Jutta Renkonen Transplantation Laboratory & Infection Biology Research Program Haartman Institute University of Helsinki Helsinki Finland
Background: Previous work in type-I pollen allergies has mainly focused on lymphocytes and immune responses. Here, we begin to analyse with a systems biology view the differences in conjunctival epithelium obtained from healthy and allergic subjects.
Methods: Transcriptomics analysis combined with light and electron microscopic analysis of birch pollen allergen Bet v 1 located within conjunctival epithelial cells and tissues from birch allergic subjects and healthy controls was carried out.
Results: Bet v 1 pollen allergen bound to conjunctival epithelial cells within minutes after the exposure even during the nonsymptomatic winter season only in allergic, but not in healthy individuals. Light- and electron microscopy showed that Bet v 1 was transported through the epithelium within lipid rafts/caveolae and reached mast cells only in allergic patients, but not in healthy individuals. Transcriptomics yielded 22 putative receptors expressed at higher levels in allergic epithelium compared with healthy specimens. A literature search indicated that out of these receptors, eight (i.e. 37%) were associated with lipid rafts/caveolae, which suggested again that Bet v 1 transport is lipid raft/caveola-dependent.
Conclusions: We show a clear difference in the binding and uptake of Bet v 1 allergen by conjunctival epithelial cells in allergic vs healthy subjects and several putative lipid raft/caveolar receptors were identified, which could mediate or be co-transported with this entry. The application of discovery driven methodologies on human conjunctival epithelial cells and tissues can provide new hypotheses worth a further analysis to the molecular mechanisms of a complex multifactorial disease such as type-I birch pollen allergy.
Allergic conjunctivitis is a type-I hypersensitivity reaction caused by antigens, such as pollen allergens, cross-linking specific anti-IgE molecules on the previously presensitized mast cells. Most of the previous work on the pathogenesis of the pollen allergy has focused on the mechanisms of the immune responses elicited by the allergen (1–5). We have begun to analyse the molecular mechanisms facilitating the pollen entry into the epithelium and thus allowing its interaction with tissue-bound mast cells Joenvaara S et al., unpublished data. We have previously introduced a noninvasive application of the confocal reflected light microscopy enabling the direct and repeatable quantitative analysis of the very early phases of conjunctival allergic inflammation in human patients (6, 7). The challenge with the birch pollen allergen Bet v 1 led within minutes to the leucocyte extravasation into the conjunctival sites of inflammation only in allergic, but not in healthy individuals (8).
The role of epithelial cells as the first line of defence has recently been emphasized (8, 5–11). The aim of this system level analysis is to begin to dissect the role of conjunctival epithelial cells in the early pathogenesis of type-I allergic reactions. We focused on this disease as it is common, i.e. up to 15–20% of the population in Europe suffer from pollen allergies (4, 11–13). Allergic symptoms cause a lot of morbidity and thus are a major indirect reason for the losses of productivity in the society (14). Type-I allergy is an example of an acute disease where its causative agents such as the birch pollen allergen Bet v 1 have been identified (1, 4, 15). Conjunctival epithelial cells and tissues can be obtained from healthy controls and allergic patients during winter (a symptom-free period for both patients and controls) before and after the in vivo conjunctival pollen challenge. To minimize the in vitro artefacts, we collected, analysed and compared only freshly isolated conjunctival epithelial cell swabs and conjunctival tissue biopsies from healthy control subjects and birch pollen allergic individuals during the nonsymptomatic winter before and after the pollen challenge.
This study is a part of a large systems biology effort to begin to understand the early events in the type-I allergy. We selected this disease as a starting point of the system level approach as it is an acute disease. These are clearly less difficult, if not easy, to study compared with chronic diseases. Here, we can challenge the subjects with a known allergen such as the birch pollen during the nonsymptomatic winter season and induce the acute disease reactions within minutes, collect cell and tissue specimens in a time series manner both from allergic patients and healthy control subjects. In this larger series of our top-down analysis of a human type-I allergy, we collected not only conjunctival but also nasal epithelial cells and tissues Joenvaara S et al., unpublished data. These specimens were analysed before and after the birch pollen challenge and the following data domains were gathered; transcriptomics with various platforms (Affymetrix, Illumina and ABI), Bet v 1-associated proteins with various HLPC methods combined with MS/MS analysis and the light and electron microscopy analysis.
In this report, we show the results obtained from conjunctival epithelial cells. Comparing the conjunctival epithelial cells and tissues from healthy and allergic subjects, we show with the light and electron microscopic analysis that the Bet v 1 pollen allergen bound to epithelial cells within minutes even during the nonsymptomatic winter season only in allergic, but not in healthy individuals. It also was transported through the morphologically intact epithelium and reached mast cells located close to or under the basal lamina during the nonsymptomatic winter season only in allergic patients. Co-localization together with caveolin 2 suggested a traffic within lipid rafts/caveolae.
Thus, the application of discovery driven methodologies on human conjunctival epithelial cells and tissues can provide a new hypothesis worth a further analysis to the molecular mechanisms of a complex multifactorial disease such as the type-I birch pollen allergy.
Material and methods
This study was performed in accordance with the Declaration of Helsinki, and the study protocol was reviewed and accepted by the Committee on Ethics of Helsinki University Eye and Ear Hospital. All subjects consisting of five allergic subjects (mean age 29.4 ± 4.7 years, range 12 years, two females, three males) and five healthy controls (mean age 33.4 ± 8.4 years, range 19 years, two females and three males) gave their informed consent. All the microscopic and transcriptomic analysis were carried out separately for each specimen.
The status of birch and other pollen allergies was tested in the routine hospital laboratory with skin prick tests. The tested allergens were; birch (Betula verrucosa), timothy grass (Phleum pratense), meadow fescue (Festuca pratensis), mugwort or common wormwood (Artemisia vulgaris), fungal allergen (Cladosporium herbarum), cat, dog, horse, cow, house dust mite, latex and histamine as a positive and saline as a negative control. The birch pollen allergic patients selected for this study showed no other pollen allergies and the healthy control subjects were negative for all tested antigens. The selected patients also reacted to the in vivo conjunctival challenge with the birch pollen, while the healthy controls did not. The allergic patients also had a positive history of seasonal allergy in May; the time birch pollen is in the air in Finland. All allergic subjects selected for this study fulfilled these criteria; positive in the prick test only to the birch allergen, reacted to the conjunctival pollen challenge and had a positive history of seasonal allergy in May–June (birch pollen season in southern Finland). Concomitantly, the healthy controls did not react to any allergen in the prick test, did not react to conjunctival birch pollen challenge and had no history of any allergy.
Conjunctival epithelial specimens
Biopsies from conjunctival epithelium as well as conjunctival epithelial cell swabs from both five healthy subjects and five allergic patients, verified with skin prick tests against the birch pollen extract, were obtained.
Bet v 1 location within the epithelial tissue was studied with polyclonal anti Bet v 1 antibodies (first a gift from R Valenta, University of Vienna, Austria and later our own polyclonal rabbit anti-Bet v 1 antibody towards recombinant Bet v 1). Rabbit IgG and the depletion of the primary antibody were used as negative controls. Secondary fluorescent anti-rabbit antibody (Alexa 488; Molecular Probes, Eugene, OR, USA) was used to detect the signal and ConA lectin coupled with Alexa 633 (Molecular Probes) was used to detect epithelial tissue.
Immuno transmission electron microscopy
Immuno transmission electron microscopy (immunoTEM) was performed to analyse the presence of the Bet v 1 allergen within the conjunctival epithelial tissue as described previously (6, 17).
The quantification of the immunohistochemistry as well as immunoTEM specimens were carried out with dedicated in silico workflows (Medicel Ltd, Helsinki, Finland), which allowed not only the automated counting, but also the traceability of results with all their metadata and their easy combination to other data domains such as mass spectrometry and transcriptomics (http://www.medical.com).
Conjunctival epithelial cells from healthy and allergic subjects were collected during winter before and after the in vivo birch pollen challenge. A basic analysis was performed from five healthy and five allergic subjects during winter when no one had any clinical symptoms. The RNA isolation was carried out according to the Queen’s Rnaeasy Mini Handbook, the concentration measurement with RNA 6000 Nano kit for Agilent Bioanalyser, Agilent Technologies, Santa Clara, CA 95051, USA and the specimens were hybridized with Affymetrix chip platform on U133A and conducted in Finnish DNA Microarray Centre (Turku Centre for Biotechnology).
During the microarray analysis, the following integrated databases were used for probe annotations and conversions between genes, transcripts and proteins: HG-U133-PLUS-2 (2006-11-15-a), NCBI/RefSeq/RNA (2007/05/24), EnsEMBL (45-a), SwissProt (53.2-a), TrEMBL (36.2), GeneOntology (2007-07-30), KeGG (2006/12/15-1) and Intact (2007-07-27).
Data analysis was performed using the Integrator (Medicel Ltd, Helsinki, Finland). The data preprocessed with the GCRMA method were normalized using quantile normalization. Only those probes known to represent one transcript were selected for the analysis. Statistically differing expressions between different groups were calculated with a workflow using anova. Enrichments–depletions of gene ontology (GO) categories were essentially computed as described Joenvaara S et al., unpublished data.
This is the first part of a study of a human disease at the system level. We identified and verified a group of birch allergic subjects (n = 5). No other allergen(s) caused reactions to these patients in a standard skin prick test. An age-matched group of healthy controls (n = 5) with no symptoms and a negative skin prick test was also identified. We then collected conjunctival epithelial cell swabs and biopsies from all subjects during winter, when there was no birch pollen in the air and thus even the allergic patients did not present any symptoms (Fig. 1). During this previously nonsymptomatic period, we could also apply an in vivo perturbation by an application of pollen allergen to the eyes, which caused rapid symptoms in allergic subjects, but no manifestation in healthy controls. We collected a new set of cell and tissue specimens and analysed all specimens with the light and TEM before perturbations, as well as 1 and 10 min after it. We also collected mRNA specimens during the winter season both from the healthy and allergic subjects.
In this systems biology approach, we try to study the type-I allergy disease on a systems level. Here, we do not restrict only to single genes or one data domain, but instead try to begin to dissect differences within a broader content. Furthermore, we restricted only to freshly isolated human cells and tissue specimens to avoid as much as possible the in vitro artefacts of cell cultures. Our study is an example of a holistic top-down approach to begin to understand the early events of an allergic reaction in patients.
Morphological alterations induced by the ocular Bet v 1 challenge to epithelial cells
During the winter season, we challenged the asymptomatic allergic patients and the healthy subjects with the conjunctival application of the birch pollen extract and collected conjunctival epithelial cell swabs before the challenge, and 1 and 10 min after it. Bet v 1 is the major allergen within the birch pollen extract used for these perturbations and thus its presence and location were analysed with the following experiments.
During the asymptomatic winter season, there were no morphological differences between conjunctival cells obtained either from healthy or allergic subjects before the birch pollen challenge (Fig. 2A,B). However, after a very short in vivo conjunctival perturbation with the birch pollen extract the epithelial cells from the allergic, but not from the healthy subjects, showed a gross vacuolization when analysed with the TEM (Fig. 2C,D). Thus, clearly the seemingly morphologically similar conjunctival epithelial cells during the asymptomatic period had the capacity to react differently to the birch pollen perturbation.
Bet v 1 binding to epithelium
We then analysed the presence, quantity and location of Bet v 1 in the conjunctival biopsies from these same individuals obtained during a different session during the same winter period. Already the 1-min in vivo ocular challenge resulted in a patchy location of the Bet v 1 binding to conjunctival epithelial surfaces only in the allergic, but not in the healthy individuals (Fig. 2E–H and Table 1). Even exposures for 10 min did not result in the Bet v 1 binding to the epithelium in healthy subjects (data not shown).
Table 1. Bet v 1 binding to conjunctival epithelial surface analysed with the immunofluorescence microscopy during the winter season after 1-min ocular pollen challenge
Bet v 1 on the epithelial surface
−, No staining is graded; ++, medium staining is graded; +++, strong staining is graded. See also Fig. 2.
Healthy subject 1
Healthy subject 2
Healthy subject 3
Healthy subject 4
Healthy subject 5
Allergic subject 1
Allergic subject 2
Allergic subject 3
Allergic subject 4
Allergic subject 5
Bet v 1 entry to epithelial cells and tissues
These observations were further verified on the immunoTEM, where the Bet v 1 binding to and entry in the epithelium were evident only in cells taken from the allergic, but not from the healthy individuals. During the asymptomatic phase before the pollen perturbation, epithelial cells both from the healthy and allergic subjects did not show essentially any anti-Bet v 1 immunogold staining (Table 2). Already 1 min after the perturbation, before the fulminant clinical symptoms, Bet v 1 was found both at the villae on the cell surface, as well as within the villae, cytoplasm, intracellular vesicles and also in the nuclei in epithelial cells from allergic patients. Concomitantly, there was no Bet v 1 in the conjunctival epithelial cells obtained from the healthy control subjects. At 10 min after the challenge, while the clinical symptoms already were fulminant, most of Bet v 1 was located within the cytoplasm and not any more on the cell surface of villae (Table 2).
Table 2. The analysis of the amount of anti-Bet v 1 gold label particles in conjunctival epithelial cells before and after the in vivo pollen challenge during the asymptomatic winter season
At villus surface
Bet v 1 rapidly bound to and entered into the conjunctival epithelial cells in allergic patients but not in healthy subjects. The numbers in brackets indicate the number of microscopic fields analysed with original magnification ×27 500.
0.1 ± 0.1
1.0 ± 0.4
0.3 ± 0.2
0.9 ± 0.3
0.5 ± 0.3
2.6 ± 1.0
0.2 ± 0.2
0.5 ± 0.3
0.5 ± 0.3
1.1 ± 0.8
0.6 ± 0.7
0.8 ± 0.7
0.5 ± 0.3
12.4 ± 7.1
8.6 ± 5.3
36.2 ± 10.4
11.6 ± 6.2
1.8 ± 1.0
1.1 ± 1.0
0.8 ± 0.5
3.5 ± 2.2
0.6 ± 0.4
0.6 ± 0.3
0.2 ± 0.3
0.7 ± 0.4
27.3 ± 11.2
1.1 ± 0.8
2.7 ± 1.4
Conjunctival biopsies were also obtained and the immuno TEM analysis showed a rapid transport of Bet v 1 into the epithelial layers (Table 3). The stratified conjunctival epithelium was analysed in the vertical direction starting from the apical surface and entering to deeper epithelial layers. Although the majority of anti Bet v 1-gold label was within the most apical epithelial cells (cells no: 1–2), the staining could be detected already at this early time point all the way into the more basal epithelial cell layers (cell no 4). This indicates the extremely rapid traffic through conjunctival epithelial cells and tissues in allergic patients. Concomitantly, there was essentially no Bet v 1 binding to and traffic through the conjunctival epithelium in healthy subjects (Table 3).
Table 3. The analysis of the location of anti-Bet v 1 gold label particles within various epithelial layers of the stratified conjunctival epithelium after 1 min of the pollen perturbation
Bet v 1
The number of anti-Bet v 1 gold particles per the whole specimen was analysed.
Epithelial cell 1
Epithelial cell 2
Epithelial cell 3
Epithelial cell 4
In most cases, Bet v 1 was found in the small clusters of anti-Bet v 1 gold label particles (Fig. 3). The number of anti-Bet v 1 gold label particles was 25 ± 3/μm2 in allergic patients and only on the background level, i.e. 4 ± 0.3 μm2 (Fig. 3B) in healthy controls. These clusters were never found prior to perturbations, but already 1 min after the birch pollen challenge, the clusters were detected at the surface and also within the conjunctival epithelial cells (Fig. 3C). Likewise, most of the gold label (over 75%) was detected within such clusters in specimens from the allergic patients. As a negative control challenge, birch pollen allergic subjects were also challenged with timothy grass (P. pratense), but no entry of this pollen was detected (data not shown).
In most cases, Bet v 1 was present in small clusters while binding to and entering into the epithelium. We next tried to identify further the mechanisms of this binding and entry. As a caveola was a good putative candidate for such a mechanism, we first analysed the presence of various caveolar markers in the conjunctival epithelial tissue clustering together with the Bet v 1 protein. By using the double immunoTEM, we could show that a caveolar marker, caveolin 2 but not caveolin 1 and 3, was present on the conjunctival epithelial surface in the same clusters as Bet v 1 (in the double immunoTEM staining. Caveolin 2 showed a strong co-localization with Bet v 1, especially at the early sites of the contact (Fig. 3D–F).
We then collected epithelial cell swabs from both allergic and healthy subjects during winter, when there was no birch pollen in the air and thus not even the allergic patients had any symptoms. The conjunctival epithelial cells displayed 182 differentially expressed transcripts during the symptomatic winter period (Table S1).
The novel observation in this study is that epithelial cells and tissues obtained from allergic patients were able to uptake Bet v 1 pollen allergen while the same was not true in the healthy subjects. To begin to characterize this difference, we performed a transcriptomic analysis of the intact epithelium from both allergic and healthy subjects during the winter season when neither group had any symptoms. The aim of this analysis was to ask if the epithelium of these two groups expressed already different transcriptomic states even in the presence of no clinical symptoms of allergy. The classical study approach is to pick a few altering genes/proteins and start to work with them. Instead systems biology tries to look these differences from a broader perspective and thus we have performed a top-down analysis of the differentially expressed genes.
We next converted these differentially expressed transcripts to corresponding proteins (Table S1). In an attempt to try to get some putatively relevant knowledge from this list of proteins, we first analysed it against GO categories. They provide a controlled vocabulary to describe gene and gene product attributes in any organism, and here in human subjects. The enrichments–depletions of GO categories within our list of differentially regulated proteins between allergic and healthy subjects were essentially computed as described (6).
Gene ontology slim is an ‘easy-reader’ or a cut-down version of the GO containing a subset of the terms in the whole GO. Gene ontology slims give a broad overview of the ontology content without the detail of the specific fine grained terms. Gene ontology slims are particularly useful for giving a summary of the results of a GO annotation of a list of genes or proteins generated with e.g. a microarray when the broad classification of the gene product function is required (http://www.geneontology.org/GO.slims.shtml).
The most enriched biological process was signal transduction (Table S2). The category “Integral to plasma membrane” was the most enriched cellular compartment and transporters, cell adhesion and endopeptidase activities were the most enriched molecular functions.
Another approach to generate some information from this gene/protein list was to try to identify putative cell-surface receptors, which might be involved in the uptake of Bet v 1 into the allergic but not healthy epithelium during winter without any perturbations. From the list of upregulated genes/proteins in the specimen of allergic, but not healthy subjects, we gathered a subset of well-known cell-surface receptors. We could list 22 differentially expressed cell-surface receptors, which displayed enhanced expression levels in allergic vs healthy controls (Table S3). Among these receptors, (many) seven were G protein coupled receptors, three receptors involved in immune response, two linked to the extracellular matrix and one lectin known to bind venom toxin and HI viruses (Table S3). An intriguing observation is that eight of them (36%) have been found in lipid rafts/caveolae according to our manual PubMed literature search (Table S3). The frequency of caveolar/lipid raft proteins in the order of 1% within the human proteome and thus this 36% proportion is significantly higher than that excepted by chance. This again could suggest that the epithelial cell surfaces of allergic and healthy epithelium differ significantly in the expression of these lipid raft/caveolar receptors. This in silico analysis provides a list of putative receptors, which could be involved in the uptake of Bet v 1.
We have initiated a large systems biology effort to begin to understand the early events in the type-I allergy. We selected this disease as a starting point of a systems level approach as it is an acute disease (12, 13), which is clearly less difficult, if not easy, to study compared with chronic diseases. Here, we can challenge the subjects with a known allergen such as the birch pollen during the asymptomatic winter season and induce the acute disease reactions within minutes and collect cell and tissue specimens in a time-series manner both from allergic patients and healthy control subjects. We began to analyse with a systems biology view the differences in the nasal epithelium obtained from healthy and allergic subjects (6) and now expand these analysis to other respiratory epithelial tissues. With the light and electron microscopic analysis, we show here that the Bet v 1 pollen allergen bound to epithelial cells within minutes even during the nonsymptomatic winter season only in allergic, but not in healthy individuals. It travelled through the epithelium together with lipid rafts/caveolae and reached mast cells only in allergic patients.
The novel observation of this study is the difference between the epithelial cells and tissues obtained from allergic patients and healthy controls. We began to characterize this difference by performing the transcriptomic analysis of the intact epithelium from both allergic and healthy subjects during the winter season, when neither group had any symptoms. A classical study approach is to pick a few altering genes/proteins and start to work with them. Systems biology tries to look these differences from a broader perspective and thus we have performed a top-down analysis of the differentially expressed genes. We first try to classify them into categories (here GO categories), which could characterize the differentially expressed genes/proteins.
We found 182 statistically significantly differentially expressed transcripts (and converted them to corresponding proteins). We continued to analyse this whole set and characterized them with GO categories. An over-representation of cell receptors and proteins involved in signal transduction were/was seen.
As there were so many receptors displaying enhanced expression levels in allergic subjects compared with healthy controls, we also took a closer look at them. It could be postulated that allergic epithelium might express receptor(s) ready to putatively uptake the birch pollen or at least co-migrate along the Bet v 1 transport. A list of 22 epithelial cell-surface receptors displayed enhanced expression levels in allergic patients vs healthy controls. Eight of them (36%) have been found in lipid rafts/caveolae according to our manual PubMed literature search. The frequency of caveolar/lipid raft proteins in the order of 1% within the human proteome and thus this 36% proportion is significantly higher than that excepted by chance. This again could suggest that the epithelial cell surfaces of the allergic and healthy epithelium differ significantly in the expression of these lipid raft/caveolar receptors. Together with the Bet v 1 co-localization to lipid rafts/caveolae in the immuno EM analysis these observation could further strengthen our hypothesis of the caveolar/lipid raft dependent Bet v 1 transport in the allergic epithelium.
Thus, an application of discovery driven methodologies on human conjunctival epithelial cells and tissues can provide a new hypothesis worth a further analysis to the molecular mechanisms of a complex multifactorial disease such as the type-I birch pollen allergy.
The kinetics of caveolar traffic of Bet v 1 resembles in many ways other known caveolar-dependent phenomena such as the entry of viruses, bacterial toxins or even parasites (17–22). We have previously identified a small set of Bet v 1-associated proteins, where 37% were of the caveolar origin Joenvaara S et al., unpublished data. Here, we analyse the putative receptors on conjunctival epithelial cells, which could participate in this Bet v 1 uptake seen only in allergic, but not healthy subjects. Altogether 22 receptors were overexpressed in allergic conjunctiva and out of these, several could play a putative role in the Bet v 1 recognition and traffic through epithelium.
Bet v 1 protein has been shown to have an affinity for a wide section of amphiphilic molecules including sterols and fatty acids (23, 24). Bet v 1 is also able to interact and associate with lipid vesicles, which induces large conformational rearrangements to the structure of Bet v 1 (25). Thus in a separate study, we have in silico modelled the Bet v 1 binding to various amphiphilic ligands and lipids (Mattila K and Renkonen R, unpublished data).
Here, we show that already during winter, the conjunctival epithelium in allergic patients was able to bind and uptake the Bet v 1 allergen putatively in a caveola-dependent manner. In our immunoTEM analysis, a proportion of the gold label could be located in the nuclei of epithelial cells. While we currently do not have any direct role for this, we can put out a hypothesis that the Bet v 1 entry to the conjunctival cells modifies directly or indirectly their transcriptomic profiles, which is the subject of our next study.
The most significant aspects of this and other top-down explorative studies using systems biology approaches is that they can provide a truly new hypothesis. Before this study, few would have argued that the conjunctival epithelium is so markedly different in healthy compared with allergic subjects already during the nonsymptomatic winter season. Now, based on these results, the mechanisms of the lipid raft/caveolar-dependent Bet v 1 pollen allergen binding to and trafficking through the epithelium can be further explored and analysed. The ability to discover aetiologically relevant disease mechanisms, which fall outside the prior medical and biological knowledge, is the major motivation for these unbiased explorative approaches.
We thank Maaret Helintö and Jussi Kirveskari for conjunctival specimens, Marko Nykänen and Mikko Tiainen for their help with the immunoTEM microscopy and Sirkka-Liisa Holm for the technical help.