Soluble CD14 and CD83 from Human Neonatal Antigen-Presenting Cells Are Inducible by Commensal Bacteria and Suppress Allergen-Induced Human Neonatal Th2 Differentiation

Abstract

CD14 is expressed on the cell surface of various antigen-presenting cells, and CD83 is a maturation marker for dendritic cells (DC). CD14 and CD83 are also present as soluble proteins, and both have immunoregulatory functions. We examined whether neonatal cord blood monocytes or DC released soluble CD14 (sCD14) or sCD83 when exposed to the commensal intestinal bacteria Clostridium perfringens, Staphylococcus aureus, Lactobacillus rhamnosus, Escherichia coli, and Bacteroides fragilis. We found that the gram-positive bacteria C. perfringens and S. aureus, but not gram-negative bacteria, induced the release of sCD14 from monocytes. DC, on the other hand, released sCD14 in response to both gram-positive and gram-negative bacteria. Moreover, the expression of the virulence factor staphylococcal protein A seemed to be important for S. aureus-induced sCD14 production from both monocytes and DC. Soluble CD83 was released from DC, but not from monocytes, when exposed to both gram-positive and gram-negative bacteria. Finally, to investigate whether sCD14 or sCD83 could modulate neonatal allergen-induced T-cell differentiation, DC were exposed to birch allergen alone or in the presence of sCD14 or sCD83 and then cocultured with autologous T cells. We demonstrate that sCD14 and sCD83 inhibited the birch allergen-induced Th2 differentiation by suppressing interleukin 13 production. Together, these results suggest that the commensal intestinal flora may be an important stimulus for the developing immune system by inducing the immunoregulatory proteins sCD14 and sCD83, which may be involved in preventing T-cell sensitization to allergens in infants.

Allergic diseases are characterized by a Th2-type immune response, resulting in immunoglobulin E (IgE) production against innocuous environmental allergens. It has been suggested that a lack of exposure to a wide variety of microbes during early childhood may result in a higher prevalence of allergic diseases in the Western world than in developing countries (12, 28). This altered microbial exposure affects the normal gastrointestinal microflora, which may be important for appropriate development of the immune system (33, 35). Although the immunological mechanisms behind the protection against allergy are not clear, innate immune cells are most likely involved, as the first interaction between bacteria and the immune system occurs via these cells.

Bacterial structures are recognized by pattern recognition receptors, such as CD14 and Toll-like receptors (TLRs) (1), which thereby represent a link between the innate and adaptive immune systems. Traditionally, CD14 has been a hallmark of monocytes and macrophages, as most subpopulations of these cells express CD14 (42). In blood, however, dendritic cell (DC) preparations contain several phenotypically and functionally distinct subpopulations, of which some express low levels of CD14 (31). CD14 plays a key role in initiating cell activation by a group of bacterially derived structures, such as lipopolysaccharide (LPS) from gram-negative bacteria and peptidoglycan from gram-positive and gram-negative bacteria (15, 39). A soluble form of CD14 (sCD14) is present in large amounts in human serum and in breast milk (14, 26) and may be an acute-phase protein with the function of protecting against LPS-induced shock (7, 21). In addition, sCD14 has immunoregulatory functions, as it interacts with activated human T and B cells, leading to inhibition of interleukin 4 (IL-4), gamma interferon (IFN-γ), and IgE production, respectively (5, 34). Interestingly, it was recently shown that the circulating levels of sCD14 were reduced in 7-year-old atopic children compared with nonatopic children (41), and house dust mite-sensitized children had significantly lower levels of sCD14 in the circulation than those not sensitized to house dust mites (36). Moreover, we have recently shown that children who harbor Staphylococcus aureus in the gut early in life have higher levels of sCD14 in the circulation than do noncolonized children and that children who developed food allergies tended to have lower levels of sCD14 in plasma (30).

Human CD83 is a 45-kDa glycoprotein and a member of the immunoglobulin superfamily, which has been widely used as one of the best surface markers for activated DC (6). Soluble CD83 (sCD83) is released from activated in vitro-cultured human DC and is detectable in small amounts in the circulation of healthy infants and adult individuals (19, 30). Soluble CD83 has been shown to regulate immune responses by inhibiting DC-T-cell clustering and DC-mediated T-cell expansion (25, 29). Moreover, administration of sCD83 has been found to ameliorate Th1-driven experimental autoimmune encephalomyelitis in mice (43). How sCD83 is generated is not clear, but proteolytic shedding of cell surface-associated CD83 or alternative splicing has been proposed as a possible mechanism (13, 19).

It is unknown whether intestinal commensal bacteria are able to induce the release of sCD14 and sCD83 from neonatal innate immune cells. Moreover, it also remains to be elucidated whether gram-positive and gram-negative bacterial species differ in the ability to stimulate the release of these two proteins. Therefore, we examined whether gram-positive commensal bacteria, including Clostridium perfringens, Staphylococcus aureus, and Lactobacillus rhamnosus, or the gram-negative commensal bacteria Escherichia coli and Bacteroides fragilis were able to induce the release of sCD14 or sCD83 from neonatal blood monocytes or DC. We also analyzed the possible impact of the virulence factor staphylococcal protein A on S. aureus-induced production of sCD14 by neonatal innate immune cells. Moreover, as the role of sCD14 or sCD83 in regulating human Th2 immune responses to allergens is unclear, we investigated whether these two proteins could modulate birch allergen-induced T-cell differentiation in an in vitro model of allergic sensitization. Autologous DC and T cells from cord blood were used, as we have previously shown that birch allergen extract induces a Th2 profile in this system (4).

MATERIALS AND METHODS

Cell separation.

Umbilical cord blood, collected in heparin-containing tubes, was obtained from unselected, healthy, vaginally delivered babies (atopic status of the parents unknown) born at the Sahlgrenska University Hospital (Göteborg, Sweden). The study was approved by the Human Research Ethics Committee of the Medical Faculty, Göteborg University, Göteborg, Sweden. Mononuclear cells were isolated by density gradient centrifugation over Ficoll-Hypaque (Pharmacia Biotech, Uppsala, Sweden). Monocytes were isolated by a positive selection technique, according to the manufacturer's instructions (MACS system; Miltenyi Biotech, Bergisch Gladbach, Germany). For isolation of blood DC, a DC isolation kit (BDCA-1) was used (Miltenyi Biotech) and T cells were purified with the Dynabead CD4 Positive Isolation kit (Dynal Biotech ASA, Oslo, Norway), according to the manufacturer's instructions. Isolated T cells were gradually cooled to −70°C. After 24 h, the T cells were transferred to a −143°C freezer.

Bacterial strains.

Bacterial strains were obtained from the commensal intestinal flora of healthy Swedish infants. Fecal samples were cultured quantitatively for all major groups of aerobic and anaerobic bacteria and speciated as previously described (2). In brief, staphylococci were isolated from Staphylococcus agar and divided into coagulase-positive (S. aureus) and coagulase-negative staphylococci by the coagulase test. Enterobacteria were isolated on Drigalski agar and speciated with the API20E biotyping system (API Systems SA, La Balme les Grottes, France). All isolates obtained from anaerobic cultures were first checked for the inability to grow under aerobic conditions. Bacteroides spp. were isolated from Bacteroides bile esculin agar and speciated by Rapid ID 32A (API Systems). Straight gram-positive rods isolated on Rogosa agar were defined as lactobacilli, which was confirmed by PCR using group- and species-specific primers (3). Clostridia, defined as straight gram-positive or gram-labile rods with or without spores, were speciated with Rapid ID32A. We also used S. aureus strain Newman and a previously described mutant (DU5873; staphylococcal protein A-deficient (SpA), derived from Newman), kindly provided by T. Foster, Department of Microbiology, Trinity College, Dublin, Ireland (32). All bacterial strains used were washed in phosphate-buffered saline (PBS) (1,000 × g, 10 min), counted in a microscope, inactivated by exposure to UV light for 20 min (inactivation was confirmed by a negative viable count), and finally stored at −70°C.

Bacterial stimulation of cord blood monocytes and cord blood DC.

Monocytes or DC (1 × 10⁶/ml) were stimulated with UV-killed commensal bacteria, including Clostridium perfringens, Staphylococcus aureus, Lactobacillus rhamnosus, Escherichia coli, or Bacteroides fragilis (1 × 10⁷ bacteria/ml), under serum-free conditions for 24 or 48 h at 37°C in 5% CO₂. Monocytes or DC (1 × 10⁶/ml) were also stimulated with S. aureus strain Newman or the SpA mutant strain DU5873 (1 × 10⁶/ml) under serum-free conditions for 48 h. Phenotypic analysis of DC stimulated with bacteria was performed by flow cytometry. The cells were suspended in PBS containing 1% fetal calf serum, 0.1% sodium azide, and 0.5 mM EDTA (fluorescence-activated cell sorter [FACS] buffer), placed in 96-well V-bottom plates, and pelleted by centrifugation (3 min at 300 × g, 4°). All of the monoclonal antibodies (MAbs) used were diluted in FACS buffer at optimal concentrations. The following MAbs were used: APC-anti-CD11c (B-ly6), PE-anti-CD14 (M5E2), or PE-anti-CD83 (HB15e) (BD-Bioscience, Stockholm, Sweden). As isotype controls, mouse monoclonal IgG1 antibodies were used (BD Bioscience). The cells were incubated with the respective MAbs for a minimum of 15 min at 4°C in the dark, followed by two washing steps and a final resuspension step in FACS buffer before analysis. We analyzed approximately 1,000 to 5,000 cells in a FACSCalibur (BD Bioscience) equipped with CellQuest software (BD Bioscience). Phenotypic analysis of stimulated monocytes was not performed, as this cell population adheres strongly to plastic during culture. The release of sCD14 by monocytes and DC was higher after 48 h of stimulation than after 24 h (see Fig. S1A and B in the supplemental material). Soluble CD83 was released only by DC, and not by monocytes, and could be measured only after 48 h of stimulation (see Fig. S2A and B in the supplemental material). The expression of CD14 on the cell surface was higher after 48 h of stimulation than after 24 h (see Fig. S3A in the supplemental material). The cell surface expression of CD83, on the other hand, was higher after 24 h of stimulation than after 48 h (see Fig. S3B in the supplemental material).

Reagents.

Birch allergen (Betula verrucosa) extract was provided by ALK-Abelló (Hørsholm, Denmark). The proportion of the protein in the allergen extract was 67.3%, and our batch of allergen extract added 5 pg of LPS per 200 μl of culture medium. Recombinant human CD14 was purchased from R&D Systems (Minneapolis, MN) and added 2 or 20 pg of LPS per 200 μl of culture medium. Recombinant human sCD83 added less than 115 pg of LPS per 200 μl of culture medium. Human serum albumin (HSA) was obtained from Octapharma (Stockholm, Sweden). The endotoxin content was assessed by a chromogenic Limulus amoebocyte lysate endpoint test (Chomogenix AB, Mölndal, Sweden).

Autologous DC-T-cell cocultures.

For generation of immature monocyte-derived DC (MD-DC), monocytes (10⁶ cells/ml) were cultured in RPMI 1640 medium (BioWhittaker, Cambrex Company, Belgium) supplemented with 2% autologous serum, 1 mM l-glutamine, 50 μg/ml gentamicin, 500 U/ml recombinant IL-4 (R&D Systems), and 800 U/ml recombinant granulocyte-macrophage colony-stimulating factor (GM-CSF) (R&D Systems). The cells were cultured for 6 to 7 days and were refed with IL-4 and GM-CSF containing medium every second day. MD-DC used for DC-T-cell cocultures were either frozen or stimulated with birch allergen alone (50 μg/ml) or birch allergen in combination with sCD14 (0.1 or 1 μg/ml), sCD83 (10 ng/ml), or HSA (0.1 μg/ml, 1 μg/ml, or 10 ng/ml) in the presence of tumor necrosis factor (TNF) (20 ng/ml) (R&D Systems), IL-1β (10 ng/ml) (R&D Systems), and prostaglandin E₂ (1 μg/ml) (Sigma-Aldrich, St. Louis, MO) under serum-free conditions. After 24 h of stimulation, MD-DC were washed and cocultured (10 × 10³ to 15 × 10³ cells/well) with autologous naïve CD4⁺ T cells (1.5 × 10⁵ cells/well) in serum-free X-Vivo15 medium (BioWhittaker). After 3 days of culture, 100 μl of the supernatant was replaced with fresh medium containing IL-2 (50 U/ml) (Nordic BioSite, Täby, Sweden). On day 6, frozen autologous MD-DC were thawed and stimulated as described above and thereafter used on day 7 for restimulation of the T cells. After 10 days of culture, supernatants were collected for analysis of secreted cytokines by enzyme-linked immunosorbent assay (ELISA). For analysis of intracellular expression of cytokines, T cells were stimulated with phorbol myristate acetate (10 ng/ml) and ionomycin (1 μg/ml) for 4 h, and GolgiPlug was added for the last 3 h. Cells were fixed with paraformaldehyde (2%) and permeabilized with saponin (0.5%), and intracellular cytokines were detected by flow cytometry using fluorescein isothiocyanate-anti-IFN-γ (B27) and PE-anti-IL-13 (JES10-5A2) MAbs (Becton-Dickinson, Erembodegum, Belgium). All reagents were purchased from Sigma-Aldrich except for GolgiPlug, which was obtained from Pharmingen.

ELISAs.

Concentrations of IL-5, IL-13, IFN-γ, or sCD14 in cell culture supernatants were determined by a standard ELISA procedure, as described in detail elsewhere (23). Costar plates (Invitrogen, San Diego, CA) were coated with the following capture MAbs: anti-IL-5 (TRFK5), anti-IL-13 (JES10-5A2), anti-IFN-γ (NIB42), or anti-sCD14 (55-3). Standard curves were generated with recombinant human IL-5, IL-13, IFN-γ, or CD14, respectively. All antibodies and standards were purchased from Pharmingen except for recombinant CD14, which was obtained from R&D Systems. The following biotinylated detection antibodies were used: anti-IL-5 (JES1-5A10), anti-IL-13 (B69-2), anti-IFN-γ (4S.B3), or anti-sCD14 (3-C39). Samples, standards, biotinylated antibodies, and streptavidin-horseradish peroxidase were diluted in the high-performance ELISA buffer Sanquin (Amsterdam, The Netherlands). In control experiments, we found that preincubation of sCD14 with various doses of LPS from E. coli (Sigma-Aldrich) or peptidoglycan from S. aureus (Sigma-Aldrich) did not interfere with the detection of sCD14 in the ELISAs (see Fig. S4 in the supplemental material).

Concentrations of sCD83 were determined with a modification of a previously described ELISA (18, 19). Costar plates (Invitrogen, San Diego, CA) were coated with monoclonal anti-CD83 (clone HB15a;Immunotech, Marseille, France). The isotype-matched CD69 control MAb (Immunotech) was used to provide a measure of the nonspecific background for each individual sample. The capture antibodies CD83 and CD69 were diluted in PBS, and the plates were thereafter blocked with 10% goat serum (Gibco-BRL, Life Technologies, New Zealand). Standard curves were generated with recombinant sCD83 (CD83-GST) (29). For detection, polyclonal rabbit anti-CD83 (RA83; kindly provided by B. Hock, Christchurch Hospital, New Zealand) was diluted in 5% goat serum, 2% mouse serum, and 1% dried nonfat milk in PBS to a concentration of 10 μg/ml (18, 19, 29). Thereafter, biotinylated monoclonal mouse anti-rabbit antibodies (RG-96;Sigma-Aldrich), diluted in reagent buffer, were added to the plates. Next, the plates were incubated with streptavidin-horseradish peroxidase (Sanquin, The Netherlands) diluted in PBS containing 0.5% bovine serum albumin. Then, 3,3′5,5′-tetramethylbenzidine (Dako, Carpinteria, CA) substrate was added to the plates, which were kept in the dark, and the reaction was stopped by the addition of 2.5 M H₂SO₄.

Statistical analysis.

The data were analyzed by the Kruskal-Wallis test or the Friedman test, followed by Dunn's multiple-comparison test or the Wilcoxon matched-pairs test, as described in the figure legends (GraphPad Prism, San Diego, CA).

RESULTS

Production of sCD14 and expression of CD14 by neonatal innate immune cells in response to commensal bacteria.

As the physiological stimulus that may trigger the production of sCD14 and sCD83 is unclear, we investigated whether human neonatal innate immune cells would release sCD14 or sCD83 in response to stimulation with commensal intestinal bacteria isolated from Swedish infants. Cord blood monocytes or DC were exposed to the UV-killed commensal gram-positive bacterial strain C. perfringens, S. aureus, or L. rhamnosus or the gram-negative bacterial strain E. coli or B. fragilis.

We found that cord blood monocytes released significantly larger amounts of sCD14 after stimulation with the gram-positive C. perfringens or S. aureus but not with the gram-negative bacteria, compared to unstimulated cells (Fig. 1A). Freshly isolated cord blood DC, on the other hand, released sCD14 in response to both the gram-positive C. perfringens or S. aureus and to the gram-negative E. coli or B. fragilis, as shown in Fig. 1B. In Fig. 1C, we show that CD14 was upregulated on the cell surface of blood DC when exposed to both gram-negative and gram-positive bacteria, but CD14 was not detected intracellularly either in cells exposed to bacteria or in unstimulated cells (see Fig. S5 in the supplemental material). In Fig. 1D, we show representative dot plots regarding CD14 expression on the cell surface of unstimulated cord blood DC and on cells stimulated with S. aureus or E. coli for 48 h.

FIG. 1.

Release of sCD14 and expression of CD14 on the cell surface in response to bacterial stimulation. Shown is the release of sCD14 by cord blood monocytes (A) or cord blood DC (B) in response to the UV-killed gram-positive bacterium C. perfringens, S. aureus, or L. rhamnosus or the gram-negative bacterium E. coli or B. fragilis for 48 h. (C) Expression of CD14 on the cell surface by DC after exposure to the bacteria listed above for 48 h. Each symbol represents one individual (panel A, n = 8; panel B, n = 5 to 10; panel C, n = 3 to 6), and the horizontal bars represent the median. (D) Dot plots represent CD14 expression on unstimulated cord blood DC and on cells stimulated with S. aureus or E. coli for 48 h. *, P < 0.05; **, P < 0.01; ***, P < 0.001 (Kruskal-Wallis followed by Dunn's multiple comparison test).

To examine the possible impact of staphylococcal protein A (SpA) on S. aureus-induced sCD14 release from neonatal monocytes and DC, the cells were stimulated with a wild-type S. aureus strain or an SpA-deficient S. aureus strain (isogenic strains with respect to SpA). In Fig. 2, we show that the SpA-deficient S. aureus strain triggered significantly lower levels of sCD14 from both cord blood monocytes and DC than did the wild-type strain. Taken together, freshly isolated cord blood monocytes appear to release sCD14 only when exposed to gram-positive bacteria, whereas the sCD14 release from DC and the CD14 expression on the DC cell surface was upregulated in response to both gram-positive and gram-negative bacteria. Furthermore, SpA seems to be one factor that influences sCD14 production by neonatal innate immune cells after S. aureus stimulation.

FIG. 2.

S. aureus-induced release of sCD14 is influenced by staphylococcal protein A. Shown are levels of sCD14 released by cord blood monocytes (A) or DC (B) in response to a wild-type S. aureus strain or a staphylococcal protein A mutant after stimulation for 48 h. Each symbol represents one individual (panel A, n = 6; panel B, n = 8), and the horizontal bars represent the median. *, P < 0.05 (Wilcoxon matched-pairs test).

Production of sCD83 and expression of CD83 by neonatal cells in response to bacteria.

When we examined the ability of innate immune cells to release sCD83 in response to commensal bacteria, we found that the blood DC released sCD83 in response to all gram-positive and gram-negative bacterial species we included in the experiments (Fig. 3A), whereas CD83 expression on the cell surface seemed to be upregulated only in response to stimulation with the gram-positive C. perfringens, S. aureus, or L. rhamnosus (Fig. 3B). In Fig. 3C, we show representative dot plots regarding CD83 expression on the cell surface of unstimulated cord blood DC and on cells stimulated with S. aureus or E. coli for 48 h. CD83 was detected intracellularly in the majority of the blood DC, but there were no differences between stimulated and unstimulated cells (Fig. 3D). In contrast to sCD14, monocytes did not release sCD83 when exposed to bacteria (see Fig. S2B in the supplemental material). In summary, cord blood DC released sCD83 when exposed to both gram-positive and gram-negative bacteria, whereas the cell surface expression of CD83 was upregulated only in response to the gram-positive bacteria.

FIG. 3.

Release of sCD83 and expression of CD83 on the cell surface in response to bacterial stimulation. (A and B) Release of sCD83 and expression of CD83 on the cell surface by blood DC in response to the UV-killed gram-positive bacterium C. perfringens, S. aureus, or L. rhamnosus or the gram-negative bacterium E. coli or B. fragilis for 48 h. Each symbol represents one individual (panel A, n = 5 or 6; panel B, n = 3 to 6), and the horizontal bars represent the median. (C) Dot plots represent CD83 expression on unstimulated cord blood DC and on cells stimulated with S. aureus or E. coli for 48 h. (D) Intracellular expression of CD83 by blood DC in response to the UV-killed gram-positive bacterium C. perfringens, S. aureus, or L. rhamnosus or the gram-negative bacterium E. coli or B. fragilis for 48 h. Each symbol represents one individual (n = 3), and the horizontal bars represent the median. *, P < 0.05; **, P < 0.01 (Kruskal-Wallis followed by Dunn's multiple comparison test).

Soluble CD14 and sCD83 suppress human neonatal birch allergen-induced IL-13.

To investigate whether sCD14 or sCD83 might modulate neonatal birch allergen-induced T-cell responses, DC were stimulated with birch allergen extract alone or in the presence of sCD14, sCD83, or the control protein HSA and were thereafter cocultured with autologous CD4⁺ T cells. DC were stimulated in the presence of TNF, IL-1β, and prostaglandin E₂, as we have previously shown that DC need to be fully mature to be able to induce birch allergen-induced T-cell expansion and differentiation (4). In order to use physiological doses of sCD14 in our culture systems, the selected concentrations (0.1 or 1 μg/ml) were based on the median levels that we have recently found to be present in the circulation during infancy (cord, 0.18 μg/ml; 4 months, 0.57 μg/ml; 18 months, 0.97 μg/ml; 3 years, 0.55 μg/ml) (30). Regarding sCD83, we used a higher concentration (10 ng/ml) than the median levels that were present in the circulation during infancy (cord, 0.1 ng/ml; 4 months, 0.14 ng/ml; 18 months, 0.1 ng/ml; 3 years, 0.1 ng/ml) (30), as it has been shown that sCD83 is elevated at the local site of inflammation but not in the circulation (20).

In Fig. 4A, we show that birch alone induced significantly higher levels of the Th2 cytokine IL-13 than did unstimulated cells. Birch in combination with the higher dose of sCD14 (1 μg/ml) or sCD83 (10 ng/ml) induced significantly lower levels of IL-13 relative to birch alone, whereas the levels of IL-13 induced by birch together with the lower dose of sCD14 (0.1 μg/ml) did not differ significantly from those induced by birch allergen alone. Soluble CD14 or CD83 alone did not affect the production of IL-13 relative to unstimulated cells. Neither sCD14 nor sCD83 appeared to affect the viability of the cells as judged by staining with trypan blue. Birch in combination with the control protein HSA did not affect IL-13 production compared to birch alone (Fig. 4B). The production of IL-5 and IFN-γ was also measured in the cell cultures, but only very low levels of IL-5 and no IFN-γ were induced, and increased levels were not seen after stimulation (see Fig. S6A and B in the supplemental material), as previously shown with neonatal T cells (4, 27).

FIG. 4.

Soluble CD14 and sCD83 suppress birch allergen-induced IL-13. DC were stimulated with birch, birch plus sCD14, birch plus sCD83, or birch plus HSA and cocultured with autologous CD4⁺ T cells. (A) On day 10, production of IL-13 was analyzed by ELISA, and boxes and whiskers summarize data from 11 subjects. The boxes show the median and the interquartile range, and the whiskers show the range. (B) IL-13 production induced by birch allergen compared to birch plus HSA (n = 7). (C) Intracellular expression of IL-13 in T cells after 10 days of coculture with DC stimulated with birch, birch plus sCD14, birch plus HSA, or sCD14 alone. Data demonstrate one experiment out of four. (D) Intracellular expression of IL-13 in T cells after 10 days of coculture with DC stimulated with birch, birch plus sCD83, birch plus HSA, or sCD83 alone. Data demonstrate one experiment out of two. The intracellular IL-13 expression was analyzed by flow cytometry, and the percentages of positive cells are indicated in the upper right quadrant. *, P < 0.05; **, P < 0.01 (Friedman test followed by Dunn's multiple comparison test).

To support our observations regarding the secretion of IL-13 into the cell culture supernatant, we also examined the intracellular IL-13 expression in CD3-positive T cells in response to DC stimulated with birch alone or birch in combination with sCD14 or sCD83 in a limited number of experiments (see Table S1 in the supplemental material). In Fig. 4C, we show data from one experiment after stimulation with birch relative to unstimulated cells and the suppressive effect of adding 1 μg/ml of sCD14 to the cultures. In Fig. 4D, we show the corresponding data for an experiment using sCD83. Taken together, stimulation of DC with birch in combination with sCD14 or sCD83 resulted in suppression of IL-13 production, manifested by reduced levels in the supernatant, a finding which was supported by lower numbers of T cells harboring IL-13 intracellularly in the majority of the experiments.

DISCUSSION

Many epidemiological studies indicate that exposure to high levels of microbes early in life may induce tolerance to environmental allergens and thereby prevent allergy development (11, 12). The immunological mechanisms behind this phenomenon are not clear, but stimulation of the developing immune system may be an important factor. Here we demonstrate that gram-positive, but not gram-negative, commensal intestinal bacteria triggered cord blood monocytes to release sCD14. Neonatal DC, on the other hand, released sCD14 and sCD83 when exposed to both gram-positive and gram-negative bacteria. Moreover, both sCD14 and sCD83 suppressed human neonatal birch allergen-induced Th2 differentiation in an autologous in vitro model for allergen sensitization. Thus, we show for the first time how the production of sCD14 and sCD83 is differentially stimulated by gut bacteria and that both proteins modulate neonatal allergen-induced T-cell responses.

We have recently found that children with food allergies are less frequently colonized with S. aureus in the gut at 2 weeks of age, and they also tend to have lower levels of sCD14 in plasma than do healthy children (30). No other groups have studied the effects of early intestinal bacterial colonization in relation to the development of food allergies. However, one study has shown that S. aureus is more frequent in the gut of 6-month-old children who develop atopic dermatitis than in children who remain healthy (9). Interestingly, in this study no difference in S. aureus colonization between the groups was seen at 1 week or 1 month of age. This is supported by data from our study showing no relation between early S. aureus colonization and the development of eczema, in contrast to the development of food allergies (30). One possible explanation for the increased frequency of intestinal S. aureus colonization at 6 months of age could be that the children had at that time developed S. aureus-infected atopic dermatitis, which could in turn lead to increased S. aureus colonization of the gut.

In our previous study, we also demonstrated that children who are colonized with S. aureus early in life have higher levels of sCD14 in the circulation than do noncolonized children, whereas early colonization with E. coli had no effect (30). However, a cause-effect relationship between S. aureus colonization and increased sCD14 levels in the circulation has not been proven in vivo. Here we demonstrate that neonatal monocytes and DC released sCD14 in response to the gram-positive bacteria C. perfringens and S. aureus. In contrast to monocytes, cord blood DC released sCD14 in response to both gram-positive and gram-negative gut bacteria. Thus, C. perfringens and S. aureus, but not E. coli or B. fragilis, seem to have characteristics that trigger both monocytes and DC to release sCD14. This might be explained by the fact that monocytes and DC express different repertoires of pattern recognition receptors (22, 24, 38). Monocytes express CD14 and TLR2 with high intensity and TLR4 with intermediate intensity on the cell surface relative to DC, which express very low levels of these three receptors (24). Moreover, it has been shown that TLR2 recognizes structures expressed only by gram-positive bacteria, such as lipoteichoic acid (37), which is in accordance with our results demonstrating that monocytes release sCD14 only in response to gram-positive bacterial stimulation. Furthermore, our group has previously shown that the pattern of proinflammatory cytokine responses to gut bacteria is changed when monocytes differentiate to DC (24). Monocytes produced higher levels of IL-12p70 and TNF after stimulation with the gram-positive bacterium L. plantarum relative the levels induced by the gram-negative bacterium E. coli, whereas DC secreted large amounts of IL-12p70 and TNF in response to E. coli (24). In line with these results, it has been shown that mucosal explants from the human gut produce TNF in response to stimulation with E. coli but not when exposed to lactobacilli (10). Taken together, not only cytokine profiles but also production of other immunoregulatory proteins, such as sCD14, are changed when monocytes differentiate to DC.

The ability of the gram-positive bacteria C. perfringens and S. aureus to induce high levels of sCD14 might be related to their ability to elicit TNF production. Our group and others have shown that gram-positive bacteria induce higher levels of TNF from cord blood mononuclear cells, peripheral blood mononuclear cells, and purified monocytes than do gram-negative bacteria (17, 23, 24), whereas there were no differences in IL-6 production (23). Indeed, the most potent gram-positive bacterium to induce TNF production by monocytes was found to be S. aureus (17). In the present study, we demonstrate that the S. aureus-induced release of sCD14 from neonatal blood monocytes and DC was influenced by the expression of SpA on the bacterial cell surface. One possible explanation for this finding could be that SpA has been shown to bind to the TNF receptor and may thereby directly trigger the release of sCD14 (16). However, it remains to be elucidated why C. perfringens, which is a poor inducer of TNF, was found to be the most potent inducer of sCD14 by neonatal monocytes and DC (17).

We demonstrate that gram-positive and gram-negative commensal gut bacteria induce release of sCD83 from cord blood DC but not from monocytes. Our results are consistent with a previous in vitro study showing that human monocyte-derived DC from adult individuals release sCD83 when stimulated with LPS (19). Previously, we found that sCD83 is present in small amounts in the circulation of human infants compared to the levels of sCD14 (30). Interestingly, clinical studies have reported that the levels of soluble CD14 are elevated in both plasma and synovial fluid from patients with rheumatoid arthritis (8, 40), but the levels of sCD83 are increased only in the synovial fluid, not in the circulation (20). Thus, the amount of sCD14 appears to be increased both locally and systemically during immune activation, whereas elevated levels of sCD83 may be found only at the local site of inflammation. This may explain why we found no relation between the levels of sCD83 in the circulation and early intestinal colonization in spite of the fact that the bacteria were able to induce release of sCD83 from DC in vitro (30). Therefore, sCD83 may well be elevated locally in the intestinal mucosa in response to the commensal microflora.

Soluble CD14 appears to have several immunological functions. In addition to its role as a receptor for bacterial structures, including LPS and peptidoglycan, sCD14 modulates cellular immune responses by suppressing T-cell proliferation and production of IFN-γ and IL-4 (34). Accordingly, we found that sCD14 suppressed birch allergen-induced IL-13 production and decreased the number of IL-13 expressing T cells. One explanation for the observed regulatory function of sCD14 on T-cell activation could be via an inhibitory effect on IL-2 production (34). As for sCD14, we found that sCD83 inhibited birch allergen-induced Th2 differentiation in our in vitro model for allergic sensitization. Our observation is in line with those of others, as it was recently shown that sCD83 has immunoregulatory functions by inhibiting disease symptoms in a murine model of experimental autoimmune encephalomyelitis in vivo, which is dominated by a Th1-type response (43). Further, sCD83 inhibits human DC-mediated allogeneic T-cell proliferation in vitro in a dose-dependent manner (29). One possible explanation for the effect of sCD83 on T-cell differentiation could be by affecting the activation of DC. Indeed, sCD83 has been found to influence the arrangement of the cytoskeleton of DC and thereby inhibiting DC-T-cell clustering, a prerequisite for DC-mediated T-cell expansion (25). Thus, our results, in combination with those of others, show that sCD14 and sCD83 are able to suppress both Th2 and Th1 responses, suggesting that these soluble proteins may have an important function in maintaining immune homeostasis.

In summary, we show that neonatal monocytes release sCD14 and neonatal DC release sCD14 and sCD83 when stimulated with commensal gut bacteria. Moreover, both proteins suppressed birch allergen-induced Th2 differentiation in an in vitro model for allergen sensitization by inhibiting the production and expression of IL-13, an important switch factor for IgE. Whether the present study may help to explain the immunological mechanisms at the cellular level behind recent epidemiological studies needs to be further elucidated. However, we suggest that sCD14 and sCD83 may be mediators that are induced by the intestinal microflora and that may be involved in down-regulating immune responses, leading to allergic diseases in children.