| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* Infection and Immunity, Department of Medical Biochemistry and Immunology, Cardiff University, College of Medicine, Cardiff, United Kingdom;
School of Applied Sciences, University of Wales Institute, Cardiff, United Kingdom;
Department of Bioanalytical Science, Nestlé Research Center, Lausanne, Switzerland;
Department of Nutrition and Health, Nestlé Research Center, Lausanne, Switzerland; and
¶ Newborn Immunity, School of Medicine, University of Wales Swansea, United Kingdom
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
80 kDa was found to be responsible for the effect on TLR4. Notably, infant milk formulations did not reproduce the modulatory activity of breast milk. Together, these findings reveal an unrecognized function of human milk, namely, its capacity to influence neonatal microbial recognition by modulating TLR-mediated responses specifically and differentially. This in turn suggests the existence of novel mechanisms regulating TLR activation. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
Notably, a lower incidence of gastrointestinal infections and inflammatory conditions as well as allergic diseases in breast-fed newborns has long been reported (5, 6, 7, 8, 9, 10, 11, 12, 13). Over the years, protection by breast-feeding has been variously ascribed to a number of mechanisms, and to milk components such as maternal immunocompetent cells, Igs, antimicrobial peptides, oligosaccharides, growth factors, cytokines, lysozyme, lactoferrin, complement, and nutrients (8, 14). However, we have sought additional and specific mechanisms related to the recognition of microbes in the gut, because microbial recognition appears as a common factor directly or indirectly involved in the pathological conditions that are claimed to be reduced by breastfeeding. Such mechanisms remain to be fully elucidated.
The activity of the mammalian TLR family is crucial to an immediate and efficient innate recognition of an array of microorganisms and their cell-wall components (15). Thirteen mammalian TLRs have so far been identified (TLR1–TLR13) and the ligand specificity described for most of them (16, 17, 18). Efficient microbial recognition by TLR2 and TLR4 requires, in most cases, the activity of a coreceptor, CD14, which is expressed as a cell surface molecule (membrane-bound CD14; mCD14)3 mainly in myeloid cells, and also in serum as a soluble (s) coreceptor (sCD14). Through its coreceptor activity, CD14 enhances cellular responses to most microbial components activating via TLR2 or TLR4 (19, 20, 21, 22, 23, 24).
TLRs are expressed preferentially in tissues that are in constant contact with microorganisms, such as the lung and the gastrointestinal tract, as well as in immunologically important cell types such as blood leukocytes and dendritic cells (DC) (25, 26, 27). The latter are also distributed along the intestinal epithelium and can sample pathogens directly into the gut lumen by extending dendrites outside the epithelium (28). Recent work highlighted the crucial role that TLRs play in the gut by controlling intestinal epithelial homeostasis and protecting it from direct injury (4, 29, 30).
TLR engagement triggers the recruitment and activation of a number of signal adaptors and intermediates. This leads to transcription factor activation and the induction of genes that code for a variety of proinflammatory and immunoregulatory cytokines, chemokines, and costimulatory molecules (31). This process results in a prompt and efficient response to the microbial challenge. However, the excessive release of some of these proinflammatory molecules may lead to serious inflammatory conditions, which may result in tissue damage and septic shock. Therefore, TLR triggering has to be tightly regulated (32, 33). This is of particular importance during bacterial colonization of the newborn’s gut, because only a finely controlled recognition of microorganisms will result in the maintenance of neonatal gut homeostasis.
We have therefore sought TLR-specific mechanisms operating in milk that may modulate microbial recognition in the gut, thereby contributing to the beneficial effects of breast-feeding. We recently reported the presence in human milk, but not in infant milk formulations, of high concentrations of the coreceptor sCD14 (34). We have proposed that milk sCD14 may enable efficient TLR activation during bacterial colonization of the neonatal gut, and suggested that milk sCD14-mediated TLR triggering might be tightly regulated (34). We have therefore looked for regulatory mechanisms. More recently, we described the existence of soluble forms of TLR2 (sTLR2) capable of regulating cell activation via cell surface TLR2, and naturally expressed in human milk as well as in plasma (35). Interestingly, sTLR2, like sCD14, was not detected in infant milk formulations. These findings prompted us to test in this study whether additional regulatory mechanisms operating in milk are involved in modulating TLR-mediated microbial recognition.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
Ultra-pure LPS (Escherichia coli O111:B4 strain), bacterial peptidoglycan, heat-killed Listeria monocytogenes, bacterial flagellin, and polyinosinic-polycytidylic acid (poly(I:C)) were purchased from InvivoGen. The synthetic bacterial lipopeptide Pam3-Cys-Ser-(Lys)4 HCl (Pam3Cys) was obtained from EMC Microcollections. Heat-killed E. coli (NLTC 1048) was provided by Dr. S. Jackson (Department of Medical Microbiology, Cardiff University, College of Medicine, Cardiff, U.K.). IL-1
and TNF-
were obtained from R&D Systems. Human milk-derived sCD14 was isolated and purified as described previously (34). The anti-human CD14 (clone MY4, IgG2b), and CD86-PE conjugate (clone IT2.2, IgG2b) mAbs were obtained from Beckman Coulter, and BD Pharmingen, respectively. The anti-TLR2 mAb (clone TLR2.1, IgG2a) was obtained from eBioscience. The anti-TLR4 mAb (clone HTA125, IgG2a) was provided by K. Miyake (Saga Medical School, Saga, Japan). The anti-TLR3 (clone 40C1285.6, IgG1) and TLR5 (clone 85B152.5, IgG2a) mAbs were donated by ImmunoKontact (AMS Biotech, Oxon, U.K.). The isotype-matched controls IgG2b, IgG1- and IgG2b-PE conjugate were obtained from Diaclone. GM-CSF (Leukine, Sargramostim) was obtained from Berlex Laboratories. IL-13 was donated by Sanofi-Aventis. Proteinase K was obtained from Roche Diagnostics. All chemicals were reagent grade.
Milk, neonatal (cord) blood, cells, and cell activation
Human milk from clinically healthy mothers of term infants (38–39 wk gestation) was obtained after written consent. Any relevant pathological condition (e.g., mastitis) was an exclusion criterion. Milk samples containing 6 pg/ml LPS (Limulus amebocyte lysate test) and/or capable of inducing detectable amounts (above background) of IL-8 or IL-6 (ELISA) by Mono Mac-6 monocytes were excluded from this study. Cell-free milk aliquots were kept at –80°C until use. Three different commercially available infant milk formulations were prepared just before using, following the manufacturer’s instructions. Newborn cord blood—from infants born at 39+ wk of gestation (with the exception of one at 37 wk 4 days)—was collected following local Health Authority Ethical Committee approval from the umbilical cord immediately after vaginal birth or after cesarean section delivery of the placenta. The blood was anticoagulated (129 mM sodium citrate), and the hemocytes were washed (600 x g for 6 min at room temperature) with phenol red-free RPMI 1640 medium (Invitrogen Life Technologies) and resuspended in autologous plasma, adult (AB) plasma, or milk, as indicated in Results, in preparation for the experiments. The adult intestinal epithelial cell (IEC) lines HT-29, SW-620, Caco-2, and T-84 were obtained from the American Type Culture Collection. The nonmalignant fetal IEC line H-4 was developed from a 20-wk-old normal fetal small intestine (36) (provided by Dr. A. Walker, Massachusetts General Hospital, Department of Pediatrics, Harvard Medical School, Boston, MA). The adult IEC lines HT-29, T-84, and Caco-2 were cultured in DMEM medium (Invitrogen Life Technologies) supplemented with 10% FBS (defined FBS; HyClone), 2 mM glutamine, and further supplemented with 1% nonessential amino acids for Caco-2. The H-4 cell line was cultured in DMEM medium supplemented with 10% FBS (HyClone), 2 mM glutamine, 1 mM sodium pyruvate, 10 mM HEPES, 1% nonessential amino acids, and 10 µg/ml insulin (Invitrogen Life Technologies). The human monocytic cell line Mono Mac-6 (provided by H. W. L. Ziegler-Heitbrock, Department of Immunology, Leicester University, Leicester, U.K.) was cultured in RPMI 1640 medium supplemented with 10% FBS (HyClone), 2 mM glutamine, 1 mM pyruvate, 1% nonessential amino acids, and 10 µg/ml insulin (all obtained from Invitrogen Life Technologies). PBMC were obtained after Ficoll density-gradient centrifugation of buffy coats from heparinized blood of healthy donors. Monocyte-derived DCs were obtained following culture of monocyte preparations from PBMC (2 h at 37°C adherence) for at least 6 days in phenol red-free RPMI 1640 supplemented with 10% FBS (HyClone), 2 mM glutamine, 2 mM sodium pyruvate, 1% nonessential amino acids, GM-CSF (800 IU/ml), and IL-13 (100 ng/ml). At day 3, fresh medium containing the cytokines was added to the cultures. The purity of the DC preparations was always 95%, as evaluated by FACS with anti-CD3, -CD19, -CD14, -CD11c, -CD40, -CD80, and -CD83 mAbs. The expression level of the particular TLRs tested in the cells shown or referred to in Results was determined by FACS (data not shown). Specifically, a relatively moderate level of expression of TLR4 was detected in adult PBMC, a low level in cord blood and H-4 cells, a very low level in HT-29 cells, Mono Mac-6 monocytes, and DCs, and undetectable expression (inferred, as indicated by the cell stimulation assays shown in Results, and Northern blots) in Caco-2 and T-84 cells. The TLR5 expression in Mono Mac-6 monocytes was low. TLR2 expression was moderate in Mono Mac-6 cells, and the TLR3 expression in DCs was relatively high. For cell stimulation experiments, the adult IEC lines, the Mono Mac-6 monocytes, PBMC (2 x 105 cells) as well as the fetal IEC line H-4 (1 x 105 cells) and DCs (1 x 104 cells), were cultured in 96-well plates in phenol red-free RPMI 1640/1 mM glutamine medium without serum in the absence or presence of the indicated concentrations of sCD14 or milk, and stimulated with different concentrations of TLR2, TLR3, TLR4, or TLR5 ligands, IL-1 or TNF-, as indicated in Results. At the indicated time points, culture supernatants were collected and tested for IL-6, IL-8, TNF-, and IP-10 by ELISA (Duoset; R&D Systems).
Immunofluorescence and FACS analysis
Detection of cell surface CD86 in DCs with the anti-CD86-PE mAb was performed by immunofluorescence, followed by FACS analysis, as described previously (37). In this study, DCs were preincubated for 10 min with 20% normal rabbit serum before washing and staining.
Gene microarray analysis
For gene expression profile analyses, quadruplicate cultures of Mono Mac-6 monocytes (5 x 106 cells) in serum-free and phenol red-free RPMI 1640/1 mM glutamine medium were supplemented or not with 0.125% human milk (n = 2; collected at day 2 and 3 postpartum). The concentration of sCD14 in the cultures was normalized by supplementing the cultures without milk with purified milk-derived sCD14 at the same concentration as that in the 0.125% milk sample used. The cultures were stimulated with 5 ng/ml LPS for 2, 4, and 16 h. At each time point, the culture supernatants were collected for cytokine determinations (IL-6, IL-8, TNF-, and IP-10), and total RNA was extracted from the cells in preparation for the gene array analysis by using the RNeasy kit (Qiagen) following the manufacturer’s instructions. Comparative analyses of gene expression between cells stimulated in the presence and absence of milk were performed by using a TLR-focused, cDNA-based microarray (GEArrays; SuperArray Bioscience) following the manufacturer’s instructions. One hundred one cDNA fragments (300–600 bp), corresponding to TLR-related (96 cDNAs; see Table I) and control (5 cDNAs: pUC18, GAPDH, cyclophilin A, RPL13A, and -actin) genes, and printed in a tetra-spot format on nylon membranes, were hybridized with biotin-labeled cDNA probes prepared from the RNAs (3 µg) extracted from the experimental samples. The specifically hybridized probes were detected as spots in the microarrays by using a chemiluminescent detection system (SuperArray Bioscience). Data were collected by scanning the x-ray films, and pixel levels were analyzed with the ScanAlyze software (SuperArray Bioscience). Expression values of transcripts scoring as present after background subtraction and normalization, according to an internal program algorithm, were compared. Transcripts with a ratio of normalized expression levels between the two experimental conditions (with or without human milk) altered by 2-fold (average of three independent experiments) were considered modulated.
|
|
In preparation for the cell stimulation experiments described (see Fig. 6), 0.125% milk samples were either: 1) boiled (5 min), 2) subjected to protease digestion (proteinase K, 0.5 U/ml milk, 30 min, 37°C), 3) ultrafiltered (Ultra-free MC centrifugal filters, PL-10 10-kDa cutoff membrane; Millipore), or 4) the milk sample was fractionated by using a restricted access media column (provided by Tosoh Biosciences) consisting of size-exclusion (exclusion limit 80 kDa) and anion-exchange chromatographic media. Up to 4 ml of defatted and casein-depleted milk were loaded onto the column and fractionated as described recently (38). The flow-through and eluted fractions were dialyzed (PBS), and their sCD14 concentrations normalized by addition of purified sCD14 before cell stimulation experiments.
|
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
To test for additional TLR-specific regulatory mechanisms operating in milk, we first focused on TLR4, and evaluated the role that milk sCD14 plays in IEC stimulation via this receptor. IEC stimulation via TLR4 strictly depended on milk sCD14, because an anti-CD14 mAb (MY4) blocked the milk-mediated stimulation of the human (mCD14–) IEC lines HT-29 (Fig. 1) and SW620 (data not shown) induced by LPS or whole Gram-negative bacteria. These results confirmed our previous findings (34). We found, however, that milk sCD14 was necessary, but not sufficient, for efficient TLR4-mediated IEC stimulation, because purified milk-derived sCD14 did not reproduce the effect of whole milk, even when sCD14 was used at concentrations up to fifty times higher than those in the milk samples (Fig. 2A). This enhancing effect of milk was observed in the adult IEC line HT-29, the nonmalignant IEC line of fetal origin H-4 (mCD14+), human monocytes (Mono Mac-6), and DCs (Fig. 2A), as well as in PBMC and the adult IEC lines Caco-2 and T-84 (data not shown). The production of IL-8, IL-6, TNF-, and IP-10 was found to be affected (Fig. 2A). The effect of milk, however, was not restricted to the LPS-induced production of cytokines, because costimulatory molecule (CD86) expression in DCs was also found to be enhanced (Fig. 2B). The enhancing effect was not observed after a week postpartum (Fig. 2C) and, notably, was not reproduced by using infant milk formulations, even when they were supplemented with sCD14 (Fig. 2D). This effect of milk was not limited to stimulation via TLR4. In fact, cell stimulation via TLR5 by the TLR5 ligand bacterial flagellin was found also to be enhanced (Fig. 2E).
|
|
The enhancing effect of milk posed the question of whether milk exerts only a positive regulation on TLR-mediated cell stimulation. In contrast to the enhancing effect on TLR4 and TLR5, in the presence of milk cell stimulation induced via TLR2 by a number of TLR2-specific ligands was found significantly reduced (Fig. 3A), as compared with that in the presence of the coreceptor sCD14 (used at the same concentration as in the milk sample) or to stimulation in the absence of any supplement (cells alone) at high ligand concentration. A similarly negative effect of milk was observed when DCs were stimulated via TLR3 with the viral dsRNA mimic TLR3 ligand poly(I:C) (Fig. 3B). Interestingly, poly(I:C)-induced CD86 expression in DCs was not affected by milk (Fig. 3C). This was also in contrast to the positive effect observed when TLR4 was tested (Fig. 2B). The inhibitory effect was not reproduced by using infant milk formulations (Fig. 3D), like the enhancing effect on TLR4 (Fig. 2C).
|
We asked whether the regulatory activity of human milk was restricted to TLRs, or whether other receptors were also affected. To address this issue, we tested the effect of milk on cell stimulation via the IL-1R, which shares with TLRs the (MyD88-dependent) signaling pathway, because both receptors have a conserved region in their intracytoplasmic domain, known as the Toll/IL-1R domain, which is crucial for signaling (31). In addition, the effect of milk on signaling via a TLR nonrelated receptor, the TNFR, was tested. Fig. 4 shows that cell stimulation via either the IL-1R or TNFR was not affected by milk. This finding, together with the capacity of anti-CD14 mAb to block milk-mediated cell stimulation (Figs. 1 and 2A), and the lack of effect of milk on cells alone (Figs. 1, 2, and 3), indicated that the modulatory effect is exerted in a specific manner on the ligand-induced TLR-mediated responses.
|
We then asked whether the modulatory effect of milk observed in a number of cell types and lines as well as in adult PBMC can be reproduced in normal newborn immunocompetent cells. It has also recently been demonstrated that newborn (cord) plasma confers greatly reduced sensitivity to TLR4-mediated leukocyte stimulation, as compared with adult plasma (39). Therefore, we also asked whether milk behaves differently from neonatal plasma. To address these issues, we used whole neonatal (cord) blood as a minimally perturbed ex vivo model system, and compared the capacity of newborn plasma and milk to mediate cell stimulation induced via TLR4. Washed newborn hemocytes were resuspended in 100% autologous plasma, 100% adult (AB) plasma, or 1–2% human milk. Notably, just a 2% concentration of milk was sufficient toenhance IL-8 and TNF- release by newborn hemocytes in response to LPS, as compared with neonatal plasma (Fig. 5). This finding indicated that newborn peripheral immunocompetent cells are also susceptible to regulation by milk, and that milk can compensate for the low capacity of neonatal plasma to support responses to LPS.
|
To evaluate the extent of milk’s regulatory effect and obtain an insight into the underlying mechanism, we used a gene array approach. We focused on TLR4, and used the TLR4-mediated (Mono Mac-6) monocyte stimulation by LPS as a model system. We asked whether the enhancing effect observed on the LPS-induced cytokine and costimulatory molecule expression reflected a general (positive) effect of milk on the ligand-induced TLR4-mediated gene expression. To address this issue, we performed a comparative, TLR-focused, microarray analysis of the expression of 96 TLR-signaling-related genes by monocytes stimulated with LPS in the presence, as opposed to the absence, of human milk (preliminary gene array studies and the results shown in Figs. 1, 2, and 3 indicated that milk has no effect on nonstimulated cells). To normalize the concentration of sCD14, cell cultures stimulated in the absence of milk were supplemented with sCD14. Primary and secondary transcriptional changes in gene expression were evaluated by conducting the analysis at relatively early (2 and 4 h) and late (16 h) time points poststimulation.
Eighty four percent of the genes (81 genes) were found affected (2-fold) by milk upon LPS stimulation (Table I). Eighty three percent of the genes affected (67 genes) were up-modulated 2 h poststimulation (Table I, early up-modulation category). These early up-modulations were accompanied or not by later effects. Notably, however, a total of 34 genes (42%) were found down-modulated, including 10 genes after 2 h or 4 h (Table I, early down-modulation category), 3 after 16 h (Table I, late modulations category), and a further 21 preceded by earlier modulations (Table I, early up-modulation category). The relative down-modulation of 16 of these genes was observed at 4 h, and was preceded by an earlier (2 h) up-modulation (Table I, early up-modulation category). This reflects, most likely, the effect of milk on the expression kinetics of these genes (i.e., faster kinetics), rather than an absolute negative effect on their expression levels at 4 h.
Early up-modulations affected many genes coding for TLR signal intermediates, e.g., ECSIT, IL-1R-associated kinase family members, and isoforms of p38MAPK. The transcriptional factor IFN regulatory factor-3 and those of the NF-
B family, I-REL (REL-B), KBF1 (p105/p50), NFKB2 (p100/p52), NFKB3 (REL-A), proinflammatory mediators and cytokines (IFN-1, IFN-
, IL-1, IL-1, IL-2, IL-6, IL-8, IL-12, IL-12, IP-10, TNF-, TNF-), TLRs (TLR2, TLR4, TLR5, TLR7, TLR8, TLR9), the TLR4 accessory molecule MD2, as well as the apoptosis intermediate and NF-B inducer FLICE (caspase-8) were also found up-modulated. Some TLR signal intermediates were, however, down-modulated early in the presence of milk, e.g., JIP3, MKK3, TAK1, TIRAP (MAL) as well as the apoptosis intermediate Fas-associated via death domain protein. Notably, down-modulation of mRNA levels (2 h) also strongly affected the transcript coding for TLR3. Late (16 h, positive or negative) effects of milk, accompanied or not by earlier modulations, were observed in 
31% of the genes affected (25 genes). These late effects most likely depended on protein synthesis induced in the initial wave of gene induction. Interestingly, 64% of the late-modulated genes (16 genes) were subjected to down-regulation. Late modulations not accompanied by earlier effects were observed in only four genes (Table I, late modulations), and were negative, with the notable exception of the relatively strong up-regulation of the gene coding for the TLR-negative regulator molecule SIGIRR (TIR8).
The validity of the microarray analysis was confirmed by the correlation between the relatively high concentrations of IL-8, IL-6, TNF-, and IP-10 detected in the cultures supplemented with milk and the early up-modulation of the corresponding transcripts (Fig. 2A and Table I) as well as by performing Northern blots for RNA analysis of prototypical examples of early down-modulated (Fas-associated death domain protein) and late-modulated (IKK1) genes (data not shown).
A milk protein component is responsible for the modulatory effect on LPS-induced cell stimulation
To obtain an insight into the molecular nature of the milk component responsible for the modulatory effect of milk on LPS-induced cell activation, the milk samples were subjected to heat, protease digestion, and size fractionation. Fig. 6 shows that boiling as well as proteinase K digestion of the milk samples before the experiments abrogated the enhancing effect (Fig. 6, A and B). Furthermore, the enhancing effect of milk was reproduced by using the retentate, but not the filtrate following size fractionation of the samples by ultrafiltration through a 10-kDa cutoff membrane (Fig. 6C). The milk samples were fractionated further by using a combination of anion-exchange and size-exclusion (exclusion limit, 80 kDa) chromatography in a single column. The sCD14 concentration in the flow-through and eluted fractions was normalized by the addition of purified milk-derived sCD14, and the capacity of both fractions to mediate HT-29 cell stimulation by LPS was tested. Fig. 6D shows that the enhancing effect of milk was reproduced by using the flow-through, but not the eluted fraction, and thecombination of both fractions restored the original effect of the unfractionated milk. Together, these results indicated that a proteinaceous milk component(s) of 80 kDa is responsible for the TLR4-enhancing effect.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
The use of individual TLR ligands, mostly corresponding to or representing main microbial components expressed and released by a variety of bacteria colonizing the neonatal gut (except for heat-killed L. monocytogenes and poly(I:C)), allowed us to dissect and analyze the response of the TLR tested separately.
Following ligand-induced cell stimulation, the modulatory effects exerted by human milk on TLR2 and TLR3, and those on TLR4 and TLR5 were found to be opposite. Although a negative modulation was exerted on TLR2- and TLR3-mediated responses, those via TLR4 and TLR5 were enhanced. These effects were observed only when relatively early milk samples (days 1–5 postpartum) were tested. The fact that milk did not affect IL-1R signaling (Fig. 4), and that it affected TLR2 and TLR3 differently from TLR4 and TLR5, indicates that the primary effect of milk is exerted upstream of the signaling pathway, most probably proximal to ligand recognition.
As regards the enhancing effect on TLR4-mediated responses, the mRNA level for TLR4 was found up-modulated (2-fold) early (2 h) in the presence of milk (Table I). Increased TLR4 expression may contribute to the enhancing effect. However, it cannot explain the modulatory effect of milk on genes that were found affected as early as the one for TLR4, nor how TLR4 was up-modulated early (2 h) in the first place. The potency of the receptor, rather than receptor numbers, seems to be affected, because the sensitivity to low doses of ligand was increased preferentially (Fig. 2A; HT-29), whereas flow cytometric analysis did not show variations in TLR-4 cell surface expression (data not shown). Notably, the mRNA level for the accessory molecule MD2 was strongly up-regulated in the presence of milk at a relatively early time point (Table I). It is well documented that this secreted molecule is crucial to LPS recognition and the function of the LPS receptor complex (29, 40, 41). It is thus possible that an increased level of MD2 results in selective (up- or down-) modulation or recruitment of signal intermediates following LPS triggering of the receptor. This in turn may influence the quality and extent of the biological response. However, this cannot explain the effect of milk on genes that were found affected as early as the one for MD2 (Table I). It could be argued that MD2 present in milk contributes to this effect. This would be consistent with the proteinaceous nature of the TLR4-enhancing milk component (Fig. 6). Although this possibility cannot be excluded, we failed to detect MD2 in a number of milk samples (our unpublished observations). The presence of the LPS binding protein (LBP) in milk would also explain the enhancing effect of milk on cytokine and costimulatory molecule expression, because LBP—at low concentrations—is known to facilitate the interaction of LPS with CD14, thereby increasing cell sensitivity to LPS substantially (42). However, this is unlikely because of the following: 1) we have reported that the concentration of LBP in milk is extremely low, at the limit of detection (34), and at the concentrations of milk used in this study (0.1–0.2%), milk LBP was undetectable; 2) we also found that preincubating milk with anti-LBP Ab does not affect IEC sensitivity to LPS (43); 3) supplementing milk samples with highly purified LBP (0.5 µg/ml) does not increase the enhancing capacity (our unpublished observations); and 4) the Mr of LBP (60, 000) is not consistent with that estimated for the TLR4-enhancing milk component (80,000; Fig. 6D). We also tested for the possible involvement of GM-CSF, because this growth factor was shown to up-modulate TLR-mediated responses (44, 45). However, in our hands, GM-CSF (100 or 200 IU/ml) either alone or as a milk supplement did not affect Mono Mac-6 sensitivity to LPS (our unpublished observations).
Further experimentation will be necessary to determine the nature of the primary event resulting in the modulatory effect of milk on TLR4 signaling. Similarly, the primary events responsible for the positive effect on TLR5 and negative effects on TLR2 and TLR3 remain to be elucidated. With regard to the negative effect, milk markedly inhibited the TLR3-mediated poly(I:C)-induced cytokine production by DCs without affecting costimulatory molecule expression (Fig. 3, B and C), suggesting that the MyD88-dependent (cytokine production), but not the independent pathway, which controls costimulatory molecule expression and is the main signaling pathway for TLR3 (31), was affected. This was in marked contrast to the effect of milk on TLR4, because here both cytokine production and costimulatory molecule expression were affected, and in a positive manner (Fig. 2, A and B). Similar to TLR3, the negative regulation of TLR2 appears to result from an effect on the MyD88-dependent pathway, because this is the only signaling pathway used by this receptor. At present, we are evaluating the extent of the milk sTLR2’s contribution to this negative regulation. Interestingly, TLR5 was positively affected by milk despite only using the MyD88-dependent signaling pathway, like TLR2. These findings support the claim that the primary event(s) affecting TLRs is exerted upstream of signaling.
The individual expression of TLRs in the initial cultures of the cells tested is unlikely to contribute to the differences observed, because the different capacity of milk, sCD14, and nonsupplemented cultures (cells alone) to mediate ligand-induced cell stimulation were observed in the same cell line by testing and comparing parallel cultures of the same cell line preparation. Moreover, the modulatory effect of milk was observed in cells irrespective of the expression level of the TLR involved, e.g., the TLR4-enhancing effect was observed in cells with moderate (PBMC), low or very low (H-4, cord blood, Mono Mac-6, HT-29, DCs) as well as undetectable—inferred—(Caco-2 and T-84 cells) TLR4 expression.
The TLR4-mediated cellular response was used as a model system in comparative, TLR-focused, gene array analysis to determine the extent of modulatory activity, and to obtain an insight into the underlying mechanism. The majority of the TLR-related genes tested were up-modulated by milk upon cellular stimulation by LPS. However, a substantial number of genes were down-modulated or not affected, and others were only modulated late as a result of secondary transcriptional events (Table I). These findings indicated that milk behaves as a general TLR modulator, and not as an enhancer of a discrete number of genes. Notably, among the genes up-modulated early were those coding for TLR signal intermediates, including three of the four isoforms of p38MAPK, which are involved in the MyD88-dependent and -independent signaling pathways of gene induction, and the transcriptional factor IFN regulatory factor-3, which is crucial to the MyD88-independent signaling pathway (31). Consistent with these findings, milk enhanced not only cytokine production (Fig. 2A), which is controlled by the MyD88-dependent and -independent signaling pathways, but also IP-10 and costimulatory molecule expression (Fig. 2, A and B), which follows a MyD88-independent signaling pathway (31). Transcripts coding for family members of the transcription factor NF-B and the protease FLICE (caspase-8) were also found up-modulated early (Table I). Interestingly, caspase-8, although involved in cellular apoptosis, has recently been shown to be required for the efficient nuclear translocation of the transcription factor NF-B following TLR4 stimulation (46). These findings are consistent with the enhancing effect of milk, because NF-B plays a pivotal role in LPS-mediated gene induction (31). The gene array analysis also identified transcripts coding for molecules influencing T cell activation and responses that were up-modulated by milk, including IL-1, IL-1, IL-12 (IL-12p35), and IL-12 (IL-12p40). It will thus be important to evaluate the effect that this modulation may have on the adaptive immune response. A late (16 h), but relatively strong up-modulation of the transcript coding for SIGIRR was observed. This cell surface molecule is known to negatively regulate TLR4 signaling (47). The putative overexpression of SIGIRR, together with that of other TLR-negative regulators whose transcripts were found up-modulated at 2 h, namely IL-1R-associated kinase-M and TOLLIP (Table I), may be part of a negative feedback to control the extent of the milk’s modulatory effect.
No correlation between the milk-induced enhanced expression of CD86 and the level of the corresponding mRNA was observed (Fig. 2B and Table I). It is possible that the (up) regulation of CD86 was exerted at the protein synthesis or posttranslational levels. Alternatively, after LPS stimulation, milk may have induced mobilization to the cell surface of an internal pool of this molecule.
Overall, the data support the notion that the primary effect of milk is exerted upstream of the signaling pathway. In this regard, recent studies have demonstrated that different cell surface molecules associating with TLRs impart specificity to TLR-mediated microbial recognition (48, 49, 50). Furthermore, it has been shown that, following LPS stimulation, TLR4, MD2, and a number of additional cell surface molecules are recruited to lipid rafts, and that the composition and stoichiometry of this receptor cluster may dictate the type of signal transduced (51). We therefore hypothesize that milk component(s) may induce or facilitate the differential association of cell surface molecules with TLRs. These differential molecular associations may affect the capacity of TLRs to recognize or respond to ligands and constitute the primary events in milk-mediated TLR signaling. This would be consistent with documented observations, indicating that the reactivity pattern of gut immunocompetent cells is influenced by the local environment (52, 53, 54). Our findings thus suggest that a human breast milk-, but not an infant milk formulation-rich environment may affect the reactivity pattern of gut immunocompetent cells by modulating TLR reactivity specifically and differentially.
The biological reason for the opposite modulatory effect of milk on TLR2 (negative) and TLR4 (positive), the two main receptors for microbial components, is not clear. It is possible that following birth, the opposite modulation by milk promotes the efficient recognition of and response to potentially harmful LPS-producing Gram-negative bacteria via TLR4, whereas allowing the establishment of bifidobacteria (Gram-positive) as the predominant intestinal microbiota through a low TLR2 reactivity. Consistent with this possibility is the observation that breast-fed, but not formula-fed, full-term infants have a preferred intestine microbial flora with predominance of the probiotic bifidobacteria (55). More experimentation will, however, be required to determine whether the modulatory effects on TLR-mediated immune responses described in this study contribute to the reported protective effects of breastfeeding against necrotizing enterocolitis, an inflammatory condition affecting premature infants. Testing the effect of preterm milk samples may help to clarify this issue.
The modulatory effects of human milk seem to extend to neonatal peripheral leukocytes (Fig. 5), suggesting that if the milk component(s) ultimately responsible for modulating TLR responses reach the periphery, they would influence neonatal innate immune responses to systemic infections. This study paves the way for the identification of such milk components.
The findings reported in this study indicate that, in addition to sCD14 and sTLR2, other TLR modulatory molecules are present in milk and involved in novel mechanisms regulating the extent and quality of TLR activation. The data increases our understanding of how breast milk components contribute to adequate interactions between microorganisms and immunocompetent cells. The identification of these molecules will advance our knowledge of the mechanisms controlling innate immune responses and may inform the design of improved infant milk formulations, as well as being of relevance to gut inflammatory conditions where optimizing TLR-mediated tissue responses may be of therapeutic potential.
|
Acknowledgments |
|---|
|
Disclosures |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
|
Footnotes |
|---|
1 This work was supported by the Wellcome Trust of Great Britain. 
2 Address correspondence and reprint requests to Dr. Mario O. Labéta, Infection and Immunity, Department of Medical Biochemistry and Immunology, Cardiff University, College of Medicine, Henry Wellcome Research Building, Heath Park, Cardiff CF14 4XX, United Kingdom. E-mail address: wmdmol{at}cardiff.ac.uk
3 Abbreviations used in this paper: mCD14, membrane-bound CD14; s, soluble; DC, dendritic cell; poly(I:C), polyinosinic-polycytidylic acid; IEC, intestinal epithelial cell; LBP, LPS binding protein.
Received for publication September 1, 2005. Accepted for publication December 28, 2005.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
induction by bacterial lipopeptides, lipopolysaccharide, and imiquimod, but preserves the response to R-848. J. Immunol. 173: 4627-4634. B activation by antigen receptor. Science 307: 1465-1468. -glucans. J. Exp. Med. 197: 1119-1124. This article has been cited by other articles:
|
|
 |
|
  M. Kalliomaki, M. Carmen Collado, S. Salminen, and E. Isolauri Early differences in fecal microbiota composition in children may predict overweight Am. J. Clinical Nutrition, March 1, 2008; 87(3): 534 - 538. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||
