Respiratory Epithelial Cells Can Remember Infection: A Proof-of-Concept Study

Abstract

Human bronchial epithelial cells play a key role in airway immune homeostasis. We hypothesized that these sentinel cells can remember a previous contact with pathogen compounds and respond nonspecifically to reinfection, a phenomenon called innate immune memory. We demonstrated that their preexposure to Pseudomonas aeruginosa flagellin modify their inflammatory response to a second, nonrelated stimulus, including live pathogens or lipopolysaccharide. Using histone acetyltransferase and methyltransferase inhibitors, we showed that this phenomenon relied on epigenetic regulation. This report is a major breakthrough in the field of multimicrobial respiratory tract infections, wherein control of inflammatory exacerbations is a major therapeutic issue.

trained immunity, bronchial epithelial cells, infection, inflammation, innate immune memory

The human immune system includes the innate and adaptive immune responses. Adaptive immunity results from pathogen-specific immunological memory, whereas innate immunity, the first line of defense against pathogens, is nonspecific, without memory of the first contact [1].

Innate immune cells, originating from hematopoietic progenitors are capable of modulating their response on reinfection, a phenomenon called trained immunity or innate immune memory [2]. This memory is initiated by interactions between pathogen-associated molecular patterns (PAMPs) and pattern recognition receptors. A functional reprogramming that occurs through metabolic/epigenetic regulation leads the cells to become hyperresponsive (training) or hyporesponsive (tolerance) [3].

To date, innate immune memory has not been reported for bronchial epithelial cells (BECs), which are frequently in contact with inhaled microorganisms. Respiratory epithelial cells are nonhematopoietic cells playing a crucial role in orchestrating the host immune response [4]. Moreover, BECs are sensitive to epigenetic modifications in clinical contexts such as asthma [5, 6] and viral infections [7].

Sequential pulmonary infections are common in clinical contexts. For instance, bronchi in patients with cystic fibrosis are commonly chronologically colonized by Staphylococcus aureus, Pseudomonas aeruginosa, and then Aspergillus fumigatus [8]. These infections and the associated inflammation result in a progressive decrease in the respiratory function. In the current study, we investigated the potential immune memory of BECs with the hypothesis that initial exposure to PAMPs could modify their subsequent response to infection.

METHODS

Culture of BECs and Pathogens

Human BECs (BEAS-2B; American Type Culture Collection) were maintained at 37°C and 5% carbon dioxide in F-12 medium (Invitrogen) supplemented with 10% fetal calf serum, 1% penicillin-streptomycin, and 10-mmol/L HEPES and stimulated in F-12 without antibiotics. Primary human BECs (Epithelix SaRL) were isolated from a 72-year-old man and a 72-year-old woman; both were white and nonsmokers, without reported disease. The BECs were cultured according to the manufacturer’s instructions.

The A. fumigatus DAL strain (CBS 144.89) was grown as described elsewhere [9]. The conidia concentration was adjusted to obtain a multiplicity of infection of 2. The P. aeruginosa PAK strain and a Stenotrophomonas maltophilia clinical strain were grown as described elsewhere [10]. The concentrations were adjusted to obtain a multiplicity of infection of 0.25.

Stimulation of BECs

A flow diagram of the BEC stimulation is shown in Figure 1A. On day −1, cells were cultured in 96-well plates (1.5 × 10⁴ cells per well) and then prestimulated on day 0 (mean [standard error of the mean (SEM)], 2.8 [0.5] × 10⁴ cells per well) with ultrapure flagellin (InvivoGen) from P. aeruginosa (10⁻⁶ to 5 µg/mL) or Pam3CSK4 (10⁻⁶ to 1 µg/mL), a synthetic triacylated lipopeptide (InvivoGen), for 48 hours. The prestimulated supernatants were collected, and the cells were washed twice with medium, followed by a resting period of 4 days. On day 6, cells reaching confluence (10.3 × 10⁴ cells per well) were stimulated or not (control) with 10 ng/mL Escherichia coli lipopolysaccharide (LPS; Sigma-Aldrich), live A. fumigatus conidia or S. maltophilia for 15 hours, or P. aeruginosa for 4 hours.

Figure 1.

Protocol and cytokine secretion by bronchial epithelial cells (BECs) during trained immunity. A, Protocol used to study the memory effect in BECs. B, Interleukin 8 (IL-8) and interleukin 6 (IL-6) secretion after prestimulation of BEAS-2B cells with different concentrations of flagellin and stimulation with Aspergillus fumigatus conidia (10⁶/mL), Escherichia coli lipopolysaccharide (LPS) (10 ng/mL), Pseudomonas aeruginosa (1.3 × 10⁵ cells/mL), or Stenotrophomonas maltophilia (1.3 × 10⁵ cells/mL) on day 6. C, IL-8 secretion after prestimulation of primary human BECs isolated from donor 1 (72-year-old man) and donor 2 (71-year-old woman), with different concentrations of flagellin and infection with conidia of A. fumigatus (10⁶/mL) on day 6. Data are presented as means with standard errors of the mean for 3–8 independent experiments performed in triplicate. Statistical analysis was performed using analysis of variance, followed by Bonferroni multiple comparison test. *P < .05; †P < .01; ‡P < .001 (vs control [nonprestimulated cells]). Abbreviations: PAMPs, pathogen-associated molecular patterns; FCS, fetal calf serum.

Pharmacological inhibitors of histone acetyltransferase, epigallocatechin-3-gallate (EGCG), or histone methyltransferase, BIX01294 trihydrochloride-hydrate (BIX; Sigma-Aldrich), were used to investigate the role of epigenetic modifications. Cells were preincubated for 1 hour with 15-µmol/L EGCG or 4-µmol/L BIX, before prestimulation. Subsequently, the inhibitors were maintained during the prestimulation and resting phases, and removed when cells were stimulated on day 6. The supernatants were collected on day 7 and stored at −20°C until interleukin 8 (IL-8) and interleukin 6 (IL-6) quantification with enzyme-linked immunosorbent assay (R&D Systems).

Measurement of Toll-Like Receptor 4 Gene Expression

Toll-like receptor 4 gene (TLR4) messenger RNA expression was quantified by means of quantitative reverse-transcription polymerase chain reaction (TaqMan probe Hs_00152939) on day 6 in cells prestimulated with flagellin (1 µg/mL) and in nonstimulated cells. The expression level of TLR4 was normalized to that of glyceraldehyde-3-phosphate dehydrogenase (TaqMan probe Hs_2786624), with control cells used as a calibrator, and relative expression was calculated using the 2^−ΔΔCt method (where Ct indicates cycle threshold) [10].

Statistical Analysis

Data are presented as means with standard errors of the mean. Differences among groups were assessed using Prism 7.00 software (GraphPad). Differences with were considered statistically significant at P < .05.

RESULTS

To test the hypothesis of an immune memory in BECs, we adapted to the BEAS-2B cell line, the protocol developed by Ifrim et al [3] to study monocytes innate memory (Figure 1A). To mimic the first infection with P. aeruginosa or S. aureus, we prestimulated BECs with flagellin or Pam3CSK4 and determined whether their inflammatory response was modified when the cells were subsequently stimulated with LPS, A. fumigatus, P. aeruginosa, or S. maltophilia. The most striking results were obtained after prestimulation with flagellin, a TLR5 ligand (Figure 1B). Indeed, prestimulation with 5-µg/mL flagellin led to a trained immune response, compared with nonprestimulated cells, with increased secretion of IL-8 (mean [SEM] 3253 [574] vs 1663 [107] pg/mL, respectively) and IL-6 (275 [38] vs 148 [13] pg/mL) when A. fumigatus conidia were used as a second stimulus. In contrast, 1-µg/mL flagellin induced a tolerant response, compared with nonprestimulated cells, with decreased secretion of IL-8 (mean [SEM], 66 [24] vs 250 [49] pg/mL, respectively) and IL-6 (9 [2] vs 24 [5] pg/mL, respectively) when the cells were subsequently challenged with LPS.

Similar decreased responses were obtained when BECs were infected with 2 strains of live gram-negative bacteria, P. aeruginosa and S. maltophilia. Indeed, we observed that 5-µg/mL flagellin induced a tolerant response, compared with nonprestimulated cells, with decreased secretion of IL-8 (mean [SEM], 145 [14] vs 1279 [235] pg/mL and 341 [62] vs 961 [56] pg/mL) and IL-6 (26 [13] vs 237 [41] pg/mL and 46 [3] vs 167 [1] pg/mL) when the cells were subsequently infected with P. aeruginosa and S. maltophilia, respectively. This regulation seemed to be linked to the flagellin signaling pathway, because prestimulation with Pam3CSK4, a TLR2 ligand, did not modify IL-8 and IL-6 secretion by cells subsequently challenged with LPS, flagellin, or A. fumigatus (data not shown). Importantly, using primary BECs from human donors, we confirmed the modulation of the inflammatory response after flagellin prestimulation and subsequent A. fumigatus infection, with a trend for a tolerant (not significant) and a trained immune response observed for the BECs from donor 1 and donor 2, respectively (Figure 1C).

To assess whether the level of cytokine secretion induced by the first stimulation influences the secondary response, we compared IL-8 and IL-6 concentrations in the cell supernatants after 48 hours of prestimulation with PAMPs. We found that 1- and 5-µg/mL flagellin triggered similar levels of IL-8 (mean [SEM], 854 [114] and 817 [97] pg/mL, respectively, vs 227 [54] pg/mL for nonstimulated cells) and IL-6 (67 [10] and 84 [20] pg/mL, respectively, vs 20 [3] pg/mL for nonstimulated cells), whereas these 2 concentrations could induce either a trained or a tolerant immune response, depending on the second stimulus. Furthermore, we observed that 1-µg/mL Pam3CSK4, which did not lead to any modification in the secondary response, induced IL-8 and IL-6 secretion at levels (mean [SEM], 791 [48] and 90 [11] pg/mL, respectively) similar to those induced with 1- or 5-µg/mL flagellin (Supplementary Figure 1A and 1B). These results suggest that the innate immune memory would not depend on the level of the inflammatory response of prestimulated cells but rather on the nature and concentration of PAMPs, by way of pattern recognition receptors engaged during the first contact.

To test the role of epigenetic mechanisms, we inhibited enzymes involved in the chromatin condensation or decondensation. Inhibition of histone acetyltransferase with EGCG significantly diminished the flagellin-induced IL-8–trained immune response to A. fumigatus (mean [SEM], 1229 [120] pg/mL with EGCG vs 1747 [170] pg/mL without EGCG), without modifying the IL-8 response to A. fumigatus in nonprestimulated cells (845 [73] pg/mL with vs 1024 [93] pg/mL without EGCG) (Figure 2A). Similarly, inhibition of histone 3 at lysine 9 (H3K9) methylation with BIX diminished the flagellin-induced IL-8–trained immune response to A. fumigatus (mean [SEM], 742 [38] pg/mL with BIX vs 1236 [132] pg/mL without BIX), without affecting IL-8 synthesis in nonprestimulated cells in response to A. fumigatus infection (482 [16] pg/mL with vs 557 [31] pg/mL without BIX) (Figure 2B). In contrast, neither EGCG nor BIX treatment modified the IL-6–trained immune response (data not shown).

Figure 2.

Involvement of epigenetic mechanisms in innate immune memory in bronchial epithelial cells (BECs). A, B, Effects of epigenetic modulators on cytokine secretion. Cells incubated with (+) or without (−) 15-µmol/L epigallocatechin-3-gallate (EGCG) (A) or 4-µmol/L BIX (B) were prestimulated with 5-µg/mL Pseudomonas aeruginosa flagellin and infected with Aspergillus fumigatus conidia (10⁶/mL). Data are presented as means with standard errors of the mean for 3–7 independent experiments performed in duplicate. Statistical analysis was carried out using analysis of variance, followed by Bonferroni multiple comparison test. *P < .05; †P < .01; ‡P < .001 (vs control [nonprestimulated cells]). C, Expression of TLR4 messenger RNA on day 6 after prestimulation with flagellin (1 µg/mL). Data are presented as means with standard errors of the mean for 9 independent experiments. Statistical analysis was conducted using the Mann-Whitney test. *P < .05 (prestimulated vs nonprestimulated group). Abbreviations: AU, Arbitrary unit; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IL-8, interleukin 8; TLR4, toll-like receptor 4.

Altogether, our data suggest the involvement of epigenetic mechanisms in the reprogramming of the transcriptional response in BECs. This conclusion was further supported by a significant decrease (26%) in the expression of TLR4, an LPS receptor, in BECs preincubated with 1-µg/mL flagellin, compared with that in nonprestimulated cells (Figure 2C).

DISCUSSION

Innate immune cells, originating from hematopoietic progenitors such as macrophages/monocytes and natural killer cells, are capable of immune memory [2]. This phenomenon has also been described in nonprofessional immune cells, such as epithelial stem cells, fibroblasts, and intestinal stromal cells, which, similarly to BECs, exhibit immune properties, express pattern recognition receptors, and promote cell replacement and tissue regeneration [11]. However, to our knowledge, this phenomenon has not been reported for BECs that are frequently in contact with inhaled environmental microorganisms.

In the current study, we demonstrated that BEAS-2B cells could be reprogrammed, like monocytes [3], after the first contact with flagellin and were able to reduce or exacerbate their inflammatory response to a second microbial stimulus. We further confirmed the memorization capacity of BECs using primary cells and demonstrated that the innate memory observed in the BEAS-2B cell line was not a consequence of its immortalization process. Interestingly, the modulation differed between the donors, with either a reduction or exacerbation of the inflammatory response that may be explained by their previous exposition to different pathogens during their lifetime. Furthermore, immune responses and epigenetic mechanisms are known to be influenced by sex [12], which differed between the 2 donors.

A limitation of our study is that PAMPs were used during preexposure, instead of living pathogens, which cause cellular toxicity after 48 hours of infection. In addition, other inflammatory molecules may be modulated on preexposure of BECs to PAMPs.

Thus, our data highlight the capability of BECs to develop innate immune memory. We also demonstrated the existence of epigenetic mechanisms involved in innate immune memory of BECs, similar to those reported elsewhere for macrophages/monocytes [13]. Indeed, by inhibiting histone acetylation or methylation, known to play a role in the condensation and decondensation of the chromatin, thus affecting the transcriptional machinery, we observed a dramatic decrease in the IL-8–trained memory of flagellin-prestimulated cells in response to A. fumigatus infection.

These results require further investigations to identify histone modifications involved. The expression of TLR4, an LPS receptor, was reduced in flagellin-prestimulated cells on day 6; this may explain, at least in part, the tolerance to subsequent stimulation with LPS or LPS-expressing bacteria. However, TLR4 expression was not completely abolished to fully account for the drastic decrease in cytokine levels. The expression of TLR4 coreceptors should be studied to further elucidate the mechanism. A similar approach cannot be used for A. fumigatus, because its receptor has not yet been identified in BECs.

There are some differences between the response of flagellin-prestimulated BECs, observed in our study and that of monocytes, described elsewhere. Indeed, Ifrim et al [3] demonstrated that the monocytes initial exposure to low concentrations of flagellin elicited an enhanced response upon secondary exposure to LPS, contrary to that of BECs. This is probably owing to the difference in TLR expression levels [14]. Moreover, it is important to note that our BEC model differs from monocytes/macrophages by their capacity to divide.

Therefore, whereas Ifrim et al [3] prestimulated and subsequently challenged the same monocyte population, in our model, BECs challenged on day 6 were daughter cells, derived from cells prestimulated on day 0. Thus, as epigenetic changes, the memory of BECs seems to be transmitted during cell division. This capacity for memory transfer to daughter cells may be of major importance in an inflammatory context, especially in cystic fibrosis, chronic obstructive pulmonary disease, or bronchiectasis, in which epithelial lesions, due to inflammatory exacerbations and infections, frequently lead to tissue repair and cell renewal [15].

In conclusion, we demonstrated that the inflammatory response of BECs could be reprogrammed after the first contact with a microbial ligand. Intracellular processes associated with this innate immune memory should be identified to assess various epigenetic and metabolic defects that may occur in immune-related diseases associated with chronic infections or inflammation.

Note

Financial support. This work was supported by a grant from a French nonprofit cystic fibrosis organization, Vaincre la Mucoviscidose.

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Author notes

Christophe Hennequin and Viviane Balloy contributed equally to this work.