Airway epithelial cells modify immune responses by inducing an anti-inflammatory microenvironment



The upper airways are prone to contact with pathogenic as well as non-pathogenic microbes, therefore immune recognition principles have to be tightly controlled. Here we show that human BEAS-2B bronchial epithelial cells inhibited secretion of the pro-inflammatory cytokines TNF-α and IL-12 by monocytes, macrophages and dendritic cells. This inhibitory effect could be transferred by supernatant of resting BEAS-2B cells and was also observed when primary murine tracheal epithelial cells were prepared. In contrast to inhibition of pro-inflammatory cytokine secretion epithelial cell-conditioned dendritic cells showed increased expression of IL-10 and arginase-1, thus displaying properties of alternative activation. Accordingly, Toll-like receptor-mediated up-regulation of CD40, CD86 and PD-L2 (CD273) on murine dendritic cells was reduced in the presence of bronchial epithelial cell supernatant. However, expression of negative regulatory PD-L1 (CD274) was increased and dendritic cell induced proliferation of T lymphocytes was diminished. Epithelial cells also showed a direct inhibitory effect on T lymphocyte proliferation and this was due to the constitutive secretion of TGF-β by bronchial epithelial cells. Moreover, epithelial cell-conditioned T lymphocytes showed increased differentiation towards IL-10-producing Tr1 cells. The results indicate that bronchial epithelial cells induce a non-inflammatory microenvironment that regulates local immune homeostasis.


BM-derived DC


epithelial cell-conditioned medium


thymic stromal lymphopoietin


Due to its large surface, airway epithelium is prone to encounter airborne microorganisms. Hence, bronchial epithelial cells have developed sophisticated defense mechanisms including formation of tight junctions, mucociliary clearance and secretion of anti-microbial chemicals 1. Moreover, it was recently recognized that bronchial epithelial cells also express various pattern recognition receptors thus being able to actively sense and respond to microbial contact 2.

Recognition of pathogen-associated molecular patterns conceptually was proposed to be the recognition principle within innate immunity 3. Nowadays it is known that prototypical pattern recognition receptors including Toll-like receptors (TLR) are not only expressed on professional innate immune cells but also cover epithelial cells within various organs including the gut 4, the airways 2, 5, 6, the oral cavity 7 and the skin 8. In addition, it is now clear that TLR not only sense microbial components derived from pathogenic microbes but also recognize microbial compounds in general 9. In other words, TLR cannot discriminate pathogenic from apathogenic microorganisms. These both facts raise questions about the concept of pattern recognition at non-sterile epithelial surfaces.

Taking these limitations into account, Eyal Raz 9 recently proposed a concept of organ-specific and graded immune responses. Accordingly, each organ senses infectious danger in a specific way and organ physiology modulates and instructs local immune responses. An important regulating variable is the degree of non-pathogenic microbial flora at a given surface. From such a concept two hypotheses emerge: pattern recognition sensitivity has to be controlled at non-sterile epithelial surfaces and organ-specific stroma cells control local immune responses.

These hypotheses have now been tested to some extent for the gut where microbial encounter is permanent and strong 4, 1013. However, only limited information is available for the situation in the upper airways. It has been reported that the sensitivity and the activation threshold of TLR in bronchial epithelial cells is indeed regulated 6, 1416, supporting the concept that airway epithelial cells actively restrict pattern recognition principles.

However, little is known about the impact of airway epithelial cells on the control of local immune responses. In a recent report alveolar macrophages were shown to be constantly inhibited by transforming growth factor-β (TGF-β) presented by epithelial cells 17. During infectious encounter this tonic inhibition was transiently released. Moreover, it is known that, in general, mucosal DC differ from splenic counterparts. It has been proposed that epithelial cells mount a specific microenvironment affecting the phenotype of professional immune cells and soluble factors like TGF-β, nitric oxide (NO), prostaglandins, IL-10 and thymic stromal lymphopoietin (TSLP) were suggested as possible mediators 1719. However, no detailed analysis of upper airway bronchial epithelial cells and their impact on immune homeostasis of innate and adaptive immunity is available.

In this study we analyze whether bronchial epithelial cells are able to manipulate professional innate and adaptive immune responses, including TLR-mediated activation of monocytes and dendritic cells and modulation of T lymphocyte activation and differentiation. The findings ascribe a new function to airway epithelial cells and impact our understanding of immunological disorders within this organ, including asthma, chronic inflammation and infection.


Human bronchial epithelial cells inhibit TLR-mediated activation of CD14+ monocytes and macrophages

To analyze whether bronchial epithelial cells modulate professional immune cells, we co-incubated graded numbers of human bronchial BEAS-2B cells with CD14+ monocytes, added LPS as a microbial stimulus and measured secretion of TNF-α (Fig. 1A). BEAS-2B cells 6 as well as monocytes both express TLR-4; however, only monocytes respond with induction of TNF-α. We observed that BEAS-2B cells starting from a 1:10 ratio (BEAS-2B:monocytes) inhibited LPS-induced TNF-α secretion from monocytes. To analyze whether this inhibition was dependent on formation of cell contacts, we repeated the experiment in a transwell setting (Fig. 1B). BEAS-2B cells in a ratio of 1:2 were still effective in the inhibition of monocytes, yet the decrease in TNF-α secretion was weaker than in the co-incubation experiment. However, when BEAS-2B cells were cultivated for 1 day prior to addition of monocytes, inhibition was about 80%. This indicated that bronchial epithelial cells produced soluble factors that are sufficient to inhibit monocytes; cell contact-dependent mechanisms are also operative and account for about 20% of the inhibitory effects.

Figure 1.

Bronchial epithelial cells inhibit activation of monocytes. (A) BEAS-2B were seeded as indicated, co-cultured with 105 CD14+ monocytes and left unstimulated (open bars) or stimulated with 10 ng/mL LPS overnight (filled bars). TNF-α secretion was analyzed by ELISA (mean + SD, one out of three independent experiments). (B) BEAS-2B cells (B2B) and human monocytes were co-cultured in a transwell system. Additionally BEAS-2B cells were cultivated for 1 day prior to addition of monocytes. Cells were then stimulated as in (A). (C) ECCM was collected at the indicated time point. Monocytes were stimulated with LPS in the presence of ECCM (50%) and analyzed as in (A).

As LPS is able to stimulate both cell types, BEAS-2B cells and monocytes, and because the inhibitory effect of BEAS-2B cells was stronger when they were rested prior to addition of monocytes, we next investigated whether epithelial cells without microbial activation are constitutively able to influence monocyte activation. We prepared supernatants from unstimulated BEAS-2B cells that were rested for different time periods, and epithelial cell-conditioned supernatants were added to monocytes, which subsequently were stimulated with LPS (Fig. 1C). We observed that under such homeostatic conditions epithelial cells constitutively produced soluble inhibitory mediators that decreased TNF-α secretion by TLR-stimulated monocytes. However, depending on the experimental situation only 24–48-h conditioned supernatants were inhibitory. Inhibition could be titrated by adding different amounts of BEAS-2B supernatants (data not shown).

We further analyzed the inhibition of LPS-stimulated human macrophages by BEAS-2B-conditioned supernatant. As for the experiments with monocytes, we observed an inhibition from 11.9 ± 1.2 ng/mL TNF to 4.2 ± 0.5 ng/mL in the presence of BEAS-2B supernatant.

Bronchial epithelial cells inhibit activation of DC by various TLR ligands

We next tested whether the inhibitory effects of bronchial epithelial cells are also operative on DC, which are central interface cells connecting innate with adaptive immunity. We observed that BEAS-2B supernatant also inhibited activation of human DC by LPS (Fig. 2A). Furthermore, the observed inhibitory effect was not specific for TLR4 signaling because stimulation of DC by other TLR ligands, including lipoteichoic acid, Pam3CSK4 and poly(dI:dC) was also inhibited (Fig. 2A). These findings argued for a general modulation of innate immune cells by bronchial epithelial cells. To determine whether this constitutive inhibitory property is a characteristic of airway epithelium, we generated supernatants from two bronchial epithelial cell lines, BEAS-2B and IHAEo, and compared them to control supernatants prepared from PBMC or HEK293 cells, which were seeded in even higher cell concentrations than the bronchial epithelial cells (Fig. 2B). We observed that only supernatants from IHAEoand BEAS-2B cells had inhibitory properties.

Figure 2.

Bronchial epithelial cells inhibit TLR-mediated activation of APC. (A) Monocyte-derived DC (105) were stimulated with 10 µg/mL LTA, 10 ng/mL LPS, 5 µg/mL Pam3CysK4 and poly(I:C), respectively, in the absence (open bars) or presence of 50% ECCM (ECM, filled bars). IL-12p40 secretion was analyzed by ELISA (mean + SD, one out of two independent experiments). (B) Supernatant (SN) was collected from BEAS-2B cells, IHAEo cells, HEK293 cells or PBMC as described in Materials and methods. Monocytes were stimulated with LPS in the presence of conditioned medium (50%) and left unstimulated (open bars) or stimulated with LPS (filled bars) (mean + SEM of normalized values, n = 2). (nd, not detectable)

Primary murine tracheal epithelial cells inhibit pro-inflammatory activation of DC

To further confirm that airway epithelium affects innate immunity, and to broaden our findings, we next prepared primary murine tracheal epithelial cells according to a published protocol 20. These cells showed the typical characteristics of epithelial cells (tight junction formation, transepithelial resistance and expression of specific marker genes including defB1 and cftr; data not shown) and had no significant contamination with myeloid cells or fibroblasts. We prepared epithelial cell-conditioned medium (ECCM) and stimulated bone marrow-derived DC (BMDC). We observed that epithelial-derived supernatant inhibited TNF-α and IL-12p40 secretion by LPS-stimulated DC (Fig. 3A and B) in a dose-dependent manner. Moreover, using primary, polarized epithelial cells allowed us to generate an ECCM from the luminal as well as the basolateral side. Both preparations were able to inhibit LPS-induced IL-12p40 secretion from BMDC in a similar way (Fig. 3C).

Figure 3.

Primary tracheal epithelial cells modulate TLR-mediated activation of dendritic cells. BMDC (2 × 105) were seeded into a 96-well plate and left without stimulus (open bars) or stimulated with 100 ng/mL LPS (filled bars) overnight in the presence of graded amounts of ECCM (ECM) from primary air-liquid interface cultures. Cell-free supernatant was collected and detection of TNF-α (A) and IL-12p40 (B) secretion was performed by ELISA (mean + SD, one out of three independent experiments). (C) BMDC were stimulated with LPS in the presence of ECCM (50%) that had been generated either from the luminal or the basolateral side of primary air-liquid interface epithelial cultures. (D) BMDC were preincubated with ECCM (50%) overnight and subsequently stimulated with 100 ng/mL LPS for 2 h. Expression of the indicated genes was examined by quantitative RT-PCR. Transcript expression was normalized to the housekeeping gene β-actin (mean + SD, n=3). *p<0.05. (E) ECCM-conditioned BMDC were stimulated overnight and expression of CD40, CD80, CD86, PD-L1 and PD-L2 was analyzed by flow cytometry (one representative of three independent experiments).

Epithelial cell-conditioned DC show properties of alternatively activated cells

To analyze further genes induced by LPS in murine DC, we performed quantitative RT-PCR experiments (Fig. 3D). Confirming the above results, we observed that ECCM inhibited expression of IL-12 and TNF-α at the mRNA levels. Surprisingly, ECCM itself increased the induction of iNOS as well as IL-10. Furthermore, DC conditioned by epithelial cells showed increased expression of arginase-1, which is a marker enzyme of alternative activation of macrophages and DC. This indicates that DC are not inhibited by bronchial epithelial cells in a general way but rather are modulated to express a different transcription program.

DC stimulated by LPS in the presence of epithelial cell-derived supernatant were further analyzed regarding their expression of costimulatory molecules (Fig. 3E). In accordance with the phenotype of alternative activation, we observed that primary tracheal epithelial cells inhibited TLR-mediated up-regulation of CD40 and CD86. In contrast, expression of PD-L1 was up-regulated by epithelial cell supernatant. All the murine data were confirmed for another TLR ligand CpG-DNA (data not shown). These results show that bronchial epithelial cells modulate activation of professional immune cells by microbial stimuli.

Bronchial epithelial cells inhibit T lymphocyte proliferation

Having shown that bronchial epithelial cells modulate APC (monocytes, DC) to adopt an alternatively activated phenotype, we asked whether this would also result in altered activation of T lymphocytes. We, therefore, pre-incubated human monocyte-derived dendritic cells with BEAS-2B-conditioned medium for different time periods. Subsequently, DC were washed intensively to remove any BEAS-2B medium and allogenic CD3+ T lymphocytes were added. After 4 days, T cell proliferation was determined by [3H]thymidine incorporation (Fig. 4A). We observed that DC pre-conditioned with epithelial supernatant were less effective in the induction of T lymphocyte proliferation than untreated DC. This effect was significant; however, reduction of proliferation was not very effective.

Figure 4.

Airway epithelial cells inhibit T lymphocyte activation. (A) Human monocyte-derived DC were incubated for the indicated time with ECCM (ECM, 50%) and washed intensively. CD3-sorted T cells from a different donor were added. Proliferation was determined by [3H]thymidine incorporation (mean + SEM, n=5). p<0.05. (B) CD3-sorted T cells were incubated with DC and graded amounts of ECCM were added. In contrast to (A) ECCM was present over the whole incubation period. Proliferation was measured (one out of three experiments). (C) T cells were stimulated with beads coupled with anti-CD3/CD28/CD2 antibodies in the presence of ECCM and analyzed as in (B). (D) BEAS-2B cells were examined by quantitative RT-PCR for the expression of TGF-β, IL-10 and TSLP mRNA. Transcript expression was normalized to the housekeeping gene β-actin. (E) TGF-β protein from unstimulated BEAS-2B cells or cells stimulated with heat-killed P. aeruginosa (MOI 10) was detected by ELISA over the indicated time period. (F) DC were incubated with allogenic T cells in the presence of increasing amounts of ECCM and neutralizing anti-TGF-β antibody, isotype control antibody or without any antibody (mean + SEM, n=3).

Next we analyzed proliferation of T cells induced by human DC in the presence of BEAS-2B supernatant. In contrast to the previous experiment, ECCM was present over the whole incubation period (Fig. 4B). We observed a decrease in T cell proliferation, which was dependent on the amount of ECCM added, and which was much more effective than with pre-conditioned DC.

This led us to hypothesize that BEAS-2B supernatant might also provide a direct effect on T lymphocytes. To analyze this question, CD3+ T lymphocytes were activated independently of APC by the addition of beads bearing antibodies against CD2/CD3/CD28 (Fig. 4C). We found that supernatant of bronchial epithelial cells was operative in inhibiting T cell proliferation independently of the presence of APC. Finally, we observed that even the addition of epithelium-derived medium into an ongoing T lymphocyte response 24 h after primary activation was able to decrease T cell proliferation (data not shown). Taken together, the results show that bronchial epithelial cells modify activation of T lymphocytes by modulating APC as well as by directly influencing T cells.

TGF-β contributes to the inhibitory effects of bronchial epithelial cells on T lymphocytes

Based upon the observation that airway epithelium cells directly affected T cell activation, we decided to test candidate factors that had been shown previously to play an inhibitory role in airway or intestinal epithelial cells, including TGF-β, IL-10, TSLP and NO 2123. When analyzing BEAS-2B cells for transcription of known inhibitory mediators, we found that TGF-β transcripts were detectable at high levels (Fig. 4D). This translated into a steady increase of TGF-β protein secreted into the supernatant of BEAS-2B cells (Fig. 4E) and the kinetics of protein accumulation correlated with the inhibitory potency (Fig. 1C). Interestingly, constitutive TGF-β secretion was decreased when cells were stimulated with heat-killed Pseudomonas aeruginosa (Fig. 4E). IL-10 and TSLP were expressed at very low copy numbers (Fig. 4D) and no IL-10 protein was determined in BEAS-2B cell supernatant or in supernatant of primary air-liquid interface cultures (data not shown). We detected no significant amount of NO in ECCM, and application of L-NAME, which inhibits NO production, did not affect the inhibitory potency of the conditioned medium (data not shown).

To further study the role of epithelium-derived TGF-β on T cell proliferation, we used a neutralizing antibody. We observed that recombinant TGF-β was able to inhibit T cell proliferation (reduction to 44% of control), which was reverted by addition of the neutralizing antibody (restoration to 95%) but not by an isotype control. Blocking TGF-β activity resulted in a decrease but not complete loss of inhibitory activity of BEAS-2B supernatant on T cell proliferation (Fig. 4F), which argues for a partial role of TGF-β in this setting. Interestingly, neutralization of TGF-β did not affect the inhibition of ECCM on LPS-induced TNF-α secretion in monocytes and DC (data not shown).

Bronchial epithelial cells expand regulatory T cells

APC with an alternatively activated phenotype have been proposed to expand regulatory T cells 24. As such DC were induced by bronchial epithelial cells (Fig. 3) and because epithelial cells directly affected T cell proliferation in a manner dependent on TGF-β secretion, a cytokine that is involved in generation of regulatory T cells and Th17 cells 25, we decided to study polarization of T lymphocytes in more detail. To address this question we analyzed murine bronchial epithelial cells for their effects on T cell differentiation in vitro. T lymphocytes were stimulated with beads carrying antibodies against CD3 and CD28 either in the presence or absence of ECCM. After 3 days of stimulation we observed that ECCM increased the concentration of secreted IL-10 strongly (Fig. 5A). To study this effect further, we repeated the experiment now driving T cell polarization by addition of IL-6 and TGF-β. Without exogenous addition of cytokines we observed that ECCM slightly increased the frequency of IL-10-producing cells (Fig. 5B and C). This effect was much more pronounced when TGF-β and IL-6 were added. Under these conditions ECCM strongly increased the frequency of IL-10-producing Tr1 cells (Fig. 5B and C). In addition, an increase of Th17 cells was also visible. Furthermore, when differentiation was done only with TGF-β, ECCM increased the frequency of CD25/Foxp3+ regulatory T cells (Fig. 5D).

Figure 5.

Bronchial epithelial cells expand IL-10-producing Tr1 cells. (A) Murine T lymphocytes were activated with anti-CD3/CD28 beads either in the absence (open bar) or presence (filled bars) of murine tracheal epithelial cell conditioned medium (ECM). On day 3, IL-10 was determined in the supernatant by ELISA. (B, C) T cells were activated as in (A) either without addition of cytokines or in the presence of recombinant IL-6 and TGF-β. On day 3 PMA/ionomycin restimulation was performed and IL-10 and IL-17 were analyzed by intracellular cytokine staining. (B) depicts one typical analysis, (C) shows data (mean + SEM) from n=4 experiments. p<0.05. (D) T cells were stimulated as in (A) either without additional cytokines or in the presence of TGF-β. On day 3, cells were stained for Foxp3 and CD25.


It has been shown convincingly that epithelial cells at surfaces that regularly or permanently encounter microbes express various TLR and are able to actively sense danger due to infectious agents 48. This has two important implications: First, epithelial cells at non-sterile surfaces have to regulate their TLR sensitivity to avoid continuous threat of inflammation and, second, organ-specific mechanisms of regulating immunity have to exist. The first fact has now been intensively studied and multiple mechanisms including threshold regulation of TLR stimulation 6, 10, anatomical sequestration 14, 26, inhibitory signaling molecules 27, 28, co-receptor modulation 6 and specific signaling pathways 29 have been described. The second hypothesis was the subject of this study.

We here show that bronchial epithelial cells modulate activation of monocytes, macrophages, DC and T lymphocytes thus contributing to the generation of a specific bronchial microenvironment that affects the way the body copes with microbes. Such a concept of organ-adopted immunity was recently proposed by Eyal Raz 9. Whereas several reports have now provided evidence for a role of epithelial cells in immunity in the gut, we demonstrate here that airway epithelial cells also actively engage in regulating immunity. This is of special importance as airway epithelium presents a huge entry portal for pathogens.

It is known that mucosal DC differ from their splenic counterparts by showing a predominant Th2-biasing phenotype and by stimulating B lymphocytes to mediate IgA switching 22, 30. We can show in this study that bronchial epithelial cells under homeostatic, non-inflammatory conditions release soluble factors that down-regulate TLR-mediated production of pro-inflammatory cytokines and costimulatory molecules. However, this lack of pro-inflammatory, possibly Th1-biasing, function does not represent a state of complete ‘anergy’ as at least some genes (IL-10, PD-L1, CD80) did not undergo inhibitory regulation. Instead, the increased expression of IL-10 and arginase-1 identifies these cells as having an alternatively activated phenotype. For epithelial cell-conditioned DC in the intestine similar findings of slightly up-regulated CD80, IL-10 and IL-6 have been shown, and this was interpreted as generation of a gut-specific anti-inflammatory environment 21, 31. Our findings clearly show that this mode of regulation is not specific to the gut but also can be observed at other mucosal surfaces.

Our results indicate that bronchial epithelial cells limit the pro-inflammatory capacity of innate immune cells within the lung and also induce a non-inflammatory microenvironment. In addition, we have shown that T lymphocyte activation is altered. Of note, this went along with an increased frequency of Tr1, IL-10-producing T cells as well as Foxp3+ regulatory T cells. These findings mirror results of others obtained for regulation of local immunity in the intestine. It has been reported that intestinal epithelial cells inhibit DC and reduce T cell activation 31, 32. Local epithelial cells modulated intestinal DC to be less inflammatory but more active in attracting Th2 cells 21. We now show that bronchial epithelial cells also regulate immune homeostasis.

Regarding the nature of inhibition, different modes have been suggested. It was recently reported that TGF-β bound by epithelial integrin mediated tonic inhibition of alveolar macrophages in a cell contact-dependent manner 17. However, soluble mediators including IL-10, TSLP, TGF-β, NO and prostaglandins have also been implicated in the generation of an organ-specific microenvironment 18. Also, the anti-microbial peptide LL-37 has been reported to alter DC phenotypes in the lung 33. In our experiments we observe that soluble mediators can transfer the inhibitory potential of epithelial cells; however, cell contact-dependent mechanisms might also been operative as suggested for intestinal epithelial cells and DC 31. With regard to cell contact-dependent mechanisms, it has been described recently that respiratory epithelial cells express PD-L1 and PD-L2 34, which are associated with negative regulatory activity. Our data confirm these findings in primary tracheal epithelial cells that were used in this study (data not shown).

TGF-β played a role in T cell inhibition, yet this cytokine was dispensable for the effects on monocytes and DC. We did not find a role for IL-10, TSLP or NO, which were not sufficiently expressed in bronchial epithelial cells. It is possible, however, that multiple factors (including cell-bound and soluble mediators) cooperatively mediate homeostatic inhibition. Support for this notion comes from preliminary experiments indicating that heat-labile factors account for the observed effects, and size-exclusion chromatography identifies at least four different fractions with inhibitory potential (data not shown).

So far, our results clearly show that bronchial epithelial cells constitutively produce mediators that modify the functional status of APC and T lymphocytes. We speculate that, especially within the lung, pro-inflammatory responses have to be tightly controlled to protect the easily damageable lung tissue from destructive side effects associated with inflammation during microbial contact. An interesting question would be whether, in a physiological setting, low-dose TLR stimulation contributes to such a homeostatic circuit. Indeed, it has recently been shown that epithelium integrity is dependent on intact MyD88 signaling in lung as well as intestinal non-infectious disease models 11, 35, 36.

A still more intriguing question is whether under infectious conditions epithelial cells release professional immune cells from inhibition. Corroborating such a hypothesis, we here observed a slightly down-regulated secretion of TGF-β in epithelial cells stimulated by Pseudomonas aeruginosa. Moreover, in a model of alveolar macrophage homeostasis, it was shown that a transient release of inhibition could be observed upon inflammation, which was due to loss of intimate cell contact 17.

In our setting it may be that TGF-β under homeostatic conditions not only inhibits T cell proliferation but also actively mediates induction of adaptive regulatory T cells, here exemplified by induction of Tr1 and Foxp3+ T cells. Infectious microorganisms can activate IL-6 production within epithelial cells by stimulating TLR 6, which together with TGF-β might switch T helper cell differentiation into the Th17 direction 37. Indeed, it was shown that IL-6 together with IL-1 reverted regulatory T cell generation 38. Another candidate could be IL-23 39 derived from epithelial cell-conditioned DC. Thus, airway epithelial cells would actively shape local immune responses. It is obvious that such an understanding ascribes epithelial cells a much more active function in regulating immunity. In turn this will impact our understanding of immune disorders including asthma, and possibly will pave the road for the development of new therapeutic approaches.

Taken together our results indicate that bronchial epithelial cells constitutively inhibit pro-inflammatory activation of professional immune cells and modulate T cell activation. Airway epithelial cells thereby contribute to the generation of an organ-specific microenvironment, which adapts local immunity to the necessities of epithelial surfaces.

Materials and methods

Reagents and antibodies

RPMI 1640 and sodium pyruvate were obtained from Biochrom (Berlin, Germany). FBS was from Biowest (Nuaillé, France), PBS, penicillin and streptomycin were obtained from PAA (Coelbe, Germany). Dulbecco's modified Eagle's Medium (DMEM), Ham's F12 medium, human recombinant insulin and CFDA-SE/CFSE were purchased from Invitrogen (Karlsruhe, Germany). Ultroser® G serum substitute was obtained from Pall Biosepra (Cergy-Saint Christophe, France). Highly purified lipoteichoic acid (LTA) from Staphylococcus aureus was a kind gift from C. Hermann (Konstanz, Germany). LPS from Salmonella minnesota was kindly provided by U. Seydel (Borstel, Germany). Poly(dI-dC), human type VI placental collagen, protease from Streptomyces griseus and FeN3O9 were purchased from Sigma-Aldrich (Schnelldorf, Germany). Pam3CSK4 was received from EMC Microcollections (Tübingen, Germany). Human recombinant M-CSF, GM-CSF and IL-4 were obtained from Tebu (Frankfurt, Germany). DNase I was from Roche (Mannheim, Germany). Human recombinant TGF-β1, monoclonal anti-TGF-β1, -β2, -β3 antibody and mouse IgG1 isotype control were purchased from R&D Systems (Minneapolis, MN, USA). Pseudomonas aeruginosa (ATCC 27853) was received from S. Zimmermann (Heidelberg, Germany).

Isolation and culture of primary murine tracheal epithelial cells

The isolation and culture of tracheal epithelial cells was performed as described 20. In brief, tracheae of 4–10-week-old C57BL/6 mice were excised and surrounding tissue was removed. The trachea were cut open lengthwise, washed in pre-warmed PBS and transferred to collection medium (1:1 mix of DMEM:Ham's F12; penicillin G (100 IU/mL) and streptomycin sulfate (100 IU/mL)). Eight trachea were transferred into pre-warmed dissociation medium (Ca2+/Mg2+-free PBS; 44 mM NaHCO3; 0.25 µM sodium pyruvate, 60 IU/mL penicillin G and streptomycin sulfate; 0.1 mg/mL DNase I and 1.4 mg/mL pronase E were added shortly before use) and incubated at 37°C for 1 h. Enzymatic digestion was stopped by adding 5 mL FBS with ten gentle inversion of the tube. The tracheal residues were transferred to 10 mL culture medium (collection medium supplemented with 5% FBS and 120 IU/L insulin). The cell suspension was centrifuged at 200 × g for 10 min, resuspended in 5 mL culture medium and seeded into 100-mm culture dishes for 2 h at 37°C to separate epithelial cells from contaminating cells. The unattached cells were collected, centrifuged and resuspended in 800 µL culture medium; 200 µL cell suspension was seeded into an insert of a transwell (Costar Transwell clear; 6.5-mm diameter inserts; 0.4 µm pore; Costar Corning, Corning, NY). Transwells were precoated the day before with 100 µL type VI acid-soluble human placental collagen (0.5 mg/mL collagen in distilled water with 0.2% glacial acetic acid), air-dried overnight and washed twice with PBS before use. The cells were cultured at 37°C in a humidified atmosphere in the presence of 5% CO2 for 3–5 days until confluence was reached. Then medium was removed and replaced in the lower compartment with 600 µL Ultroser G medium (collection medium supplemented with 2% Ultroser G serum substitute). The apical surface appeared dry when cells formed a confluent cell layer. Cells were checked by measuring transepithelial resistance and immunostaining for absence of contaminating fibroblasts (<5%).

Cell culture and ECCM

Human bronchial epithelial cell lines BEAS-2B and IHAEo were maintained essentially as described previously 6. For preparing ECCM, 2.5 × 104 cells were seeded in 96-well plates and cultivated for 48 h prior of harvesting the supernatant. For non-epithelial cell supernatant 6 × 104 HEK293 or 1 × 105 PBMC were treated in a similar way. Primary murine tracheal epithelial cells were cultivated for 20–30 days as air-liquid interface culture prior to collection of 48-h-old cell-free conditioned medium.

Isolation of CD14+ cells, generation of human macrophages and DC and isolation of T lymphocytes

Human PBMC were isolated from heparinized blood of healthy donors by standard Ficoll-Paque density gradient centrifugation. CD14+ or CD3+ cells were positively selected by magnetic-associated cell sorting (AutoMACS, program: possel; Miltenyi Biotec, Bergisch-Gladbach, Germany). For generation of human macrophages, CD14+ cells were differentiated by adding 5 ng/mL M-CSF for 8 days. DC were prepared by seeding CD14+ cells at a density of 2 × 106 cells/mL in 24-well, flat-bottom cell culture plates. Cultures were supplemented with 10 ng/mL recombinant human (rh) GM-CSF and 500 IU/mL rhIL-4. After 6 days, cells were harvested by incubation on ice for 30 min.

Generation of primary murine DC

BMDC were generated from female, 4–10-week-old C57BL/6 mice. Briefly, BM cells were placed in 70-cm2 tissue culture flasks in differentiation medium (RPMI 1640 medium supplemented with 10% FBS, 50 mM 2-mercaptoethanol, antibiotics and 200 U/mL GM-CSF). After 24 h, 107 nonadherent cells were re-seeded into 175-cm2 tissue culture flasks in differentiation medium. On day 5, fresh differentiation medium was added and on day 9 nonadherent, immature DC (CD11c+) were harvested. Culture supernatant of a GM-CSF-transfected cell line was also used as a source of GM-CSF.

Transwell experiments

BEAS-2B cells (105) were seeded in 24-well cell culture plates overnight. The next day 2 × 105 human monocytes were seeded into the transwell inserts and cells were stimulated with 10 ng/mL LPS overnight. Cell-free supernatant was analyzed for TNF-α secretion.

Mixed lymphocyte reaction and T cell proliferation assay

Human mixed lymphocyte reactions (MLR) were performed in an allogenic setting. Highly purified CD3+ T lymphocytes (105) were incubated with 104 APC (DC) from a different donor for 96 h with or without ECCM. For APC-independent activation, 2 × 105 T lymphocytes were incubated in 96-well plates in the presence of 1 µL anti-Biotin MACSiBead Particles loaded with antibodies against human CD2, CD3 and CD28 (human T Cell Activation/Expansion Kit, Miltenyi Biotec). During the last 16 h cultures were labeled with 22.5 kBq [3H]thymidine to measure cell proliferation. Alternatively, T lymphocytes were labeled with 1 µM CFSE and proliferation of T lymphocytes was analyzed by flow cytometry.

Preparation and activation of murine T cells

Spleens of C57BL/6 mice were minced, put through a 100-µm nylon mesh and subjected to density separation using a 1.090 g/mL Biocoll solution (Biochrom). T lymphocytes were prepared by MACS using negative selection (pan T cell isolation kit, Miltenyi Biotec). T cells (6 × 105) were activated by addition of anti-CD3/anti-CD28-coated beads (Dynabeads mouse T cell expander, Invitrogen) (1:2 bead/cell ratio). Where indicated, 20 ng/mL IL-6 and 5 ng/mL TGF-β1 (R&D Systems) were supplemented. After 3 days of incubation, cells were restimulated with 50 ng/mL PMA and 800 ng/mL ionomycin (Sigma) for 5 h prior to intracellular cytokine staining for IL-17 and IL-10 (BD Biosciences) employing a standard protocol. Also, Foxp3 (eBioscience) and CD25 were determined by flow cytometry.

Quantitative RT-PCR

Total RNA from 2 × 105 BEAS-2B or 8 × 105 BMDC was isolated using HighPureTM RNA kit (Roche), which included DNase I digestion. Total RNA (1 µg) was reverse-transcribed with a cDNA synthesis kit (MBI Fermentas, St. Leon-Rot, Germany), diluted 1:5 and used as template in the quantitative PCR mix (Absolute SyBR Green Rox Mix, Abgene, Epson, UK). PCR was performed on a 7900HAT platform (Applied Biosystems, Weiterstadt, Germany) using specific primers. All primer sequences are available on request. Quantifications were made using SYBR Green. Specificity of RT-PCR was controlled using either no template or no RT and melting curve analysis. Results are expressed relative to the expression of the housekeeping gene β-actin.

Flow cytometry of surface markers

BMDC (8x105) were pre-incubated with or without ECCM for 3 h and stimulated with 100 ng/mL LPS overnight. Cells were harvested and stained with antibodies against CD40 (HM 40–3), CD80 (16–10A1), CD86 (GL1), MHC class II (2G9), PD-L1 (MIH5) and PD-L2 (TY25) (BD Pharmingen, Heidelberg, Germany) and cells were analyzed on a FACS Canto (BD Biosciences).

Determination of cytokine secretion

Cell-free supernatants were harvested and analyzed for cytokines by commercially available ELISA kits (OptEIA; BD Pharmingen). TGF-β ELISA was purchased from R&D Systems (DuoSet®; Minneapolis, MN).


Data were analyzed by GraphPad Prism 4.03 program (GraphPad Software, San Diego, CA). Significant differences were assessed by analysis of variance (ANOVA) to compare three or more groups followed by Dunnett's test to compare selected groups. In all figures * represent p values <0.05. All experiments were performed at least three times if not indicated otherwise.


This work was supported by grants of the German Research Foundation to A.D. (Da592/1, SFB405-B18). We appreciate the excellent technical help from Adelina Dillmann, Aline Gierschke and Mario Muehmer.


The authors declare no financial of commercial conflict of interest.


  1. 1


  2. 2


  3. 3


  4. 4


  5. 5