Analysis of gene expression in peripheral blood eosinophils from patients with atopic dermatitis and in vitro cytokine-stimulated blood eosinophils

Abstract

Investigation of differentially expressed genes in eosinophils of patients with allergic diseases such as atopic dermatitis (AD) will provide important information for elucidating possible mechanisms of pathology. To identify novel genes that are expressed in AD, we compared gene expression in samples of peripheral blood eosinophils from AD patients and healthy volunteers. RNA was extracted from peripheral blood eosinophils. The expression of various genes, such as those for cytokine receptors, eosinophil activation marker, platelet activating factor (PAF) receptor, eosinophil-specific granular proteins and apoptosis-related genes, was confirmed using real-time reverse transcription–polymerase chain reaction (RT-PCR). Peripheral blood eosinophils of healthy volunteers were also isolated and stimulated for introduction of various cytokines. RNA was extracted and gene expression was monitored. Several genes, such as those for cytokine receptors (granulocyte-macrophage colony-stimulating factor (GM-CSF) receptor α and β chain and interleukin (IL)-3 receptor α chain), CD44 and PAF receptor were expressed at significantly higher levels in AD patients than in healthy volunteers. In addition, the anti-apoptotic genes, bcl-2 and bcl-xL, were expressed at increased levels in AD patients. No single gene expression correlated with clinical markers, such as eosinophil count or IgE levels. Expression of GM-CSF receptor β chain and IL-3 receptor α chain in isolated blood eosinophils of healthy volunteers was stimulated by IL-5, IL-4, interferon (IFN)-γ and GM-CSF. Expression of bcl-2 and bcl-xL was also increased after stimulation with IL-5, IL-4 or IFN-γ. The in vitro enhancement of cytokine-stimulated gene expression correlated well with the enhancement observed in clinical samples of eosinophils, suggesting that cytokines may affect gene expression in vivo in eosinophils of patients with AD.

Introduction

Allergic diseases such as asthma and atopic dermatitis (AD) are considered to have a multi-factor aetiology. These diseases are likely to be caused by a complex interplay of the expression of many different genes and by multiple environmental factors that affect gene expression. Recent developments in the technology of gene expression analysis, such as differential display and reverse transcription–polymerase chain reaction (RT-PCR), are now making it possible to analyse and compare the expression of large numbers of genes in different clinical samples. Investigation of differentially expressed genes will provide important information for elucidating a contribution to disease aetiology.

Eosinophils play important roles in the pathogenesis of allergic diseases such as asthma [1,2] and AD [3,4]. Their role is especially significant for the occurrence of tissue damage during the chronic phase of the disease. However, the expression of disease-related genes in these cells has not been examined carefully. Therefore, to identify genes relevant to allergic diseases, especially AD, we compared samples of peripheral blood eosinophils derived from both AD patients and healthy volunteers for differentially expressed genes.

The study of gene expression in eosinophils presents some difficulty, because the percentage of the cells in peripheral blood from healthy donors is less than 3%. In addition, only a few leukaemic eosinophil cell lines, such as Eol [5], YY-1 [6] and AML14.3D10 [7,8], are available for in vitro studies. In this study, we also describe an in vitro model of short time culture of peripheral blood eosinophils for analysing gene expression after cytokine stimulation. Our model has enabled us to evaluate the mechanism of eosinophil activation in peripheral blood from patients with allergic pathological conditions such as AD.

Methods

Clinical blood samples for real-time RT-PCR

Table 1 shows the number of AD patients and healthy volunteers used in this study. Peripheral blood samples were collected from healthy adults or from children with AD of differing severity. Written informed consent to participate in the study was obtained in all cases from volunteers or parents after providing them with detailed information about the study and subjects’ rights. Our experiments were authorized by the ethical committees of both Genox Research Inc. and the National Children's Medical Research Center.

Table 1.

Patient population

Healthy controls	13 people (6 males, 7 females) age 26–63 years (average 39·3)
Mild AD	15 people (9 males, 6 females) age 0–21 years (average 10·2)
Moderate AD	15 people (11 males, 4 females) age 3–18 years (average 9·4)
Severe AD	18 people (9 males, 9 females) age 0–29 years (average 11·3)

Patients with AD were diagnosed according to the criteria of Hanifin [9], and the severity of AD was judged using a modified version of Leicester's scoring system, as we have reported previously [10]. Briefly, five clinical features (erythema, papule, excoriation, oozing and lichenification) were evaluated at six body sites (hand, trunk, elbow, hand, knee and foot). The percentage area of clinical features per whole body surface defined the AD severity (mild, <10%; moderate,>10% <30%; severe,>30%). The AD patients were treated with topical glucocorticoid ointments. A milder steroid (mainly hydrocortisone acetate) was used for treating the face, and a stronger steroid (mainly dexamethasone valerate) for the body. None of the AD patients had received systemic glucocorticoids. According to the diagnostic standards of Haniffin [9], we diagnosed characteristic eczema when the diseased areas appeared in both face and chest. Chronic and recurrent dermatitis was also characterized by continuing for more than 2 months in patients below 1 year of age. Moreover, all selected patients below 1 year of age were IgE-positive and had case histories of dry skin and staphylococcal infection. In cases over 1 year of age, persistence of the diseased area for more than 6 months was a diagnostic criterion.

Measurement of blood indices

Total IgE level in each serum sample was measured using Unicap total IgE EIA systems (Pharmacia Diagnostics AB, Uppsala, Sweden). The number of eosinophils in peripheral blood samples was determined with Stromatolyser Eo II by automated blood haematology analyser SE-9000 (Sysmex, Kobe, Japan) [11].

Purification of peripheral blood eosinophils

Granulocytes were isolated from 10–50 ml heparinized venous blood from AD patients and healthy volunteers with Ficoll-Paque (Amersham Pharmacia Biotech, Uppsala, Sweden). Red blood cells were removed by hypotonic lysis. The CD16-negative flow-through eosinophil fractions were collected with MACS separation columns (Myltenyi Biotec, Bergisch Gladbach, Germany), and the purity was checked after Diff-Quick Staining (always greater than 99%).

Culture of peripheral blood eosinophils after stimulation by cytokine

Eosinophils from 100 ml of peripheral blood from healthy volunteers with a purity of more than 99% were suspended in Iscove's minimum essential medium (MEM) supplemented with 10% immobilized fetal calf serum (FCS) and 5 × 10⁻⁵m 2-mercaptoethanol. The 24-well cell culture plates were coated with 1% bovine serum albumin (BSA) that had been heated at 70°C for 1 h in order to avoid eosinophil binding to the plates. 1 × 10⁶ cells were inoculated with 10-fold-decreasing concentrations, ranging from 10 to 0·1 ng/ml, of the cytokines under investigation (interleukin (IL)-5, IL-4, interferon (IFN)-γ, granulocyte-macrophage colony-stimulating factor (GM-CSF) and eotaxin). Incubation was for 3, 6 and 18 h, and the reaction was stopped by addition of the Isogen solution (Nippon Gene, Toyama, Japan).

All cytokines using in the in vitro experiments were purchased from R&D Systems, Inc., Minneapolis, MN, USA. The endotoxin level is <0·1 ng per 1 µg of the cytokine as determined by the LAL method, indicating that the concentration in our experimental media was <1 pg/ml. The effects of contaminating endotoxin are likely to be negligible.

These in vitro experiments using normal volunteers were performed three times, twice using the same volunteer and once using a different volunteer. Duplicate wells of eosinophils were used in each experiment.

Extraction of RNA and preparation of cDNA

Isolated eosinophils or cultured cells were solubilized in the Isogen solution and total RNA was isolated. The RNA was DNase treated and converted to cDNA. The copy number of the transcripts was quantified in an ABI PRISM 7700 sequence detection system (Applied Biosystems, Foster City, CA, USA) [12,13].

Real-time RT-PCR

The 15 genes (CD44, GM-CSF receptor α chain, GM-CSF receptor β chain, IL-3 receptor α chain, IL-5 receptor α chain, IFN-γ receptor, C-C type chemokine receptor (CCR3), very late activation antigen 4 (VLA-4), platelet activation factor (PAF) receptor, leukotriene D4 (LTD4) receptor, major basic protein (MBP), eosinophil-derived neurotoxin (EDN), bcl-2, bcl-xL and bax) in the cDNA samples were amplified in ABI PRISM 7700 by introducing the forward and reverse primers and TaqMan probes designed by Primer Express software (Applied Biosystems) as shown in Table 2.

Table 2.

The primers and TaqMan probes for the 15 genes

Gene	Forward primer	Reverse primer	TaqMan probe
CD44	5′-GAC CTC TGC AAG GCT TTC AA	5′-TCC GAT GCT CAG AGC TTT CTC	5′FAM-ACC TTG CCC ACA ATG GCC CAG AT-3′TAMRA
GM-CSF receptor α chain	5′-ATC CAA ATT CAG GAA GGG AGG	5′-TCG CCC AGG TAC AGT TCA TTA A	5′FAM-CCG CTG CTC AGA ATT TCT CCT GTT TCA TCT-3′TAMRA
GM-CSF receptor β chain	5′-TGG AGT GGC CTC TGG TTA TG	5′-GGG AAC TAG GGA GAC AGA CGA G	5′FAM-CTG CAG ACC TGG TAT TCA CCC CAA ACT CA-3′TAMRA
IL-3 receptor α chain	5′-ACC CAC CAA TCA CGA ACC TAA G	5′-GGT CAC ATT TCT GTT AAG GTC CC	5′FAM-ATG AAA GCA AAG GCT CAG CAG TTG ACC-3′TAMRA
IL-5 receptor α chain	5′-CTG CAG AAC GAC CAC TCA CTA CT	5′-AAT TGA GGT TCC AGG AGA CCC	5′FAM-CAG CTG GGC TTC TGC TGA ACT TCA TG-3′TAMRA
IFN-γ receptor	5′-GAA TGA ACG GAA GTG AGA TCC AG	5′-CCC CAC ACA TGT AAG ACT CCT T	5′FAM-TGA CGA GAT TCA GTG CCA GTT AGC GA-3′TAMRA
CCR3	5′-CAT TGT CCA TGC TGT GTT TGC	5′-AGG TGA CGA TGC TGG TGA TGA	5′FAM-TTC GAG CCC GGA CTG TCA CTT TTG GT-3′TAMRA
VLA-4	5′-GTC CTT GTT TAA TGC TGG AGA TGA T	5′-GCT TCT CTT CCA GCT CTA AAA TCT T	5′FAM-ACG ACT CTA CAT GTC AAA CTA CCC GTG GG-3′TAMRA
PAF receptor	5′-TGT GGG AGC TGC ATC CTA CTT C	5′-CTC AAA GCA GCG AGT GAC GTT	5′FAM-TCA TCC TGG ACT CCA CCA ACA CAG TG-3′TAMRA
LTD4 receptor	5′-GCA CCT ATG CTT TGT ATG TCA ACC	5′-ATA CCT ACA CAC ACA AAC CTG GC	5′FAM-TTA TGA CAG CCA TGA GCT TTT TCC GGT G-3′TAMRA
MBP	5′-GGC TGT TGA GTC TAT CTC AGT GC	5′-CCC ACC ACT TTT ACT GTG TCC TC	5′FAM-ATG GTG CAC AAA AAC CTT ACG TGT CCT G-3′TAMRA
EDN	5′-CCT GTC CTA GTA ACA AAA CTC CGA	5′-GTG AGG TTA CAG TGG ATT AAA GGC	5′FAM-AAA TTG TCA CCA CAG TGG AAG CCA GG-3′TAMRA
bcl-2	5′-ACA TGA CCC CAC CGA ACT CA	5′-TGG AGG AGC TCT TCA GGG AC	5′FAM-AGG CCA CAA TCC TCC CCC AGT TCA-3′TAMRA
bcl-xL	5′-GCG TAG ACA AGG AGA TGC AGG T	5′-GGT CAT TCA GGT AAG TGG CCA T	5′FAM-TTG GTG AGT CGG ATC GCA GCT TG-3′TAMRA
bax	5′-AAA GAT GGT CAC GGT CTG CC	5′-TCC AAG ACC AGG GTG GTT G	5′FAM-TGG GCG TCC CAA AGT AGG AGA GGA AG-3′TAMRA

In order to provide templates for quantifying the copy numbers of transcripts, the coding regions of the amplified genes were cloned from human peripheral blood eosinophil cDNA libraries that were prepared in our laboratory. The expression of a human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) housekeeping gene in each sample was also assessed by using the standard primers and TaqMan probe (Applied Biosystems).

Statistics

The non-parametric Dunnet signed-rank test was used to evaluate the changes in gene expression among patient groups. The difference was considered significant when the P-value was less than 0·05. Results were expressed as mean ± standard error of the mean.

Results

Gene expression in eosinophils from clinical samples

The allergic indices obtained from clinical blood samples, total IgE and the number of eosinophils are shown in Fig. 1. The blood of patients with mild AD showed a small increase in total IgE and a significant increase in the number of eosinophils. However, indices were increased in the blood samples of patients with severe disease.

Fig 1.

The clinical scores of allergy in blood samples of study patients. Total IgE (UA/ml) (a) and the percentage ratio of eosinophils (b) in blood of healthy controls, mild AD, moderate AD and severe AD patient groups are shown (***P < 0·001).

To study the gene expression profiles in peripheral blood eosinophils, we measured expression levels of 15 selected genes which might be related to eosinophil activation and prolonged cell survival (Table 2). The expression of the CD44, GM-CSF receptor α chain, GM-CSF receptor β chain, IL-3 receptor α chain and PAF receptor genes was significantly increased in AD blood eosinophils (Fig. 2a). The gene expression levels in all groups of patients with AD, regardless of the degree of disease severity, were statistically higher than in healthy volunteers. Expression of genes for other plasma membrane receptors, IL-5 receptor α chain, IFN-γ receptor, CCR3, VLA-4 and LTD4 receptor were similar in both patients and healthy volunteer groups (data not shown).

Fig 2.

The expression levels of various genes in peripheral blood eosinophils. The expression of CD44 (a), GM-CSF receptor α chain (b), GM-CSF receptor β chain (c), IL-3 receptor α chain (d), PAF receptor (e), bcl-2 (f) and bcl-xL (g) in eosinophil samples from healthy controls and patients with mild, moderate and severe AD are shown (*P < 0·05, **P < 0·01, ***P < 0·001). The copy numbers of each transcript per 1 ng RNA, standardized to levels of GAPDH transcript, are shown on the ordinate.

The expression of the anti-apoptotic genes, bcl-2 and bcl-xL, was increased in eosinophils from AD patients, although bcl-2 was significantly higher in eosinophils from patients with only mild AD (Fig. 2f,g). The expression of the apoptotic gene, bax was not different between AD patients and healthy groups (data not shown). Moreover, there was no increase in either of two eosinophil specific granular markers MBP and EDN in the patients (data not shown).

Similar results were obtained in two independent experiments using different blood samples with 10 samples or more per single group. The results of multivariate studies of the gene expression and the clinical data are summarized in Table 3a. There were no correlations between gene expression and either total serum IgE or the number of peripheral blood eosinophils. However, the genes that were enhanced in AD patients correlated well with each other (Table 3b).

Table 3.

Multivariate analysis of clinical data and gene expression

Clinical score	CD44	GM-CSF R α chain	GM-CSF Rβ chain	IL-3 R α chain	bcl-2	bcl-xL	PAF R
(a)
Total IgE (UA/ml)	−0·065	−0·016	0·158	−0·026	−0·027	−2·77E-04	−8·93E-03
Eosinophil count (%)	0·051	0·137	0·029	0·181	−0·112	0·022	0·110
(b)
GM-CSF R α chain	0·600§
GM-CSF R β chain	0·642§	0·701¶
IL-3 R α chain	0·566§	0·505§	0·435§
bcl-2	0·466§	0·208‡	0·260‡	0·796¶
bcl-xL	0·644§	0·538§	0·510§	0·829¶	0·804¶
PAF R	0·589§	0·856¶	0·778¶	0·436§	0·169†	0·556§
Gene	CD44	GM-CSF R β chain	GM-CSF R β chain	IL-3 R α chain	Bcl-2	Bcl-xL

^¶Strong correlation r = 0·7–1·0;

^§correlation r =0·4–0·7;

^‡weak correlation r = 0·2–0·4;

^†no correlation r < 0·2.

Eosinophil gene expression in vitro by various cytokine stimuli

We analysed expression pattern of seven genes in vitro, which were expressed at high levels in eosinophils of patients with AD. Eosinophils of healthy volunteers were stimulated in the presence of cytokines for 3 h. The dose–response of some cytokines, interleukin-5 (IL-5), IL-4, IFN-γ, GM-CSF and eotaxin, all of which have the potential to stimulate eosinophil functions, are shown in Fig. 3a. Eotaxin, the ligand of CCR3, induced no gene expression in vitro. All experiments in normal volunteers, as described in the Methods, gave comparable results.

Fig 3.

In vitro dose–response of the expression of various genes in peripheral blood eosinophils of healthy volunteer after cytokine stimulation. The expression of CD44 (a), GM-CSF receptor α chain (b), GM-CSF receptor β chain (c), IL-3 receptor α chain (d), PAF receptor (e), bcl-2 (f), bcl-xL (g) and IL-5 receptor α chain (h)after a 3-h stimulation with the indicated concentrations of IL-5 (red bar), IL-4 (blue bar), IFN-γ (yellow bar), GM-CSF (green bar) and eotaxin (brown bar) is shown. An arbitrary unit contrasted with the unstimulated control level (standardized to levels of GAPDH transcript) is shown on the ordinate. These experiments were performed three times. The mean ± standard error of data from three independent experiments for each stimulation condition and the statistical results of the paired t-test are shown (*P < 0·05, **P < 0·01, ***P < 0·001).

An eosinophil activation marker CD44 expression was induced in a dose-dependent manner, coincident with the stimulation of IL-5, IL-4 and GM-CSF, but not with IFN-γ and eotaxin. Expression of the GM-CSF receptor α chain gene was stimulated by IL-5 and IL-4. The dose-dependent gene expression for GM-CSF receptor β chain was also increased, along with IL-5, IL-4, IFN-γ and GM-CSF. The IL-3 receptor α chain gene was also stimulated dramatically by IL-5, IL-4 and GM-CSF, but only slightly with IFN-γ. PAF receptor gene expression was enhanced only by IL-5. However, IL-5 receptor α chain gene expression was not increased by these cytokine stimuli (Fig. 3h). Genes for other plasma membrane markers and receptors, such as IFN-γ receptor, CCR3 and VLA-4, which were expressed at similar levels in eosinophils from both AD patients and healthy volunteers, showed no alteration by the cytokines (data not shown).

Bcl-2 genes, which were expressed higher in AD than in healthy volunteer peripheral blood eosinophils, were enhanced with IL-5 (Fig. 3f). Bcl-xL genes, which were also expressed higher in AD, were enhanced with IL-5, IL-4, IFN-γ and GM-CSF, but not with eotaxin (Fig. 3g). These cytokines did not induce expression of the bax gene (data not shown).

The time-course of the expression of these genes with IL-5, IL-4, IFN-γ and GM-CSF stimuli was monitored. Most of the genes were expressed within 3 h, reached their highest levels in approximately 6 h and decreased after overnight culture (data not shown).

Discussion

We have identified genes which are expressed at significantly higher levels in peripheral blood eosinophils derived from AD patients than healthy volunteers. In this study, the patients and controls were not age-matched. However, age is unlikely to account for the differential expression because there were no differences between the younger and older AD patients (data not shown). In our two previous papers, we compared differentially expressed genes between healthy volunteers and allergic disease patients of various ages [14,15] and showed that there were no statistical differences between the younger and older AD patients at similar stages of severity. The eosinophils of human neonates were harder to activate than those of adults [16]. The finding that shows enhancement of various mRNA even in such inactive young eosinophils supports our results.

The levels of IgE and eosinophil numbers are much higher in AD patients than in healthy volunteers (Fig. 1). There were no patients with negative serum IgE levels. All samples were collected from patients with allergic AD. However, we could not find a correlation of specific gene expression in eosinophils with the elevation of these clinical markers of AD disease (Table 3a). To clarify the interrelationships of gene expression in eosinophils, we measured 15 eosinophil marker genes in clinical samples (Table 2). The expression profiles of some of these genes have been analysed during studies of eosinophil proliferation and maturation from CD34⁺ progenitor cells [17]. The genes for an adhesion receptor, CD44, some cytokine receptors (GM-CSF receptor α chain, GM-CSF receptor β chain and IL-3 receptor α chain) and PAF receptor were expressed at significantly higher levels in AD patients than in healthy volunteers (Fig. 2). The correlation ratio between gene transcripts from two independently amplified genes was very high in a series of 60 clinical samples (Table 3b).

CD44 is one of the typical eosinophil activation markers, which is expressed in parallel with the morphological change to hypodense eosinophils both in vivo and in vitro [18]. These findings suggest that the eosinophils of the AD patients are activated and more sensitive to cytokine receptor-signalling through high expression of such cytokine receptors. The multivariate analysis of the expression of these genes and high correlation ratio suggests that both CD44 and cytokine receptor expression are controlled by the same transduction signals (Table 3b).

The gene for PAF receptor may be increased in AD blood eosinophils [19], and IL-5 stimulation results in enhanced functional PAF receptor expression in vitro in human eosinophils [20]. However, this is the first report to investigate the expression of a battery of genes in eosinophils and to study the interrelationships between in vivo and in vitro transcript levels. Yamada T et al. have reported detection, by flow cytometry, of IL-3, IL-5 and GM-CSF receptor expression in peripheral blood cells of patients with atopic conditions [21]. However, they could not find enhancement of the expression of the corresponding gene products. Our studies show that the genes for the cytokine receptors are activated in eosinophils from patients with AD conditions. Such results suggest that the signals for prolonged eosinophil survival are elevated by IL-5, GM-CSF and IL-3 through GM-CSF receptor α chain, GM-CSF receptor β chain and IL-3 receptor α chain. These signals were elevated, not only in severe but also in mild AD conditions. In the mild AD conditions, total IgE did not increase dramatically (Fig. 1a). The ability to differentiate the eosinophils of patients with even mild AD from normal individuals points to the potential diagnostic value of the gene expression evaluation. Elevation of the cytokine receptor genes, followed by activation of the eosinophil survival signals such as IL-5, GM-CSF and IFN-γ (Fig. 3) correlated well with each other in both in vivo studies and prolongation of the life span of cells in vitro. Eotaxin activated eosinophils via CCR3, but did not prolong the cell life span. It is possible that cell survival was not prolonged by eotaxin activation because neither the cytokine receptors nor anti-apoptotic genes were enhanced.

We also found that anti-apoptotic genes, bcl-2 and bcl-xL, were expressed at higher levels in AD patients compared to healthy controls (Fig. 2f,g), suggesting that the eosinophils in the AD patients may be resistant to apoptotic stimuli. IL-5, GM-CSF and IL-3 are anti-apoptotic signals for eosinophils. Therefore it is reasonable to expect that bcl-2 and bcl-xL, but not bax expression, were increased in allergic clinical samples in conjunction with the increase of cytokine receptor gene expression. There are the reports that bcl-2 is expressed in eosinophils from sputum of patients with acute asthma [22] and blood eosinophils from a patient with hypereosinophilia [23]. Our studies extend the type of patients for which there is an increase in bcl-2 expression to those with AD. The increase of the anti-apoptotic marker expression was detected in early stages of mild AD, although there was no correlation between the expression of bcl-2 and bcl-xL genes and the number of peripheral blood eosinophils (Table 2a). Therefore, the measurement of anti-apoptotic gene expression in blood eosinophils may also have diagnostic value. In the multivariate studies, the genes for the bcl-2 and bcl-xL and IL-3 receptor α chain showed a very high correlation ratio (more than 0·7), but there was a lower correlation with the other cytokine receptors (Table 3b). These data suggest that the bcl gene family expression is regulated by transduction signals that are different from those for the cytokine receptors, including GM-CSF receptor α chain and GM-CSF receptor β chain.

The subject of in vitro eosinophil apoptosis is controversial. Ochiai et al. showed that IL-5, but not IFN-γ, inhibits eosinophil apoptosis by up-regulation of bcl-2 [24]. Others report that bcl-2 protein was not induced by IL-5 or GM-CSF [25]. In our results, induction of GM-CSF receptor in eosinophils of healthy volunteers was weak, making it difficult to identify a specific effect of GM-CSF on bcl-xL. In a recent microarray analysis, eosino-phils revealed an increase in a number of survival and apoptosis genes, but bcl-2 and bcl-xL were not studied [26]. In addition, IL-4 was previously thought to be a signal for apoptosis [27]. Our data may help to resolve some of these conflicting data. Distinct dose–response in bcl-xL expression suggests that IFN-γ and GM-CSF, as well as IL-5, are the important antiapoptotic signals in vitro.

The IL-5 receptor α chain, IFN-γ receptor, CCR3, VLA-4, LTD4 receptor, MBP, EDN and Bax were expressed at similar levels in AD patients and healthy volunteers. We have reported previously that mRNA levels of eosinophil granule proteins such as MBP and EDN were significantly decreased during maturation of eosinophils from bone marrow cells to peripheral eosinophils [17]. Such a tendency may be related to the observed differences between healthy and the disease conditions. We have not yet determined the level to which these genes are translated in eosinophils, either in vivo or in vitro. However, we did find a good correlation between the in vivo and in vitro gene transcript levels. Almost all the genes, which were enhanced in eosinophils of AD conditions, were enhanced after IL-5 treatment, as well as other cytokines in vitro (Table 4).

Table 4.

Eosinophil gene expression in vitro after cytokine stimulation

Gene	IL-5	IL-4	IFN-γ	GM-CSF	Eotaxin
CD44†	○	○	×	○	×
GM-CSF R α chain†	○	○	×	×	×
GM-CSF R β chain†	○	○	○	○	×
IL-3 Rα chain†	○	○	○	○	×
PAF R†	○	×	ρ	ρ	×
bcl-2†	○	ρ	ρ	×	×
bcl-xL†	○	○	○	○	×
bax‡	×	×	×	×
IL-5 Rα chain‡	×	ρ	×	×	×
IFN-γ R‡	×	×	×	×
CCR3‡	×	×	×	×
VLA4‡	×	×	×	×

×, Not increased; ρ, marginally increased; ○, significantly increased.

^†Higher in AD patients than in healthy volunteers;

^‡similar in both AD patients and healthy volunteers.

The genes that were expressed at similar levels in both AD patients and healthy people, specifically: IL-5 receptor α chain, IFN-γ receptor, CCR3, VLA-4 and bax, were not enhanced after various cytokine stimuli in eosinophils of healthy people. IL-5 regulates bax function post-translationally by inhibiting its translocation to mitochondria [28]. Therefore, the IL-5 stimuli in peripheral blood are likely to be pathologically important in AD even though increases in plasma levels are not detected. Blood eosinophils might be activated in tissues undergoing pathology and recirculate into the peripheral blood. In case of bcl-2 and bcl-xL, the dose–response stimulation by IL-4 was not clear, but IL-5 and IFN-γ-induced anti-apoptotic gene expression in a dose-dependent manner (Fig. 3f,g), suggesting that these cytokines are also important during in vivo eosinophil activation in AD conditions.

The role of eosinophils in the pathology of allergic diseases, especially asthma, is controversial [29–32]. The decrease in eosinophil cell number in the pathological tissues of asthmatics was not correlated with the clinical recovery of lung functions. The change in eosinophil number might be a result, rather than a cause, of the disease. However, our results suggest that eosinophils are changed by activation of expression of many genes before the elevation of the number of the cells, regardless of the cause or outcome of AD.

In summary, we have compared the expression of these gene transcripts in peripheral blood eosinophils derived from AD patients and healthy volunteers, and after in vitro activation of eosinophils of healthy persons with cytokines. The clinical markers, such as total IgE or eosinophil number, did not correlate with any single gene expression. However, the marker gene expression was enhanced in patients with both mild and severe AD. Most mild AD patients, classified by a <10% area of dermatitis per whole body surface, are in the early phase of AD. Thus, determination of gene expression in blood eosinophils may become a useful diagnostic test for the early phase of AD.

Moreover, in vitro gene expression of eosinophils of healthy volunteers after various cytokine stimuli correlated with the results from peripheral blood eosinophils in pathological conditions of AD patients. Our comprehensive gene expression analysis suggests that eosinophils in allergic conditions are activated according to a pattern of increase in expression of multiple genes. Therefore, the stimulation of the in vitro eosinophil culture system by cytokines may provide a useful model for study of molecular mechanisms of progression of allergic disease.