K. I. L. Röschmann, Department of Otorhinolaryngology, Academic Medical Center, Room L3-106, Meibergdreef 9, 1105 AZ Amsterdam, the Netherlands. E-mail: email@example.com
By definition, allergens are proteins with the ability to elicit powerful T helper lymphocyte type 2 (Th2) responses, culminating in immunoglobulin (Ig)E antibody production. Why specific proteins cause aberrant immune responses has remained largely unanswered. Recent data suggest that there may be several molecular paths that may affect allergenicity of proteins. The focus of this study is the response of airway epithelium to a major allergen from Phleum pratense Phl p 1. Instead of focusing on a few genes and proteins that might be affected by the major allergen, our aim was to obtain a broader view on the immune stimulatory capacity of Phl p 1. We therefore performed detailed analysis on mRNA and protein level by using a microarray approach to define Phl p 1-induced gene expression. We found that this allergen induces modulation and release of a broad range of mediators, indicating it to be a powerful trigger of the immune system. We were able to show that genes belonging to the GO cluster ‘cell communication’ were among the most prominent functional groups, which is also reflected in cytokines and chemokines building centres in a computational model of direct gene interaction. Further detailed comparison of grass pollen extract (GPE)- and Phl p 1-induced gene expression might be beneficial with regard to the application of single components within diagnosis and immunotherapy.
Allergens are defined as proteins that have the ability to elicit powerful T helper lymphocyte type 2 (Th2) responses, culminating in immunoglobulin (Ig)E antibody production. Why specific proteins cause such an aberrant T and B cell response has remained largely unanswered. Recent data suggest that there may be several molecular paths that may affect allergenicity of proteins. These paths seem to depend on some intrinsic biological activity of the allergens, indicating that allergens can be more than passive players whose sole function is to cross-link IgE-molecules on the cell surface of mast cells.
Proteolytic activity as a general feature of major allergens has been proposed to be involved in the pathogenesis of allergies by enabling the passage of allergens through the epithelial barrier, cleaving adhesion molecules and affecting the functions of various cells and immune responses [1–4].
Recently it has become apparent that some allergen-binding lipids may contribute to the allergenicity of these proteins by affecting the response of cells to these lipids. For instance, Der p 2 was identified to resemble an important co-factor for Toll-like receptor (TLR)-4 signalling (MD-2), not only structurally but also functionally [5,6]. Several other members of the MD-2-like lipid-binding family are major allergens , referring to a more general importance of major allergens being able to bind lipids. Furthermore, a broad role for complex carbohydrates, in particular glucans, in allergen-associated Th2 responses is emerging .
However, the allergenicity of these proteins might be independent of their intrinsic biological function. Co-exposure of a tolerogenic protein with a protease can induce allergic sensitization so that environmental proteases derived from bacterial or viral origin might have accessory roles. Furthermore, microorganism-derived components, usually present in allergen extracts, could act as adjuvants with regard to the induction of immune responses.
Due to recent publications there is a growing awareness of the role of structural cells concerning allergen-induced immune responses. In particular, respiratory epithelium is likely to be important within the process of sensitization. It is accepted to be an active player in the response to the biological activity of aeroallergens, where they contribute to the immunological response by generating a microenvironment for the attraction and regulation of the activity of immune competent cells [9–12].
The focus of this study is the response of airway epithelium to the purified major allergen Phl p 1. Instead of focusing on a few genes and proteins that might be affected by the major allergen, our aim was to obtain a broader view on the immune-stimulatory capacity of Phl p 1. Furthermore, the detailed investigation of the gene expression and mediator release induced by the single major allergen might provide new insights into the contribution of the major allergen to the grass pollen extract (GPE)-induced response of airway epithelium. We therefore performed a detailed analysis on mRNA and protein level by using a microarray approach to define Phl p 1-induced gene expression to define the contribution of the purified major allergen to the induced allergic response.
NCI-H292 human airway epithelial cells (American Type Culture Collection, Manassas, VA, USA) were cultured in RPMI-1640 medium (Invitrogen, Breda, the Netherlands) supplemented with 1·25 mm of l-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin and 10% (v/v) fetal bovine serum (HyClone, Logan, UT, USA). Cells were grown in fully humidified air containing 5% CO2 at 37°C and were subcultured weekly.
NCI-H292 cells were cultured in six-well plates to 80% confluence. Before stimulation, culture medium was removed and the cells preincubated with Hanks's balanced salt solution (HBSS) for 24 h. Cells were then stimulated with grass pollen major allergen Phl p 1 (5 µm) diluted in HBSS or with HBSS alone; 24 h after stimulation supernatants were collected and stored for further analysis. The cells were used for RNA isolation.
Preparation and purification of Phl p 1
Timothy grass pollen major allergen Phl p 1 was purified from pollen (ARTU Biologicals, Lelystad, the Netherlands), as described previously [13,14].
Total RNA from each sample was extracted using Trizol (Life Technologies, Inc., Gaitersburg, MD, USA), according to manufacturer's instructions, followed by purification by nucleospin RNA II (Machery-Nagel, Düren, Germany). Concentration of extracted RNA was measured on the nanodrop ND-1000 (NanoDrop Technologies, Inc., Wilmington, DE, USA). RNA quality was checked by using Agilent 2100 bio-analyser (Agilent Technologies, Palo Alto, CA, USA).
Microarray Affymetrix U133 plus 2
Human Genome U133 Plus 2·0 Genechip Array (Affymetrix, Inc., Santa Clara, CA, USA) was used to analyse Phl p 1-induced gene expression. The technical handling was performed at the MicroArray Department at the University of Amsterdam (Amsterdam, the Netherlands), a fully licensed microarray technology centre for Affymetrix Genechip® platforms and official Dutch Affymetrix Service Provider, as described previously .
Microarray data analysis and statistics
Experiments were performed in triplicate and simultaneously. Gene chip images and data sets were uploaded into the server of the Netherlands Bioinformatics Center using the Rosetta Resolver Biosoftware package, which was also used for statistical analysis. Triplicates were compared in a factorial design using an error-weight one-way analysis of variance (anova). The quality of the images was checked by visual inspection and all raw data passed the quality criteria for average background, scale factors, percentage present calls, 3′/5′ ratios glyceraldehyde 3-phosphate dehydrogenase (GAPDH), 3′/5′ ratios beta-actin, hybridization spike-in controls and poly-A spike-in controls. The data also passed a set of quality control checks provided by the affy, affyPLM and affyQCreport packages from Bioconductor (http://www.bioconductor.org/). Expression values were calculated using the robust multi-array average (RMA) algorithm . Statistical analysis for differential gene expression was performed using anova (maanova package, version 0·98·8 ). The permutation-based Fs test was used for hypothesis testing  and all P-values were adjusted for false discovery rate correction . For gene ontology and network interaction analysis, GeneSpring GX11 (Agilent Technologies) was used. Correlations between observed fold changes within the microarray data and real-time PCR data were determined using Pearson's correlation.
Quantitative real-time PCR was performed with the Light Cycler 2·0 System (Roche Diagnostics, Mannheim, Germany) using LightCycler® Fast Start DNA MasterPLUS, according to the manufacturer's instructions. Relative cDNA amount was calculated compared to the expression of the housekeeping gene hypoxanthine–guanine phosphoribosyltransferase (HPRT).
Protein multiplex Luminex Bio-Plex assay
Cell free supernatants of Phl p 1- and control-treated cells were stored at −20°C until analysis. Cytokine levels in the supernatants of the treated cells were determined using a Bio-Plex Human Cytokine 30-Plex Panel Kit in combination with a Bio-Plex workstation (Bio-Rad, Veenendaal, the Netherlands). All standards were diluted in HBSS. Concentrations were calculated from a dilution series of standards using the Luminex software. Fold changes (FC) were calculated compared to the detected release in control-stimulated cells. If the measured mediator concentration in the control was below the detection level, the FC was calculated on basis of the detection limit for the respective mediator.
Phl p 1-induced changes in gene expression level in airway epithelial cells
The overall gene expression ratio between HBSS and Phl p 1-stimulated cells was measured and calculated using the Rosetta resolver software. Expression was considered to be expressed significantly differentially with P ≤ 0·01 after one-way Bonferroni multiple test correction. The Affimetrix U133 plus 2·0 contains 54·675 transcripts, representing about 38 500 genes. Stimulation of NCI-H292 cells with Phl p 1 induced a huge response. In total, 7218 transcripts showed a significant regulation of gene expression. The subsequent analysis of transcripts, that were changed by more than threefold, revealed 86 transcripts to be up-regulated and 16 transcripts to be down-regulated, corresponding to 54 and 11 known genes, respectively.
For validation of the microarray data we first investigated the influence of stimulation with Phl p 1 on the expression of housekeeping genes. Analysing the expression of these genes revealed that their regulation is not affected by stimulation with Phl p 1. For example, the average expression ratio between control and stimulated condition for GAPDH has been 1·03 (±0·00), and beta-2 microglobulin (B2M) showed a fold change of 0·90 (±0·01). To confirm further the observed expression levels, we chose 11 genes from the 65 known genes that showed an up- or down-regulation by more than threefold for confirmatory real-time polymerase chain reaction (PCR). In Table 1 the ratios calculated from the microarray data and the real-time PCR derived expression ratios are shown. Expression levels derived from real-time PCR were calculated as fold change between control and Phl p 1-stimulation and normalized to GAPDH and HPRT. The real-time PCR data strongly resemble the ratios calculated on the basis of the microarray data (see Fig. 1). Statistical analysis of the two sets of data revealed a high correlation (R = 0·79 with R2 = 0·63), pointing towards a correlation of 63% between the microarray data and the results of the real-time PCR.
Table 1. Validation of the Phl p 1-induced expression profile of selected genes using real-time polymerase chain reaction (PCR). All values are given as mean (± standard deviation). Real-time PCR values are stated as fold change between control and Phl p 1-stimulated cells and normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and hypoxanthine–guanine phosphoribosyltransferase (HPRT).
Real time-PCR (fold induction)
After successful validation of the microarray data we continued with sorting the genes that were regulated at least threefold up or down after stimulation with Phl p 1 into functional classes. Using their ascribed gene ontology, genes were assorted by their molecular function or the biological process in which they are involved (Table 2). After correcting the assorted gene categories for the group ‘unclassified genes’, we found 10 genes (16·4%) to be in the group ‘immunity and defence’, four genes in the group ‘receptor activity’, ‘enzyme inhibitor activity’ (two) and ‘ion binding’ (three). Comparable to previous results we obtained with GPE the groups ‘cell communication’, representing cytokines, chemokines and growth factors, and ‘metabolism’ appeared to be the most prominent ontology groups, with 23 genes (38%) and 16 genes (26·2%), respectively. All genes identified to be single members of a certain ontology group were pooled within the category ‘miscellaneous’.
Table 2. Genes expressed differentially in NCI-H292 cells after stimulation with Phl p 1, sorted by ontology into functional groups. Stated are gene aliases, fold changes calculated in relation to control-stimulated cells, gene names, P-values and accession numbers.
Solute carrier family 5 (inositol transporters), member 3
Selenoprotein P, plasma, 1
Epithelial membrane protein 1
Aldehyde dehydrogenase 1 family, member L2
Citrate lyase beta-like
Potassium channel, subfamily K, member 3
Transmembrane protein 46
Immediate early response 3
FERM domain containing 3
Chromosome 20 open reading frame 175
Chromosome 8 open reading frame 4
Similar to WDNM1-like protein
Chromosome X open reading frame 58
Chromosome 13 open reading frame 15
Chromosome 6 open reading frame 155
Chromosome 11 open reading frame 70
KN motif and ankyrin repeat domains 4
RCSD domain containing 1
Network analysis and Phl p 1-induced significant signalling pathways
To be able to identify interactions within genes that displayed an altered expression after stimulation with Phl p 1, we performed a network analysis· The evaluation of genes with a significant change in expression level by at least threefold up or down revealed 50 genes to interact directly, with 31 genes showing Phl p 1-induced changes in expression (Fig. 2). Resembling the observation that cell communication was identified to be among the most prominent GO groups, a predominant proportion of the genes in the network are related to chemokines and cytokines, with interleukin (IL)-1A, IL-1B, IL-6 and IL-8 forming hubs within the computational model.
Next we analysed signalling pathways induced significantly after stimulation of NCI-H292 cells with Phl p 1. We therefore screened all significantly Phl p 1-regulated probe sets independent of their fold change for their participation in specific signalling cascades. As shown in Table 3, the following pathways appeared to be regulated by Phl p 1: epidermal growth factor receptor 1 (EGFR1), IL-1, IL-2, IL-3, IL-4, IL-6, IL-7, Notch, Wnt, transforming growth factor beta receptor (TGF-βR), tumour necrosis factor (TNF)-α, and nuclear factor kappa-light-chain enhancer of activated B cells (NF-κB).
Genes matched with Agilent technology. IL, interleukin; TNF, tumour necrosis factor; EGF, epidermal growth factor; NF-κB, nuclear factor kappa B.
Phl p1-induced protein expression
GO analysis indicated the group ‘cell communication’ to be the dominant functional group among Phl p 1-regulated genes. We therefore continued with the determination of Phl p 1-induced release of mediators in general. As shown in Table 4, measurement of the release of a panel of chemokines, cytokines and growth factors showed a massive Phl p 1-induced increased release of IL-1RA, IL-6 and IL-8, which was also reflected by a clear increase in gene expression levels. Albeit at rather low level, vascular endothelial growth factor (VEGF) showed an increase in release and also a moderate Phl p 1-induced up-regulation of gene expression. Granulocyte–macrophage colony-stimulating factor (GM-CSF), granulocyte colony-stimulating factor (G-CSF), epidermal growth factor (EGF), interferon gamma-induced protein IP-10 and monokine induced by gamma interferon (MIG) showed a moderate increase on mediator levels, with gene expression levels remaining unaltered upon stimulation. Although the monocyte chemotactic protein-1 (MCP-1) showed an increased release of 5·4-fold, the corresponding gene showed no altered expression upon stimulation, which is likely to result from the fact that 24 h were chosen for stimulation. IL-12 also showed a clear enhanced release upon stimulation, with an increase of detectable IL-12p40/p70 by fourfold. Interestingly, this was not reflected in the Phl p 1-induced expression level of the IL-12 gene, with IL12A (p35) and IL12B (p40) gene expression remaining unaltered. For IL-1β we were not able to detect an increased release, but the stimulation with Phl p 1 resulted in a clear up-regulation of gene expression by 3·6-fold, indicating IL-1β to be more probably part of a secondary response induced by Phl p 1. A similar picture emerged for TNF-α, IL-15 and fibroblast growth factor (FGF)-β that were also undetectable within the supernatant of Phl p 1-stimulated airway epithelial cells (AECs), with gene expression levels being only slightly up-regulated. IL-2R was identified to be released on basal level after stimulation with Phl p 1 and the gene expression level remained unaltered.
Table 4. Mediator release from NCI-H292 cells after stimulation with Phl p 1 for 24 h. After stimulation cell free supernatants were collected and analysed for the presence of chemokines and cytokines. The stated values represent the average (± standard deviation) of a triplicate experiment and are given in pg/ml; with detection limits in brackets for each of the mediators. Fold change (FC) was calculated in relation to level of control-stimulated cells.
Phl p 1
*P ≤ 0·01.
Not significant; n.d., not detectable; ELISA, enzyme-linked immunosorbent assay; HBSS, Hanks's balanced salt solution; IL, interleukin; TNF, tumour necrosis factor; MCP, major chemoattractant protein; GM-CSF, granulocyte–macrophage colony-stimulating factor; VEGF, vascular endothelial growth factor; FGF, fibroblast growth factor beta; EGF, epidermal growth factor; IP, interferon gamma-induced protein; MIG, monokine induced by gamma interferon; FC, fold change.
Although in this experimental set-up we stimulated AECs with a single component derived from GPE only, we still observed a huge number of Phl p 1-regulated genes and significant Phl p 1-induced release of mediators. In total, 7218 transcripts showed a significant regulation of gene expression, with 71 genes showing a Phl p 1-induced altered gene expression level of at least threefold. The observed extent of this response might result partially from the usage of the cell line NCI-H292. In previous studies we have already observed more pronounced fold changes and a higher number of affected genes in stimulated NCI-H292 cells when compared to the response induced in primary nasal epithelial cells . Nevertheless, we decided to use this established model, as it offers the advantage of limited experimental variation and lack of potentially contaminating cells. The absence of intra-individual differences enables the detection of small changes in gene expression levels and therefore might add valuable information.
Phl p 1 has already been shown to induce an increased expression of IL-6, IL-8 and TGF-β from AECs on both mRNA and protein levels , but our data show the true extent of this response. Subsequent analysis of prominent GO clusters showed that the group ‘cell communication’ belongs to the most prominent Phl p 1-induced groups, reflecting the ability of the major allergen to induce the release of mediators that, in turn, can attract and influence the activity of immune-competent cells. On the assumption that the response of NCI-H292 cells has been shown to rather reflect the healthy epithelial immune response towards allergens , our data support the observation that epithelial cells from healthy donors release high levels of cytokines and chemokines upon stimulation with house dust mite extract (HDM) or birch pollen allergens [21,22].
Network analysis of direct interactions within genes that showed a change in gene expression level of at least threefold revealed central roles for IL1A, IL6, IL8, but also for IL1B. Phl p 1 induced a pronounced up-regulation of the expression of this proinflammatory cytokine (3·6-fold). Interestingly, polymorphisms within IL1B are associated with allergic rhinitis , confirming the importance of this mediator within atopic diseases. Vroling et al. already investigated the response of AECs to HDM allergens in detail on both a respiratory epithelial cell line and primary nasal epithelial cells from healthy donors and allergic individuals [20,24]. Network analysis of the aforementioned samples revealed a set of overlapping genes, describing an epithelial core-response towards HDM. Interestingly, some of these genes (namely IL1B, CXCL1, CCL20, IL8, CXCL2, PTGS2 and SERPINB3) were also regulated by Phl p 1 and part of the interaction network created upon genes that showed Phl p 1-induced alterations in expression level by at least threefold (Fig. 2). It is tempting to speculate about the existence of a general epithelial core-response towards aeroallergens, with genes being part of the core-response and genes showing changes in expression level upon stimulation with the respective allergenic extract that are specific for the source of allergens. Mutations within genes that are part of the core-response or mutations in genes that are induced specifically might explain why individuals become mono- or polysensitized. However, further experiments are required to follow this hypothesis.
Interestingly, the chemokine receptor CXCR4 displayed a pronounced up-regulation for both, GPE- and Phl p 1-stimulation of AECs. This receptor has been shown to be expressed on Th2 cells and is relevant during Th2-type allergic airway responses . Besides its role as major co-receptor for HIV-entry , CXCR4 was also shown to be expressed in human intestinal epithelial cells and is up-regulated in inflammatory bowel diseases . CXCR4 is functionally expressed on the cell surface of various immune cells and plays a role in cell proliferation and migration. This G-protein-coupled receptor (GPRC) is endocytosed upon activation. Furthermore, it has been shown that blocking of CXCR4 has a significant effect in down-regulating the inflammation of allergen-induced immune response . Taken into account that agonist-occupied CXCR4 undergoes clathrin-dependent endocytosis in human epithelial cells [29,30], it is tempting to speculate that the massive up-regulation of CXCR4 we observed after stimulation of AECs with GPE (11·5-fold up) or Phl p 1 (7·8-fold up) might contribute to the epithelial uptake of grass pollen allergens. Furthermore, after stimulation of epithelial cells from allergic donors with the birch pollen major allergen Bet v 1 an over-representation of the GO terms ‘response to viruses’ and ‘cell receptors’ could have been detected [31,32]. However, further experiments are required to investigate this possible mechanism.
Many major allergens seem to posses functional attributes which may, at least partially, explain their allergenicity. It might well be that the immune activating molecules are not identical to the proteins triggering the allergic response, although this would still raise the question of why those particular proteins are recognized among the many present during exposure. Extracts from different pollen have been shown to cause detachment of airway epithelial cells dependent on a proteolytic activity of the extracts . Also, the activating capacity of HDM extract and HDM allergens is caused probably by a proteolytic activity of the allergens with, for example, Der p 1 identified to be a serinprotease. Der p 1 has been shown to degrade tight junction proteins, thereby inducing the release of mediators and facilitating the route of the allergen through the epithelial barrier [34–36]. Although we were not able to observe any detachment of Phl p 1-stimulated AECs, members of the family of serinprotease inhibitors (SERPIN) showed an increase in gene expression level and CDH1, encoding for cadherin-1, a calcium-dependent cell–cell adhesion glycoprotein, appeared to be down-regulated upon stimulation with GPE by more than threefold. Phl p 1 is discussed to exert proteolytic activity, with a recombinant in Pichia pastoris expressed form acting as protease, whereas natural Phl p 1 failed to act as protease both in vitro and ex vivo. In contrast to GPE-stimulated AECs, Phl p 1 only induced a moderate down-regulation of CDH1 (−1·5-fold), which might allow the suspicion that proteases potentially present within the whole extract might allow the passage of allergens (including Phl p 1) through the epithelial barrier by disrupting cell–cell adhesion.
However, regardless of the speculated ability to act as a protease itself, this major allergen induces the up-regulation and release of a broad range of mediators, indicating it to be a powerful trigger of the immune system. Taken together, we were able to show that the major allergen Phl p 1 induces a huge response in AECs, both on the gene expression level, as well as on the level of mediator release. Genes belonging to the GO cluster ‘cell communication’ were among the most prominent functional groups, which is also reflected in IL1A, IL1B, IL6 and IL8 building centres in a model of direct gene interaction of genes showing an altered gene expression level of at least threefold. Further detailed comparison of GPE-and Phl p 1-induced gene expression might be beneficial with regard to the application of single components within diagnosis and immunotherapy.
We thank Marisa Böttger (Research Center Borstel, Germany) for excellent technical assistance. This study was supported by Deutsche Forschungsgemeinschaft (DFG)-TR22-Z1/A3 and by the Interuniversity Attraction Poles Programme (IUAP) – Belgian state – Belgian Science Policy P6/35.
None of the authors has any conflict of interest related to this manuscript.