Comparative metabolomics analysis of bronchial epithelium during barrier establishment after allergen exposure

Abstract Background Several studies have shown a correlation between an altered metabolome and respiratory allergies. The epithelial barrier hypothesis proposes that an epithelial barrier dysfunction can result in allergic diseases development. Der p 1 allergen from house dust mite is a renowned epithelial barrier disruptor and allergy initiator due to its cysteine‐protease activity. Here, we compared the metabolic profile of the bronchial epithelium exposed or not to Der p 1 during barrier establishment to understand its active role in allergy development. Methods Calu‐3 cells were cultivated in air‐liquid interface cultures and exposed to either Der p 1 or Ole e 1 allergens during barrier establishment. The comparative metabolomics analysis of apical and basolateral media were performed using liquid chromatography and capillary electrophoresis both coupled to mass spectrometry. Results We showed that epithelial barrier disruption by Der p 1 was associated with a specific metabolic profile, which was highly dependent on the state of the epithelium at the time of contact. Moreover, an apical‐basolateral distribution of the metabolites was also observed, indicating a compartmentalization of the response with differential metabolic patterns. A number of metabolites were changed by Der p 1, mainly related to amino acids metabolism, such as L‐arginine, L‐kynurenine and L‐methionine. Conclusion This work is the first report on the metabolic response in human bronchial epithelial cells associated with cysteine‐protease Der p 1 activity, which could contribute to allergy development. Moreover, it supports a reformulated epithelial barrier hypothesis that might help to explain allergies and their increasing prevalence.


| INTRODUCTION
During the last years, the development of inflammatory diseases such as respiratory allergies has been correlated with changes in particular metabolites, either generated by the oxidative metabolism or in response to exposome components, such as allergens. [1][2][3][4][5] The complete analysis of metabolites (metabolomics) in a biofluid has the potential to provide information about alterations in metabolic pathways that underlie a disease. 6,7 In addition, this technology performs a rapid and accurate analysis of a wide range of low-molecular mass molecules, which can enable to uncover new biomarkers and targets that may be applied in diagnosis, prognosis and disease treatment. To date, metabolomics based on nuclear magnetic resonance spectroscopy as well as liquid or gas chromatography coupled to mass spectrometry have been applied to numerous clinical disorders, including allergy, cancer and Alzheimer's disease. [8][9][10] Bronchial epithelial cells play a key role in the orchestration of lung immune response by secreting a wide variety of mediators, including cytokines and chemokines, among others. 11 The function of the epithelial barrier is highly dependent on the formation of apical junctional complexes between adjacent cells, which are formed by adherens and tight junctions (TJs). Allergic diseases are characterized by a disrupted epithelial barrier. 12 The 'epithelial barrier hypothesis' proposes that an epithelial barrier dysfunction can be a pivotal first-step towards the development of allergy and others diseases. 13 Der p 1, the main allergen of house dust mite (HDM), is a well-recognized epithelial barrier disruptor, 14 as well as an 'initiator allergen', 15 because of its cysteine-protease activity which can promote an allergic response to unrelated allergens. Therefore, a comparative metabolomics analysis of the bronchial epithelium exposure to Der p 1 during barrier establishment will be key to study the link between defective airway epithelial barrier and allergy development, and understand its role in this disease.
In this work, air-liquid interface (ALI)-cultured Calu-3 cells were exposed to a Der p 1 or Ole e 1, an olive pollen allergen without protease activity, 16 during barrier establishment.
Comparative metabolomics analysis was performed using multiplatform analyses. To our knowledge, this is the first comparative metabolome study reported on the response of human bronchial epithelium cells to allergen exposure. Our data showed that the disruption of epithelial barrier by Der p 1 was associated with a specific metabolome profile, highly dependent on the epithelial barrier establishment at the time of exposure. In addition, an apical-basolateral distribution of the metabolites was also detected. Finally, the metabolic pathways of particular metabolites that could be involved in the allergic response were also discussed. Overall, these data support the epithelial barrier hypothesis, which proposes that a dysfunctional barrier may underlie allergic diseases development and explain their increasing prevalence.

| Proteins
Natural Der p 1 (LTN-DP1-1, Lot. 38190) was supplied from Indoor Biotech as endotoxin-free protein (≤ 0.03 EU/μg). Der p 1 was activated by incubation with 0.1 mM reduced glutathione (GSH; Sigma-Aldrich) in phosphate buffered saline (PBS) at 37°C for 15 min, and its cysteine-protease activity was checked as described. 17 Olive pollen (IBERPOLEN SL) was used to isolate and purify Ole e 1. 16  ALI-cultured cells were exposed apically to the allergen in 0.1 ml PBS containing 0.1 mM GSH on days 2 and 7: Ole e 1 (25 μg/ml) or Der p 1 (10 μg/ml). To the basolateral side, 0.5 ml of complete DMEM-phenol free-medium (Thermo Fisher Scientific) with 5% FBS were added. After 24 h, the apical and basolateral media were collected, and kept at −80°C until the metabolomics analysis ( Figure 1A). Cells exposed only to PBS/GSH were used as control.

F I G U R E 1 Characterization of air-liquid interface (ALI)-cultured Calu-3 cells. (A) Schematic diagram of experiential protocol. Calu-3 cells
were cultured under ALI conditions and exposed to Der p 1 or Ole e 1 for 24 h during barrier establishment. Cells exposed to PBS/GSH were used as controls. Apical and basolateral media were collected, and metabolic profile was determined using liquid chromatography Although some Q 2 values were low, the separation between groups observed in the obtained PLS-DA models suggest that ALI-cultured Calu-3 cells exposed to Der p 1 exhibited a distinct pattern in the metabolic profile on the apical compartment. Nevertheless, for most of the basolateral dataset, no PLS-DA models were obtained, indicating that there were no notable significant differences between groups with the type of exposure at different times of barrier establishment ( Figure 3). Only one model was built for Der p 1exposed group versus control on 2 days using CE-MS platform, and this suggest that the differences in metabolic profile between both groups could be mainly due to the amino acids as this technique allows their detection. And, a single PLS-DA model was obtained for Ole e 1 versus control on day 7 using LC-MS−. On the basis of these data, the apical and basolateral compartments exhibited different metabolic profiles, mainly, in Der p 1-exposed cells.

| Annotation and identification of metabolites released by ALI-cultured Calu-3 cells after allergen exposure during barrier establishment
Univariate analysis was performed to define which metabolites were differentially released among the three groups. A total of 292 and  Figure 5). Interestingly, metabolites that were increased on day 7, tended to be decreased on day 2, and vice versa (being more evident on the apical compartment). Significant features were annotated, as illustrated in Tables S1 and S2, amino acids such as L-alanine, L-isoleucine or L-serine were increased between 7 and 10 fold on the apical compartment of Der p 1 exposed cells on day 7 compared to day 2.
After annotation, a total of 67 and 20 extracellular metabolites were obtained for the apical and basolateral compartments, respectively (Table S1- Interestingly, metabolites related to lipid metabolism, including sphingolipids, glycerophospholipids, inositol phosphates and fatty acid thioesters were decreased after Der p 1-exposure, both on days 2 and 7. Analysis of metabolic pathways altered in ALI-cultured Calu-3 cells after allergen exposure was performed using IMPaLA (Table S5). Pathway analysis showed that most of the significantly changed pathways after allergen exposure in comparison to control cells were related to amino acid metabolism and transport, including arginine metabolism, glycine, serine and threonine metabolism, phenylalanine and tyrosine metabolism and tryptophan metabolism.

F I G U R E 4
Heatmaps of significant metabolites from air-liquid interface-cultured Calu-3 cells exposed to Der p 1 or Ole e 1 allergens in comparison to control cells, on days 2 and 7 during barrier establishment. Significantly differed metabolites were selected using Mann-Whitney U test (p < 0.05), and hierarchical clustered using Ward's algorithm, with dendrograms to represent the distance between samples: apical data (n = 18, metabolites n = 409); and basolateral data (n = 18, metabolites n = 69). Colours represent the levels of metabolites (rows) from the biological samples (columns): red, high levels and blue, low levels . 24 Similarly, a differentiation state-dependent of cytokine release by ALI-cultured NHBE cells in response to Ole e 1 has also been described. 25 Moreover, the multivariate analysis of the samples revealed further differences between the epithelial cell response to Ole e 1 and Der p 1 allergens. Exposure to Der p 1 significantly altered the metabolic profile of the cells but no significant differences were detected between metabolic profiles of Ole e 1-exposed cells compared to control cells. These differences could be partly explained by the cysteineprotease activity of Der p 1 that previous studies have shown to contribute to the allergic response by different mechanisms. 14  Methionine plays an important role in the synthesis of proteins involved in the immune response. This has been identified as a biomarker of asthma, along with glutamine and histidine, 29 and all three altered at seizure stage of pollinic patients. 30 On the other hand, L-tryptophan, L-arginine and L-kynurenine were significantly increased on the apical compartment of ALIcultured Calu-3 cells after Der p 1 exposure but not on the basolateral one, compared to both, control and Ole e 1-exposed cells.
These metabolites have been associated to allergy. 31,32 The essential amino acid L-tryptophan is breakdown by the indoleamine 2,3-dioxygenase (IDO) enzyme in N-formyl-kynurenine via the kynurenine pathway, which also include the metabolite kynurenine. 33 IDO-1 is the predominant isoform expressed in the lung by epithelial cells among others. 34 This enzyme has been involved in allergic inflammation by controlling local levels of tryptophan. [35][36][37][38] Defective IDO activity has been reported in rhinitis allergic patient and in airway epithelial primary cultures and cell lines after HDM exposure. 39,40 Moreover, high tryptophan levels have been described in serum of pollinic patients 35 and linked to unresponsiveness to immunotherapy in different animal models. [41][42][43][44] We found a decreased kynurenine/ tryptophan ratio on the apical compartment of ALI-cultured cells exposed to Der p 1 on day 2, when the barrier is not yet established.
Interestingly, this ratio was not altered on day 7, when the barrier is established. It has been shown that kynurenine contributes to tolerance induction against allergens and has been associated with low activity of IDO. 41,45,46 Overall, our data support the notion that exposure to Der p 1 induces an inflammatory response on the apical compartment of ALI-cultured Calu-3 cells on day 2. L-arginine plays a key role in the regulation of airway function via the production of the bronchodilating nitric oxide (NO) by the nitric oxide synthase (NOS).
Both arginine and NO have been extensively reported to be increased in asthma. [47][48][49] However, controversy exits about their role on airway responsiveness and remodelling. 50 Thus, an increased NOS activity has been associated to a pro-inflammatory state, whereas an up-regulation of arginase activity has been linked to an antiinflammatory state and inhibition of Th2 response. 50 The lung is a metabolically highly active organ whose glucose consumption exceeds that of other organs. 57 Our data suggested that ALI-cultured Calu-3 cells shift their metabolism to aerobic glycolysis or Warburg metabolism for ATP production after exposure to Der p 1 on day 2, as indicated by the accumulation of pantothenic acid on the apical compartment. The pantothenic acid is precursor of the synthesis of coenzyme A, which is essential in the metabolism of fatty acids and the citric acid cycle. The low levels of pyridoxal found on the basolateral media after Der p 1 treatment supports the notion that the epithelial cells are using aerobic glycolysis as pyridoxal liberates glucose monomers form glycogen (glycogenolysis). 54 Warburg metabolism is switched on in response to cellular activation in numerous cell types and in highly proliferative cells, such as T cells and macrophages. 58,59 However, further studies are needed to understand its role in the immature epithelial barrier.
To our current knowledge, this is the first time a study has reported on the metabolic response of human bronchial epithelial cells to allergen exposure. Together, our data proved that the disruption of epithelial barrier by the cysteine-protease Der p 1 activity was associated with a specific metabolic profile, which could contribute to allergy development. Moreover, this metabolic response was highly dependent on the epithelial barrier establishment at the time of allergen exposure. The metabolic changes may provide a key to uncovering the role of the epithelial barrier in the development of allergic diseases at the molecular level. Finally, these data support a reformulated epithelial barrier hypothesis, which postulates that a dysfunctional barrier can underlie allergic diseases development and explain their increasing prevalence.

ACKOWLEDGEMENTS
We would like to thank to Inés Judel and Laura Vaca for their work in the metabolomics analysis measuring the samples and their contribution to the data treatment. This work was supported by the