4 EU/mg.

Cell stimulation by LPS. THP-1 cells, human monocytic cell line. THP-1 cells (ATCC TIB 202) were grown in RPMI 1640 medium supplemented with 10% heat-inactivated FCS, and β-mercaptoethanol. Exponentially growing cells were pretreated by adding IFN-γ (produced inE. coli, >99% pure, endotoxin level <10 EU/1 × 105 U, Boehringer Mannheim Biochemica, Germany) at a concentration of 200 U/mL 16 h before the cell experiment unless otherwise stated. IFN-γ increases the sensitivity for stimulation with LPS(31). Cells were washed by centrifugation at 1500 rpm for 10 min. The cell density was adjusted to 1 × 106 cells/mL using fresh complete medium(RPMI 1640 supplemented with 5% FCS and 1% gentamicin). The cells (400 μL) were added to the wells of a 24-well plate. Stimulation by LPS was carried out for 20-24 h at 37°C. The LF preparations or LFcin were added either before or after the addition of LPS. A control of each preparation was always included in each run. No LF preparation or LFcin induced by itself an IL-6 response. After the incubation period, the cell culture media were removed and centrifuged for 10 min at 400 × g (4°C). The supernatants were stored at -20°C. In the absence of FCS a lower response was seen, approximately 50-100 times (analyzed with both an immunoassay and a bioassay). Natural human TNF-α (purified by immunoaffinity chromatography followed by gel filtration chromatography, 99% pure, endotoxin level <50 pg/mL, Genzyme, Boston, MA), was used as another IL-6-inducing agent (400 U/mL).

The viability of the THP-1 cells incubated for 20 h ranged between 90 and 98% in the presence of 50 μg/mL of human LF, bovine LF, or LFcin as measured by trypan blue exclusion. Addition of LPS did not reduce the viability significantly (85-95%).

Human blood monocytes. Fresh monocytes were isolated from venous blood by centrifugation on Lymphoprep (Ficoll-Hypaque). The cells were washed in ice-cold RPMI 1640 to remove thrombocytes. The platelet-free cells were suspended in 5% human serum and adjusted to 4 × 106 cells/mL. The cells were allowed to adhere for 30-45 min at 37°C in 24-well plates(Nunc, Roskilde, Denmark) adding 2 × 106 cells/well. The wells were washed three times to remove nonadherent cells (approximately 10% of the total amount of cells added remained in the well). RPMI 1640 medium containing 2% FCS, LPS, and LF were added to a final volume of 1 mL/well. Culture supernatants were collected after 20 h of incubation in a CO2 incubator.

IL-6 bioassay. A subclone of the cell line B13.29, which is dependent on IL-6 for growth, was used(32, 37). The cells were harvested from the tissue culture flasks and seeded into microtiter plates (Nunc, Roskilde, Denmark) at a concentration of 5000 cells/well. Samples (diluted 1:50 and 1:250) or IL-6 standard (Genzyme Corporation, Cambridge, MA) were added to the cells and incubated for 68 h.[3H]Thymidine was added 4 h before harvesting of cells. The IL-6 release in one cell experiment analyzing the LPS dose response was verified by using a human IL-6 ELISA kit (correlation coefficient, r = 0.95)(Hycyte, Hycult b.v. An Uden, The Netherlands). LF did not significantly affect the bioassay.

RESULTS

Effects of human LF added after LPS on the IL-6 response of THP-1 cells. Human LF added in amounts of 50 μg/mL, after the LPS, reduced the IL-6 response between 50 and 100% irrespective of the LPS concentration used(10-1000 ng/mL) (Figs. 1 and2). Decreasing the concentration of LF diminished the inhibition in a dose dependent manner(Fig. 1). In one experiment an inhibition of 100, 62, and 78% were obtained in the presence of 50 μg/mL of LF for the LPS concentrations 10, 100, and 1000 ng/mL, respectively (Fig. 2). Increasing the LF concentration to 100 μg/mL suppressed the IL-6 response further to 96 and 81%, respectively, for the LPS doses 100 and 1000 ng/mL (Fig. 2).

Figure 1
figure 1

Inhibition of the LPS-induced Il-6 response of THP-1 cells by human LF added 15-30 min after the LPS (1000 ng/mL = •. 100 ng/mL = ). Because the range of the IL-6 values shifted somewhat in repeated experiments, the IL-6 levels are expressed as percent of the maximal IL-6 response at the LPS dose tested without LF. Data points with SEM bars indicate the means of three separate experiments. Each experiment was run in duplicate and the culture supernatants analyzed in triplicate.

Figure 2
figure 2

Inhibition of the LPS induced IL-6 response of THP-1 cells by human LF added 30 min after LPS in one experiment. The cells were stimulated with 1000 ng/mL (), 100 ng/mL (▪), or 10 ng/mL (•) of LPS. The data points represent the mean of two samples, each analyzed in triplicate.

As early as 4 h of incubation the inhibitory activity of LF (50 μg/mL) was demonstrated for the highest LPS concentration (Fig. 3). Low Il-6 levels were obtained when IFN-γ was omitted. LF also inhibited the production of IL-6 by IFN-γ-pretreated THP-1 cells that were subsequently stimulated with TNF (46% inhibition) (Fig. 4).

Figure 3
figure 3

The effect of human LF on the O6 LPS-induced Il-6 response of THP-1 cells at 4 and 24 h. Cells were incubated with LPS and after 15 min 50 μg/mL of human LF was added (only LPS = •. LPS and human LF= ▪). Each data point represent the mean of two supernatant samples analyzed in triplicate (± SEM). Cells were preincubated with (-) or without (- - - - -) IFN-γ for 3 d.

Figure 4
figure 4

Inhibition of the TNF induced IL-6 response in THP-1 cells by addition of human LF. The cells were stimulated by 400 U/mL of TNF. Human LF (50 μg/mL) was added 15 min after TNF. The bars represent the mean of two samples run in triplicate (standard errors of the means are indicated).

Effects of human LF added before LPS on the IL-6 response of THP-1 cells or human monocytes. When human LF (0.5-50 μg/mL) was added before LPS the inhibition of the IL-6 response in THP-1 cells was mostly not reduced to more than approximately 50% of the Il-6 response obtained without LF(Fig. 5). A maximal reduction was observed at 0.5 μg LF/mL, using LPS concentrations of 10-1000 ng/mL for stimulation of the cells. At 5-50 μg/mL of LF no further inhibition of the Il-6 production was seen. An experiment with fresh monocytes using 1 ng/mL of LPS for stimulation demonstrated a similar pattern for LF regarding its capacity to inhibit the IL-6 response (Fig. 5). An inhibition of 40-50% was seen at 0.5-5 μg/mL.

Figure 5
figure 5

Inhibition of the LPS-induced IL-6 response of THP-1 cells (•) or fresh monocytes (▪) by human LF added 30 min before the O127 LPS. A concentration of 1 ng/mL was used for the fresh monocytes. The IL-6 levels of the THP-1 cells represent the means of five experiments(±1 SEM) and those of fresh cells one experiment. Ten to 1000 ng/mL of LPS was used for stimulation of THP-1 cells.

The inhibitory effect of bovine LF and LFcin on the LPS-induced IL-6 response. The inhibitory capacity of human or bovine LF (50 μg/mL) added after LPS was stronger at 10 ng than at 100 ng of LPS/mL, comparing the reduction in percent of total IL-6 secreted (Fig. 6A). LFcin, however, showed the strongest inhibition at 100 ng of LPS/mL by this comparison. When LF (5 μg/mL) was added before LPS, the inhibitory capacity of bovine LF increased with the LPS dose (Fig. 6B). LFcin behaved in the same way as when added after LPS, i.e. a strong inhibitory activity at both LPS concentrations.

Figure 6
figure 6

Inhibition of the LPS-induced IL-6 response in THP-1 cells by addition of human (LFhum) or bovine LF (LFbov), and LFcin. The LF or LFcin was added either after (A, 50 μg/mL) or before (B, 5 μg/mL) LPS. The cell experiment was run in duplicate and each sample analyzed in triplicate.

The inhibitory activity of the LF preparations on the IL-6 secretions was confirmed by analysis with an immunoassay as shown for bovine LF and LFcin(Fig. 7).

Figure 7
figure 7

Comparison of the IL-6 levels in supernatants from LPS-stimulated THP-1 cells by the bioassay and an ELISA. Bovine LF(LFbov) or LFcin in a dose of 50 μg/mL was added 30 min after the LPS (100 ng/mL).

The inhibitory effects of LFcin on the IL-6 response induced by LPS, shed cell wall fragments, or whole bacteria. Cells were also stimulated by shed cell wall fragments or whole bacteria of the E. coli O18K1 serotype to test whether native forms of LPS would be inhibited in their IL-6 inducing capacity by LFcin. A concentration corresponding to 2 ng of LPS/mL was used for stimulation of THP-1 cells with the bacterial cell wall fragment preparation or bacterial cell suspension. LFcin (50 μg/mL) added after the endotoxin-containing preparations reduced the IL-6 response 71-92% (Fig. 8). LFcin added at a concentration of 0.5μg/mL before endotoxin preparation also suppressed the IL-6 responses except for that induced by bacterial cells. The suppression was, however, lower, being 38 and 64% for cell wall fragments (shed endotoxin) and LPS, respectively. LFcin seemed to enhance the IL-6 response induced by the bacteria.

Figure 8
figure 8

Inhibition by LFcin of the IL-6 response in THP-1 cells stimulated by shed cell wall fragments, whole bacteria or purified LPS (2 ng/mL) of E. coli O18K1. The concentration of cell wall fragments or bacteria used for the cell-stimulation corresponded to 2 ng of LPS/mL. LFcin was added either 30 min before (0.5 μg/mL) or 30 min after (50 μg/mL) the endotoxin.

DISCUSSION

In this study we report that LF from human milk reduces the LPS-induced IL-6 response when added to fresh human monocytes or cultured monocytic cells. Thus, LF may act as a potential anti-inflammatory agent. Because LF inhibited the cytokine response when added 30 min after LPS, it seems most probable that LF exerts its effects directly on the cells. This was also supported by the inhibitory effect of LF on the TNF-α-induced IL-6 response.

LF forming complexes with LPS has been suggested to abrogate the inhibitory effects of LF. Formation of LF-LPS complexes may be one reason for the lack of complete inhibition when human LF was added before LPS. However, because the same pattern of inhibition was obtained with various LPS concentrations in our study, it seems less probable that LPS-LF complexes interfered with the inhibitory activity. A similar inhibition by LF was observed by Crouchet al.(27) studying the TNF and IL-1 responses in LPS-stimulated fresh human blood mononuclear cells. In that study even lower concentrations of LF suppressed the cytokine responses. The absence of IL-6 suppression when adding LFcin before bacterial cells may be explained by binding of LFcin to the bacterial surface, particularly to outer membrane proteins(33), thereby decreasing the concentration of free LF. It should be noted that a 100 times lower concentration of LFcin was used when added before bacterial cells compared with that added after.

The difference in LF doses needed for maximal inhibition due to the time at which LF was added in relation to LPS, suggests a change in the responsiveness toward LF after LPS-stimulation. That LPS induces rapid alterations in the expression of cell-surface receptors on human monocyte is known(34). LPS has shown to down-regulate both the TNF and IL-6 receptors on human monocytes(34, 35). It remains to be determined whether or not LPS induces a shift in the number of LF receptors on the THP-1 cells.

The presence of FCS increased the LPS-induced cytokine response in THP-1 cells by at least 10 times. This would imply that the serum component LPS-binding protein and CD14 may at least be partly involved in transducing an activation signal for IL-6 production. LPS complexed with LPS-binding protein binds to CD14 present on THP-1 cells(31). This would result in an enhanced cytokine response of the cells(38). Whether LF could interfere with such a complex formation and thereby prevent LPS from binding to CD14 needs to be tested. LPS incubated with LF in excess did not, however, inhibit the Limulus activity of LPS, indicating that the toxic part of LPS, lipid A, is not affected by LF (our unpublished results). Priming by IFN-γ, which increased the LPS-induced IL-6 response, has been reported not to change the CD14 expression in THP-1 cells(31). Thus, the enhanced LPS response in IFN-γ-primed THP-1 cells may be largely CD14-independent.

Human LF, which is a 80-kD glycoprotein, exists in three isoforms possibly with partly different functions. Two isoforms fail to bind iron but express RNAse activity(39). LF is a multipotent molecule expressing bactericidal and immunomodulating/suppressing activities(36). LFcin, a fragment of the LF near the N terminus of the molecule, distinct from the iron-binding sites, has been shown to constitute the bactericidal domain(6). This fragment also seems to be a part of the binding domain of LF for the lymphocyte receptor(40). Despite chemical and antigenic differences between human and bovine LF or LFcin, suppressive effects were demonstrated by both bovine LF and its fragment on the IL-6 responses in human cells. In fact, the bovine LFcin was even better than human LF in suppressing the IL-6 response.

The capacity of human milk LF to inhibit LPS-induced cytokine release may play a role for the breast-fed neonate being colonized in the gut by LPS-containing and releasing Gram-negative bacteria. It is possible that the milk LF or LFcin might dampen cytokine release from cells in the gut wall caused by such free LPS. This might be one explanation why breast-fed infants have a significantly smaller physiologic weight loss during the 1st wk of life than the non-breast-fed(41).

In addition to high concentrations of LF, human milk contains IL-6(42, 43). This fact may not be inconsistent with our results, because there are several cell types that have the capacity to produce IL-6, such as epithelial cells including human mammary gland epithelial cells(4446). IL-6 producing cells other than monocytes/macrophages may be affected to a lesser extent or not at all by LF. The major source of IL-6 in human milk is not known.

In conclusion, our studies showed a suppressive effect of LF or LFcin on the LPS-induced cytokine response in human monocytic cells, either added before or after the LPS. In addition, the suppressive effect of human LF on the IL-6 production was confirmed by using TNF as another IL-6-inducing agent.