• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of immunologyLink to Publisher's site
Immunology. Aug 2004; 112(4): 605–614.
PMCID: PMC1782530

Danger signals derived from stressed and necrotic epithelial cells activate human eosinophils


Eosinophilic granulocytes are found in tissues with an interface with the external environment, such as the gastrointestinal, genitourinary and respiratory tracts. These leucocytes have been associated with tissue damage in a variety of diseases. The aim of this study was to evaluate whether necrotic epithelial cells can activate eosinophils. The danger theory postulates that cells of the innate immune system primarily recognize substances that signal danger to the host. We damaged epithelial cell lines derived from the genital (HeLa cells), respiratory (HEp-2 cells) and intestinal tracts (HT29 cells) and assessed their capacity to cause eosinophilic migration, release of putative tissue-damaging factors, such as eosinophil peroxidase (EPO) and eosinophil cationic protein (ECP), as well as secretion of tissue-healing factors, e.g. fibroblast growth factors (FGF)-1 and -2 and transforming growth factor (TGF)-β1. We found that necrotic intestinal cells induced chemotaxis in human eosinophils. EPO release was elicited in eosinophils stimulated with necrotic cells derived from all cell lines, as well as from viable HEp-2 and HT29 cells. Release of ECP was only seen in eosinophils incubated with necrotic intestinal or genital cells, not viable ones. Both necrotic intestinal and genital cells elicited FGF-2 secretion from eosinophils. TGF-β1 was released from eosinophils exposed to viable and necrotic HT29 cells. These findings indicate that eosinophils are able to recognize and be activated by danger signals released from damaged epithelial cells, which may be of importance in understanding the role of eosinophils in the various inflammatory conditions in which they are involved.

Keywords: eosinophil, chemotaxis, inflammation, danger signal, necrosis


The danger theory was formulated in the beginning of the 1990s1 It postulates that the immune system recognizes substances that cause danger to the host, rather than discriminate between host-derived and foreign compounds (self versus non-self).2 These substances are called danger signals and can arise from stressed or damaged host cells or pathogenic micro-organisms. In contrast, host cells that are healthy or die by apoptosis are not believed to release danger signals.

The eosinophils belong to the polymorphonuclear family of leucocytes called granulocytes and develop from haematopoietic stem cells in the bone marrow. Interleukin (IL)-5 directs eosinophils from the bone marrow to the blood,3 where the cells persist for up to one day. The adhesion molecules l-selectin, complement receptor 3 (CR3, also called CD11b/CD18) and very late antigen-4 on the eosinophil surface interact with their cognate receptors on the endothelium, e.g. E- and P-selectin, intercellular adhesion molecule-1 and vascular cell adhesion molecule-1, allowing eosinophils to migrate into the tissues,4 to which they are attracted by various chemotactic substances such as eotaxin, monocyte chemotactic protein (MCP)-4 and regulated upon activation normal T-cell expressed and secreted (RANTES).5

It has been estimated that 99% of the eosinophils are found within various tissues, particularly those with an epithelial interface with the environment, such as the gastrointestinal, genitourinary and respiratory tracts.6 However, the physiological function of the eosinophil in the healthy human being is poorly understood. The cells have mainly been studied in diseased individuals, i.e. in subjects suffering from allergic conditions such as asthma, atopic dermatitis and rhinitis, in persons infested with parasites and subjects afflicted by malignancies, both solid tumours and leukaemias.7 It has also been appreciated that eosinophils increase in numbers in the intestinal mucosa during various types of inflammation, such as inflammatory bowel disease,8 celiac disease9 and dysentery caused by Shigella.10 In addition, eosinophils are frequently found infiltrating solid organ transplants undergoing rejection11 as well as in the gastrointestinal tract of individuals suffering from graft-versus-host disease following bone marrow transplantation.12

The eosinophils have been blamed for contributing to tissue damage in the various disease entities they are associated with, mainly through their release of cytotoxic granule constituents and free oxygen radicals. However, eosinophils may also partake in tissue remodelling and wound healing by the release of growth factors.13 Presumably, eosinophils can adopt both these roles, but it is not clear which signals steer the eosinophil in either of these directions. The aim of this study was to see if blood-derived eosinophils from healthy individuals are able to recognize danger signals that arise from damaged epithelial cells, originating from the predilection organs of the eosinophil, namely the gastrointestinal, genitourinary and respiratory tracts. We tested various methods to damage epithelial cells and compared their ability to generate necrotic or stressed cells that could be recognized by eosinophils. Eosinophil activation was measured as the capacity of necrotic versus viable epithelial cells to cause eosinophilic migration, release of putative tissue-damaging factors, i.e. eosinophil peroxidase (EPO) and eosinophil cationic protein (ECP), as well as secretion of possible tissue-healing factors, fibroblast growth factors (FGF)-1 and -2 and transforming growth factor (TGF)-β1.

Materials and methods

Isolation of eosinophils

Blood eosinophils were isolated from fresh human buffy coats obtained from adult healthy blood donors at Sahlgrenska University Hospital (Göteborg, Sweden) and Kungälv's Hospital (Kungälv, Sweden). The buffy coats were diluted 1 : 1 (v/v) with 0·9% NaCl and most of the erythrocytes were depleted by dextran sedimentation (Amersham Pharmacia Biotec AB, Uppsala, Sweden). The mononuclear cell fraction was removed by density centrifugation with Ficoll-Paque PLUS (Amersham Pharmacia Biotec AB). Residual erythrocytes were eliminated by repeated hypotonic lysis in distilled water. The eosinophils were separated from neutrophils by negative depletion using anti-CD16-coated magnetic-activated cell sorting (MACS)-beads (Miltenyi Biotec Inc., Auborn, CA) and a MACS column. The eosinophils were washed and resuspended in Kreb's Ringer glucose buffer, KRG (NaCl 7·0 g, KCl 0·37 g, MgSO4 0·3 g, KH2PO4 0·3 g, Na2HPO4.2H2O 1·5 g, glucose 1·8 g, CaCl2 0·15 g, 2H2O, distilled water 1 l) and stored on ice until use. Eosinophil purity, median value 94% (min 90%, max 99%), was assessed by counting Diff-Quick-stained cells (Dade Behring AG, Düdingen, Switzerland) cytospun (Shandon Scientific Co. Ltd, London, UK) onto glass slides; the most frequent contaminants were mononuclear cells. Eosinophil viability, determined by Trypan blue (Merck, Darmstadt, Germany) exclusion, was >98%.

Cell lines and cell culture

HeLa (cervical adenocarcinoma), HEp-2 (epidermoid carcinoma from larynx) and HT29 (colorectal adenocarcinoma) cells were cultured in Eagle's minimal essential medium (MEM; Tissue Culture Laboratory, Department of Virology, Göteborg, Sweden) supplemented with 5% fetal calf serum, FCS (Tissue Culture Laboratory) at 37°, in a humidified atmosphere containing 5% CO2. All cell cultures were free from Mycoplasma infection as determined by polymerase chain reaction (PCR).14 Upon confluent growth, cells were detached by incubation with Versén (NaCl 8 g, KCl 0·2 g, Na2HPO4 × 2H2O 1·4 g, KH2PO4 0·2 g, ethylenediaminetetraacetic acid (EDTA) 0·2 g, distilled water 1 l) for 45 min (HEp-2 and HT29) or for 15 min (HeLa cells), at room temperature. Detached cells were washed and resuspended in Eagle's MEM. The cell concentration was adjusted to 2 × 106 cells/ml. Cells were placed on ice and used within an hour.

Generation of necrotic/stressed epithelial cells

Three methods were employed to damage epithelial cells: freeze–pressing, freeze–thawing and heat treatment. Freeze–pressing was performed according to Magnusson and Edebo.15 In brief, frozen epithelial cells at a concentration of 2 × 106 cells/ml were placed in a metal cylinder. The cylinder in turn was placed in an ice bath containing ethanol at −25° for 1 hr. Cells were then pressed through a narrow tube with a piston exerting a pressure of 2000 bar. The smashed frozen cells were thawed on ice, aliquoted and kept at −80° prior to use. Eppendorf tubes (Sarstedt, Nümbrecht, Germany) containing 106 epithelial cells in 0·5 ml Eagle's MEM were freeze–thawed by repeated (four times) freezing (−70°) and thawing (room temperature). Heat-treated epithelial cells were incubated in a 56° water bath for 45–50 min Their viability was checked using Trypan blue exclusion, and was routinely less than 60%. Control cells, named ‘viable cells’ consisted of cells newly detached from the culture flasks.


Eosinophil migration towards viable or damaged epithelial cells was measured using a microwell dual chamber system (ChemoTx chamber: filter pore size 3 µm, 6·0 mm diameter wells; Neuro Probe Inc. Gaithersburg, MD). Viable and freeze–pressed HEp-2, HeLa and HT29 cells at three different concentrations (corresponding to 30 000, 15 000 and 3000 whole cells/30 µl) were added in triplicate to wells in the bottom chamber and covered with a framed filter. Next, eosinophil suspensions (30 000 cells/30 µl) were placed on top of the filter over each well and the chamber system was incubated for 90 min at 37°, in a humidified atmosphere with 5% CO2. The non-migrated eosinophil suspension on top of the filter was removed using tissue paper and the chamber system was centrifuged at 400 g for 10 min at room temperature. To let migrated eosinophils adhere, the bottom chamber was incubated for 10 min at 37° and then supernatants were aspirated and discarded. The eosinophilic chemoattractant eotaxin (PeproTech EC Ltd, London, UK) at 10−8 and 10−9 m was used as a physiological positive control and control buffer (KRG supplemented with 0·3% bovine serum albumin (BSA; Sigma-Aldrich, Steinheim, Germany)) was used as a negative control to measure spontaneous eosinophil migration. In addition, 30 000 eosinophils were incubated in the bottom wells (triplicate) to mimic maximal migration (100%). For blocking experiments, cyclosporin H (kind gift of Novartis Pharma, Basel, Switzerland), an antagonist of the formyl peptide receptor, was used as follows: eosinophils were preincubated for 15 min on ice with 10−6 m cyclosporin H diluted in KRG−0·3% BSA and in addition, the same concentration of cyclosporin H was added to the lower wells in the microchamber system.16 The physiological ligand of the formyl peptide receptor, the tripeptide formyl-methionine-leucine-phenylalanine (fMLF; Sigma) was used at a concentration of 10−7 m.16

To quantitate migrated eosinophils, the cells were lysed by the addition of 1% Triton-X-100 (Sigma-Aldrich) in phosphate-buffered saline (PBS) for 15 min The cell lysate was transferred to a flat-bottomed 96-well plate (Nunc A/S, Roskilde, Denmark) containing 80 µl PBS per well and EPO contents in lysed eosinophils were determined (see below, EPO assay). The percentage of migrated eosinophils was estimated by dividing the average (triplet samples) absorbance in the wells containing the various chemoattractants with that of the control wells that mimicked 100% eosinophil migration and subtracting spontaneously migrated eosinophils (towards medium).

Stimulation of eosinophils

Eosinophils diluted in KRG (50 000 cells/well) were put in 96-well polystyrene plates with v-shaped bottoms (Nunc A/S). The eosinophils were stimulated in duplicate with various numbers (50 000; 25 000; 5000 cells/50 µl total volume) of epithelial cells (undamaged, freeze–pressed, freeze–thawed, heat-treated) for 30 min at 37°. As a control, purified lipopolysaccharide (LPS, smooth Escherichia coli 055, kind gift from Dr Liliana Håversen, Department of Clinical Bacteriology, Göteborg University) at final concentrations of 1 µg/ml, 10 ng/ml, 1 ng/ml, 0·1 ng/ml was added to the eosinophils either in the presence or absence of 2·5% heat-inactivated human serum. After the 30-min stimulation period, eosinophils were spun down for 7 min, 400 g at room temperature. The cell-free supernatants were analysed for EPO, FGF-2, ECP and TGF-β1 contents. In experiments designed to detect the secretion of ECP and TGF-β1, the amount of eosinophils and epithelial cells were doubled (100 000 eosinophils/well and 100 000 or 50 000 epithelial cells/well).

Eosinophil peroxidase (EPO) assay

Peroxidase activity in standard and test supernatant samples was determined as previously described.16 In brief, a reaction solution composed of 4 µl 30% H2O2 plus 10 mg o-phenylendiamine (OPD; Sigma-Aldrich) dissolved in citrate buffer pH 4·5 with EDTA and hexadecyl-trimethyl-ammonium-bromide was added to samples in a microwell plate, and incubated in darkness for 10 min at room temperature. The reaction was stopped by the addition of 1 m H2SO4 and the optical density was analysed with an enzyme-linked immunosorbent assay (ELISA) reader (Labsystems Multiscan MS, Labsystems, Finland) at 492 nm. The total amount of EPO in intact eosinophils was determined in lysates of 50 000 unstimulated eosinophils, which were serially diluted twofold to construct a standard curve. The quantity of EPO in supernatants was translated into numbers of completely degranulated eosinophils. However, in vivo, human eosinophils degranulate partially; it is therefore likely that the numbers of eosinophils that release granular contents extracellulary are underestimated. The EPO contents in test sample supernatants were derived from this curve, using the software program Graph Pad Prism 2·0 (San Diego, CA).

ECP assay

ECP contents in supernatants were determined with a semiautomated enzyme immunoassay using fluorochrome-labelled antibodies (UniCAP 100 Pharmacia, Södertälje, Sweden).

FGF-1 and -2 ELISA

Flat-bottomed 96-well polystyrene plates (Nunc A/S) were coated overnight with 0·5 µg/ml rabbit polyclonal antibodies (PAb) anti-human FGF-1 or -2 (BioSite, Täby, Sweden) diluted in PBS. The plates were blocked with 1% BSA–PBS for 1·5 h at room temperature. Recombinant hFGF-1 or -2 (BioSite) was serially diluted (20–0·15 ng/ml) in diluent (0·05% Tween-20–0·1% BSA–PBS) to construct a standard curve. Supernatants and standard samples were added in a volume of 70 µl to each well and incubated for 2 hr at 37°. Next, 100 µl of biotinylated rabbit anti-hFGF-1 or -2 PAb (Biosite) at final concentration of 0·15 µg/ml was added to each well and incubated for 1·5 hr at room temperature. Streptavidin–horseradish peroxidase diluted 1 : 10 000 in high-performance ELISA buffer (HPE; CLB, Amsterdam, the Netherlands) was added to the plate and incubated for 30 min at room temperature. Between each incubation step, the plates were washed four times with 0·05% Tween-20–PBS. Finally, the reaction solution, 10 mg tetramethylbenzidine (Sigma-Aldrich) dissolved in 1 ml dimethyl sodium oxide (Sigma-Aldrich) and 9 ml phosphate citrate buffer pH 5·0 was added and the absorbance measured using an ELISA reader at 450 nm. The software program Graph Pad Prism was used to calculate FGF-1 and -2 contents.


The contents of TGF-β1 in supernatants were measured by Quantikine human TGF-β1 Immunoassay kit (R & D Systems Europe, Ltd, Abingdon, UK). In short, samples were activated by acidification using 1 m HCl and neutralized with 1·2 m NaOH, 0·5 m N-(2-hydroxyethylpiperazine-N′-2-ethane sulphonic acid, and a sandwich ELISA was performed according to the manufacturer's instructions.

Endotoxin assay

To analyse endotoxin contents in the epithelial cell suspensions, a Limulus amoebocyte lysate Coamatic® Chromo-LAL kit (Associates of Cape Cod, Inc., Falmouth, MA) was used according to the manufacturer's instructions. Briefly, colyophilized Limulus amoebocyte lysate and substrate were mixed with test samples in a microplate and incubated in an ELISA reader at 37°. Absorbance was measured at 405 nm every 30 s, and the time (onset time) needed for a sample to reach absorbance 0·1 (onset OD) was calculated, using a SOFTmax program (Molecular Devices Corp., Menlo Park, CA).


Median values and statistical significance (unpaired two-tailed Student's t-test, P < 0·05) were calculated using the software program Graph Pad Prism 2·0.


Our aim was to assess the capacity of necrotic epithelial cells to induce chemotactic movement in eosinophils, to activate them to become effector cells, and to establish if thus stimulated eosinophils express tissue-destructive or tissue-healing activity or both.

Freeze–pressing is the best method to damage epithelial cells for recognition by human eosinophils

Epithelial cells originating from the gastrointestinal (HT29), genitourinary (HeLa) and respiratory tracts (HEp-2) were treated by freeze–pressing, heat-treatment and freeze–thawing to induce danger signals (Fig. 1). Next, damaged epithelial cells were coincubated with eosinophils. The freeze–pressing procedure was originally designed to disintegrate micro-organisms, and proved an efficient method to make epithelial cells necrotic. After this treatment, cells were completely disintegrated and the resulting preparation was composed of cell membranes of various sizes, which tended to form clusters (Fig. 1a). All freeze–pressed cells were necrotic. The freeze-thaw procedure appeared to leave a few percent of cells intact, but most of the preparation consisted of cell membrane fractions, some of which formed conglomerates (Fig. 1b) The mildest treatment was heating at 56°, which rendered about half of the epithelial cells necrotic as determined by Trypan blue exclusion (Fig. 1c). Whereas heat-treated and freeze–thawed epithelial cells gave rise to a modest eosinophil activation (not shown), freeze–pressed epithelial cells were outstanding activators of eosinophils. Therefore, all subsequent experiments used this treatment to generate danger signals derived from epithelial cells.

Figure 1
Various methods to generate stressed or necrotic epithelial cells. Freeze–pressed HT29 cells (a), freeze–thawed HEp-2 cells (b), trypan-blue-stained heat-treated HT29 cells (c)and viable HT29 cells in solution (d). ×400 magnification. ...

Necrotic epithelial cells elicit chemotactic movement in human eosinophils

Chemotaxis experiments were performed, in which the established eosinophil chemoattractant eotaxin as well as various concentrations of necrotic and viable HEp-2, HeLa and HT29 cells were tested regarding their capacity to attract eosinophils in a microwell dual chamber system. Necrotic HT29 cells proved to be the most potent chemoattractants among the tested epithelial cell lines. At the highest dose tested, they attracted 15% of the eosinophils (Fig. 2a). A dose–response relation was apparent, such that increasing numbers of necrotic HT29 cells gave rise to higher chemotactic index (Fig. 2a). Hence, necrotic HT29 cells attracted a statistically significant higher percentage of eosinophils than the equivalent concentration of undamaged HT29 cells both at the concentrations corresponding to 1 epithelial cell per eosinophil (30 000 cells, P = 0·024) and one epithelial cell per two eosinophils (15 000 cells, P = 0. 009; Fig. 2a). No chemotactic activity was seen at the lowest concentration tested, 1 epithelial cell per 10 eosinophils (3000 cells) (Fig. 2a). Although the highest concentration of necrotic HEp-2 cells attracted on average 7% eosinophils, this did not reach statistical significance (P = 0·15) when comparing with undamaged HEp-2 cells (Fig. 2b). HeLa cells, both damaged and undamaged, gave rise to a modest eosinophilic migration. No statistical significance was seen between the two HeLa cell preparations (Fig. 2b).

Figure 2
Necrotic epithelial cells induce eosinophil chemotaxis. The percentage of migrated eosinophils above spontaneous migration is shown. 100% migration corresponds to 30 000 cells, i.e. the total amount of eosinophils used per well. Eosinophils (30 000 cells) ...

In an effort to determine which receptors on the eosinophil surface were engaged by the unidentified chemotactic factors in the freeze–pressed HT29 cells, blocking experiments using the formyl peptide receptor antagonist cyclosporin H were performed. A 53% inhibition in chemotactic movement was seen when cyclosporin H was used compared to the medium control (Fig. 3).

Figure 3
Cyclosporin H (CyH) partly inhibits chemotactic movement of eosinophils towards necrotic epithelial cells. The percentage of migrated eosinophils above spontaneous migration is shown. Eosinophils (30 000 cells) from seven blood donors were stimulated ...

Eosinophils release granule constituents upon stimulation with necrotic epithelial cells

Necrotic HT29, HEp-2 and HeLa cells were tested for their ability to cause degranulation of eosinophils, i.e. release of EPO and ECP. Necrotic HeLa cells induced EPO release from eosinophils in a dose-dependent manner (Fig. 4a) Thus, the two highest concentrations of necrotic HeLa cells stimulated eosinophils to release statistically significant amounts of EPO compared to medium. Viable HeLa cells gave rise to negligible EPO secretion (Fig. 4a). Surprisingly, viable and necrotic HEp-2 cells alike could stimulate eosinophils to release EPO. The same was seen for HT29 cells (Fig. 4b). Viable and necrotic epithelial cells alone did not catalyse the reaction with H2O2 (data not shown).

Figure 4
Viable and necrotic epithelial cells cause eosinophilic release of EPO. EPO release into supernatants in eosinophil cultures from nine different blood donors stimulated with viable and necrotic HeLa cells (a). EPO contents in supernatants of eosinophils ...

Necrotic epithelial cells derived from HeLa as well as HT29 cell lines stimulated eosinophils to release significant amounts of ECP at both the concentration corresponding to 1 epithelial cell per eosinophil (HT29, P = 0·002; HeLa, P = 0·023) and the concentration corresponding to 1 epithelial cell per 2 eosinophils (HT29, P = 0·0001; HeLa, P = 0·010; Fig. 5). Only background levels of ECP were released when eosinophils were incubated with whole ‘viable’ epithelial cells. Hence, in contrast to EPO release, which was elicited by both damaged and whole viable cells, albeit at different levels, ECP release was only accomplished by necrotic epithelial cells.

Figure 5
Necrotic epithelial cells stimulate eosinophils to release ECP. Eosinophils from five blood donors were stimulated 1 : 1 (100 000 cells) and 1 : 2 (50 000 cells) with necrotic and viable HT29 and HeLa cells. All data represent the arithmetic mean ± ...

Eosinophils release growth factors upon stimulation with necrotic epithelial cells

Since epithelial cells can synthesize FGF-1 and -2, it was necessary to discriminate between FGF excreted by eosinophils from that produced by necrotic epithelial cells. Thus, FGF contents were determined in supernatants derived from wells incubated with whole or necrotic epithelial cells alone, eosinophils alone, and the two combined. Eosinophils incubated in medium alone released no FGF-1 or -2. The amount of FGF produced by epithelial cells was subtracted from the amount of FGF detected in cocultures of eosinophils with necrotic or viable epithelial cells. Necrotic HT29 and HeLa cells alike stimulated eosinophils to release significant amounts of FGF-2 whereas no FGF-2 was detected when eosinophils were incubated with viable epithelial cells, nor the lowest dose of necrotic cells (Fig. 6a and b). Eosinophils did not release any FGF-1 upon stimulation with necrotic epithelial cells (not shown).

Figure 6
Fibroblast growth factor-2 is released from eosinophils stimulated with necrotic epithelial cells. FGF-2 contents in eosinophil supernatants from 6 blood donors after incubation with HT29 (a)and HeLa cells (b), as determined by sandwich ELISA (cut-off ...

Epithelial cells can also synthesize TGF-β1. Therefore, TGF-β1 generated by eosinophils and necrotic epithelial cells, respectively, was analysed using the same subtraction procedure as described above. Again, no spontaneous release of TGF-β1 was seen in eosinophils incubated in medium alone. A modest release of TGF-β1 was noted in eosinophils derived from four to five of seven blood donors, incubated with viable or necrotic HT29 cells (Fig. 7). In contrast, a similarly moderate excretion of TGF-β1 was only seen in eosinophils from one of five tested blood donors, in response to stimulation with viable or necrotic HeLa cells (Fig. 7). There were no significant differences (P = 0·29) in TGF-β1 release between eosinophils stimulated with necrotic and viable HT29 cells (Fig. 7). Likewise only small amounts of TGF-β1 could be detected in cocultures of eosinophils with necrotic or viable HeLa cells (Fig. 7).

Figure 7
TGF-β1 is released from eosinophils stimulated with necrotic and viable epithelial cells. TGF-β1 in supernatants of eosinophils stimulated with HT29 and HeLa cells, n = 5 blood donors. Horizontal bars denote the median amounts (pg/ml) ...

In the absence of serum, LPS does not activate eosinophils

All epithelial cell suspensions were shown to contain various low amounts of endotoxin. We tested the capacity of purified eosinophils to respond to a range of LPS concentrations (microgram to nanogram) and found that without the addition of human serum, no degranulation or chemotaxis was seen even after exposure to microgram amount of LPS. HEp-2 cells, both viable (4 ng LPS/ml) and necrotic (7 ng LPS/ml), were not more potent activators of eosinophils despite containing about 50 times more endotoxin than the HT29 and HeLa cell preparations (0·1 ng LPS/ml). Thus, the described eosinophil activation cannot be attributed to endotoxin contamination in our serum-free system.


Since the end of the 19th century, it has been known that a variety of diseases such as asthma, acute and chronic skin diseases, helminthic infections, malignant tumours, postfebrile states and reactions to drugs may cause eosinophils to accumulate in the blood and/or tissues.17 Today the list can be made even longer by adding intestinal inflammatory conditions such as Shigella dysentery,10 ulcerative colitis,18 severe celiac disease,9 and acute intestinal graft-versus-host disease.19 Although the common denominator between these various pathological entities is unknown, tissue destruction is one possibility. However, whether this is directly caused by the eosinophils, by other agents, or the two combined is not clear.

The danger model postulates that substances derived from eukaryotic non-immune cells that are stressed or damaged can activate cells of the innate immune system.2 The aim of this study was to see if human eosinophils could recognize and be activated by danger signals derived from damaged epithelial cells. Our results show that epithelial cells subjected to freeze–pressing were potent activators of human eosinophils derived from healthy individuals. This treatment destroyed the majority of cells and was the harshest one tested. However, in some instances, what we perceived to be whole, viable cells, also gave rise to modest eosinophil activation. It may be that these cells, detached from the culture flasks, expressed stress signals that were recognized by the eosinophils.

Eosinophilic accumulation and degranulation in the body has for many years been associated with tissue destruction. Hence, eosinophils have been assigned the role of ‘bad guys’. It has even been suggested that a dose–response relationship exists such that higher numbers of tissue-bound eosinophils would entail more severe tissue injury.19 Our findings also indicate that there is a relationship between degree of eosinophil activation and dose of necrotic epithelial cells. However, few studies have attempted to answer the question what comes first, tissue damage or the eosinophils. Erjefält et al. observed that eosinophils in the tracheal epithelial basement membrane of guinea pigs moved towards damaged epithelial areas in the lumen.20 This might indicate that something within the damaged tissue attracted the eosinophils, i.e. tissue damage may have preceded eosinophilic infiltration.

Our study reveals that eosinophils exposed to necrotic epithelial cells derived from the intestinal or genital mucosa released both EPO and ECP. EPO is a heme enzyme that catalyses reactions that form reactive halogen species and NO-derived oxidants,21,22 whereas, ECP belongs to the RNAse gene superfamily. Although both enzymes and their catbolites have been reported to be toxic to mammalian cells,23,24 it is not reasonable to assume that eosinophils respond to necrotic cells by releasing highly toxic products into an already damaged tissue. Probably, these enzymes are endowed with additional properties of importance to help limit or abrogate processes that initiate or perpetuate tissue damage in the human body.

Recently, it has been proposed that eosinophils may also be involved in tissue repair since they can elaborate growth factors such as FGF-2,25 nerve growth factor,26 vascular endothelial growth factor27 and TGF-β128 Therefore, we wanted to investigate if damaged epithelial cells could stimulate eosinophils to produce and release growth factors. FGF-2 is thought to stimulate endothelial cell proliferation, haematopoiesis, angiogenesis, as well as migration and differentiation of cells of mesenchymal and neuroectodermal origin.29 Our results show that necrotic epithelial cells stimulated eosinophils to release FGF-2. This supports the theory that eosinophils may be involved in tissue healing. TGF-β1 is able to promote the growth of some mesenchymal cells while inhibiting the proliferation of most other cell types, possesses immunosuppressive effects and can enhance the formation of extracellular matrix.30 Several studies point out that eosinophils from patients with asthma or chronic bronchitis express TGF-β1 as determined by in situ hybridization or immunohistochemistry.31,32 A recent study in mice shows that there is a correlation between the presence of eosinophils, TGF-β1 and the prevention of airway hyperreactivity.33 We found that cells of intestinal origin, both cells we perceived to be viable as well as necrotic ones, caused a modest TGF-β1 release from eosinophils in the majority of blood donors. HeLa cells, both damaged and undamaged, were inferior to HT29 cells in this regard.

Because the epithelial cell lines used in this study have a cancerous origin, we cannot exclude that cancer-specific molecules may have activated the eosinophils. Eosinophils are known to infiltrate the tissue in certain solid tumours, e.g. cervical, lung and colonic carcinomas, tumours engaging the predilection organs of the eosinophil.34 It has even been suggested that the presence of eosinophils in tumour tissue may indicate a more favourable outcome regarding the survival of the patient.35 Cancer-specific eosinophil chemotactic factors have been reported in lung squamous cell carcinomas36 and cervical cancer.37 However, there are reports indicating that the accumulation of eosinophils in tumours may constitute a response to the tissue damage created by the growing tumour, rather than to the tumour antigens per se.38,39 Therefore, we believe that the putative danger molecules responsible for the eosinophil activation observed in the present study are a general feature of all epithelial cells, cancerous and non-cancerous.

A striking finding in our study was that necrotic epithelial cells derived from the large intestine were more potent than cells from the uterine cervix regarding the capacity to activate eosinophils. The majority of eosinophils in the human body are found in the intestine.40 It is possible that eosinophils have become adapted to recognize necrotic intestinal cells. Exactly how colon-derived epithelium differs from cervical epithelium is unclear, but it is likely that they are not identical regarding intracellular contents. Unfortunately, it came to our knowledge that the HEp-2 cell line believed to be representative of the airway tract may be a HeLa cell subline.41 Nevertheless, HEp-2 cells were the least potent eosinophil activators.

Different studies have shown that in a number of diseases associated with tissue eosinophilia, the epithelial cells are an important source of eosinophil chemoattractants such as eotaxin, RANTES, MCP-3 and -442 It is possible that some of these molecules are also responsible for the eosinophil mobilization noted in our study. However, as the necrotic cells used were destroyed immediately after they were detached from the cell culture bottles, there was no time for the cells to have de novo synthesized or up-regulated any putative molecule(s) responsible for the observed eosinophil activation. Rather, these undefined danger signals were likely stored inside the epithelial cells. It has been reported that adenosine triphosphate (ATP), an intracellular component released by damaged cells, was able to activate another type of leucocyte, dendritic cells, to maturate and release large amounts of IL-12 in the presence of TNF-α.43 It is possible that eosinophils could be activated by this nucleotide in conjunction with TNF-α, since activated eosinophils express ATP receptors (P2 purinergic receptors) and are able to release intracellulary stored TNF-α.44,45 It has also been reported that neutrophils migrate towards mitochondrial N-formylated proteins.46 Eosinophils have formyl peptide receptors and could thus be activated by formylated proteins of mitochondrial origin.16 In fact, we found that cyclosporin H, an antagonist of formyl peptide receptors, partly inhibited the chemotactic movement of eosinophils towards necrotic epithelial cells, suggesting this class of receptors may be involved.

Because eosinophils express Toll-like receptor-4 as well as CD14, they could respond to LPS.47,48 It was important to exclude the possibility that the observed eosinophil activation by necrotic cells was due to endotoxin contamination of the epithelial cell lysates. However, as we were unable to activate the eosinophils with LPS in the absence of serum, and our assay system is serum-free, it seems unlikely that the observed degranulation and chemotaxis caused by damaged epithelial cells could be attributed to endotoxin contamination.

In conclusion, danger signals released from necrotic epithelial cells were able to activate human eosinophils. However, thus activated eosinophils released both granule constituents with documented toxic potential as well as growth factors. Thus, it was not apparent in our system if necrotic cells skewed eosinophils towards the release of tissue-damaging or -healing substances. Epithelial cell damage may be the common denominator between the broad ranges of disease entities in which eosinophils are present. To our knowledge, this is the first demonstration that human eosinophils directly recognize and are activated by necrotic epithelial cells. In fact, Paul Ehrlich in the late 1800s had already formulated the danger theory for eosinophils. He argued that two categories of substances activated eosinophils: products of tissue breakdown or toxic products of foreign organisms.17 In the future, we would like to define which components of damaged epithelial cells actually constitute danger signals and which receptors they engage on the eosinophil surface. Hopefully, such studies may help resolve the role played by these elusive cells in health as well as in disease.


This study was supported by grants from the Swedish Research Council (K2002-16X-14180-01A), LUA-SAM (I33914), Wilhelm and Martina Lundgren's Science Fund, Adlerbertska Research Foundation, Magnus Bergvall Foundation, Lars Hierta Foundation, Swedish Medical Society and Göteborg Medical Society.

We greatly appreciate the help provided by Professor Lars Edebo, Department of Clinical Bacteriology, Göteborg University, with the freeze–pressing experiments.


adenosine triphosphate
bovine serum albumin
eosinophil cationic protein
ethylenediaminetetraacetic acid
eosinophil peroxidase
fibroblast growth factor
Kreb's Ringer glucose
magnetic-activated cell sorting
monocyte chemotactic protein
minimal essential medium
polyclonal antibody
phosphate-buffered saline
regulated upon activation normal T-cell expressed and secreted
tumour necrosis factor


Janeway CA., Jr The immune system evolved to discriminate infectious nonself from noninfectious self. Immunol Today. 1992;13:11–6. [PubMed]
Matzinger P. Tolerance, danger, and the extended family. Annu Rev Immunol. 1994;12:991–1075. [PubMed]
Collins PD, Marleau S, Griffiths-Johnson DA, Jose PJ, Williams TJ. Cooperation between interleukin-5 and the chemokine eotaxin to induce eosinophil accumulation in vivo. J Exp Med. 1995;182:1169–74. [PMC free article] [PubMed]
Dunzendorfer S, Wiedermann CJ. Neuropeptide-induced chemotaxis of eosinophils in pulmonary diseases. Ann Med. 2000;32:429–39. [PubMed]
Kita H, Gleich GJ. Chemokines active on eosinophils: potential roles in allergic inflammation. J Exp Med. 1996;183:2421–6. [PMC free article] [PubMed]
Steinbach KH, Schick P, Trepel F, et al. Estimation of kinetic parameters of neutrophilic, eosinophilic, and basophilic granulocytes in human blood. Blut. 1979;39:27–38. [PubMed]
Rosenberg H. Eosinophils. In: Gallin JI, Snyderman R, editors. Inflammation: Basic Principles and Clinical Correlates. 3. Philadelphia: Lippincott Williams & Wilkins; 1999. pp. 61–76.
Rutgeerts P, Geoboes K, Vantrappen G, Kerremans R, Coenegrachts JL, Coremans G. Natural history of recurrent Crohn's disease at the ileocolonic anastomosis after curative surgery. Gut. 1984;25:665–72. [PMC free article] [PubMed]
Marsh MNHJ. Inflammatory component of celiac sprue mucosa. 1. Mast cells, basophils, and eosinophils. Gastroenterol. 1985;89:92–101. [PubMed]
Mathan MM, Mathan VI. Morphology of rectal mucosa of patients with shigellosis. Rev Infec Dis. 1991;13:314–8. [PubMed]
Saisu K, Morozumi K, Fujita K. Significance of interstitial lesions as the early indicator for acute vascular rejection in human renal allografts. Clin Transplant. 1999;13(Suppl. 1):17–23. [PubMed]
Shulman HM, Sullivan KM, Weiden PL, et al. Chronic graft-versus-host syndrome in man. A long-term clinicopathologic study of 20 Seattle patients. Am J Med. 1980;69:204–17. [PubMed]
Todd R, Donoff BR, Chiang T, Chou MY, Elovic A, Gallagher GT, Wong DT. The eosinophil as a cellular source of transforming growth factor alpha in healing cutaneous wounds. Am J Pathol. 1991;138:1307–13. [PMC free article] [PubMed]
van Kuppeveld FJ, Johansson KE, Galama JM, Kissing J, Bolske G, van der Logt JT, Melchers WJ. Detection of mycoplasma contamination in cell cultures by a mycoplasma group-specific PCR. Appl Environ Microbiol. 1994;60:149–52. [PMC free article] [PubMed]
Magnusson KE, Edebo L. Large-scale disintegration of microorganisms by freeze-pressing. Biotechnol Bioeng. 1976;18:975–86. [PubMed]
Svensson L, Dahlgren C, Wennerås C. The chemoattractant Trp-Lys-Tyr-Met-Val-D-Met activates eosinophils through the formyl peptide receptor and one of its homologues, formyl peptide receptor-like1. J Leukoc Biol. 2002;72:810–8. [PubMed]
Hirsch JG, Hirsch B. Paul Ehrlich and the discovery of the eosinophil. In: Mahmoud AAF, Austen KS, editors. The Eosinophil in Health and Disease. New York: Grune & Stratton; 1980. pp. 3–23.
Raab YFK, Gerdin B, Hällgren R. Eosinophil activation in ulcerative colitis. Dig Dis Sci. 1998;43:1061–70. [PubMed]
Daneshpouy M, Socie G, Lemann M, Rivet J, Gluckman E, Janin A. Activated eosinophils in upper gastrointestinal tract of patients with graft-versus-host disease. Blood. 2002;99:3033–40. [PubMed]
Erjefalt JS, Korsgren M, Nilsson MC, Sundler F, Persson CG. Association between inflammation and epithelial damage-restitution processes in allergic airways in vivo. Clin Exp Allergy. 1997;27:1344–55. [PubMed]
Furtmuller PG, Burner U, Regelsberger G, Obinger C. Spectral and kinetic studies on the formation of eosinophil peroxidase compound I and its reaction with halides and thiocyanate. Biochemistry. 2000;39:15578–84. [PubMed]
MacPherson JC, Comhair SA, Erzurum SC, Klein DF, Lipscomb MF, Kavuru MS, Samoszuk MK, Hazen SL. Eosinophils are a major source of nitric oxide-derived oxidants in severe asthma: characterization of pathways available to eosinophils for generating reactive nitrogen species. J Immunol. 2001;166:5763. [PubMed]
Gleich GJ. Loegering DA. Immunobiology of eosinophils. Ann Rev Immunol. 1984;2:429–59. [PubMed]
Venge P, Bystrom J, Carlson M, Hakansson L, Karawacjzyk M, Peterson C, Seveus L, Trulson A. Eosinophil cationic protein (ECP): molecular and biological properties and the use of ECP as a marker of eosinophil activation in disease. Clin Exp Allergy. 1999;29:1172–86. [PubMed]
Hoshino M, Takahashi M, Aoike N. Expression of vascular endothelial growth factor, basic fibroblast growth factor, and angiogenin immunoreactivity in asthmatic airways and its relationship to angiogenesis. J Allergy Clin Immunol. 2001;107:295–301. [PubMed]
Solomon A, Aloe L, Peter J, Frucht-Pery J, Bonini S, Levi-Schaffer F. Nerve growth factor is preformed in and activates human peripheral blood eosinophils. J Allergy Clin Immunol. 1998;102:454–60. [PubMed]
Horiuchi TWPF. Expression of vascular endothelial growth factor by human eosinophils. Am J Respir Cell Mol Biol. 1997;17:70. [PubMed]
Phipps S, Ying S, Wangoo A, Ong YE, Levi-Schaffer F, Kay AB. The relationship between allergen-induced tissue eosinophilia and markers of repair and remodeling in human atopic skin. J Immunol. 2002;169:4604–12. [PubMed]
Okada-Ban M, Theiery JP, Jouanneau J. Fibroblast growth factor-2. J Biochem Cell Biol. 2000;32:263–7. [PubMed]
Lawrence DA. Transforming growth factor-beta: a general review. Eur Cytokine Netw. 1996;7:363–74. [PubMed]
Ohno I, Nitta Y, Yamauchi K, et al. Transforming growth factor beta 1 (TGF beta 1) gene expression by eosinophils in asthmatic airway inflammation. Am J Respir Cell Mol Biol. 1996;15:404–9. [PubMed]
Vignola AM, Chanez P, Chiappara G, et al. Transforming growth factor-beta expression in mucosal biopsies in asthma and chronic bronchitis. Am J Respir Crit Care Med. 1997;156:591–9. [PubMed]
Kobayashi T, Iijima K, Kita H. Marked airway eosinophilia prevents development of airway hyper-responsiveness during an allergic response in IL-5 transgenic mice. J Immunol. 2003;170:5756–63. [PubMed]
Pretlow TP, Keith EF, Cryar AK, Bartolucci AA, Pitts AM, Pretlow TG, Kimball PM, Boohaker EA. Eosinophil infiltration of human colonic carcinomas as a prognostic indicator. Cancer Res. 1983;43:2997–3000. [PubMed]
Bethwaite PBHLJ, Yeong ML, Thornton A. Effect of tumor associated tissue eosinophilia on survival of women with stage IB carcinoma of the uterine cervix. J Clin Pathol. 1993;46:1016–20. [PMC free article] [PubMed]
Goetzl EJ, Tashijan AH, Rubin RH, Austen KF. Production of a low molecular weight eosinophil polymorphonuclear leukocyte chemotactic factor by anaplastic squamous cell carcinomas of human lung. J Clin Invest. 1978;61:770–80. [PMC free article] [PubMed]
Inoue TIK, Chihara T, Torisu M. Cervical carcinoma and eosinophils. Clin Immunol Immunopathol. 1981;21:320–31. [PubMed]
Blumenthal RD, Samoszuk M, Taylor AP, Brown G, Alisauskas R, Goldenburg DM. Degranulating eosinophils in human endometrosis. Am J Pathol. 2000;156:1581–8. [PMC free article] [PubMed]
Samoszuk M, Lin F, Rim P, Strathearn G. New marker for blood vessels in human ovarian and endometrial cancers. Clin Cancer Res. 1996;2:1867–71. [PubMed]
Rothenberg ME, Mishra A, Brandt EB, Hogan SP. Gastrointestinal eosinophils. Immunol Rev. 2001;179:139–55. [PubMed]
Masters JR. HeLa cells 50 years on: the good, the bad and the ugly. Nat Rev Cancer. 2002;2:315–9. [PubMed]
Ying S, Meng Q, Zeibecoglou K, Robinson DS, Macfarlane A, Humbert M, Kay AB. Eosinophil chemotactic chemokines (eotaxin, eotaxin-2, RANTES, monocyte chemoattractant protein-3 (MCP-3), and MCP-4), and C-C chemokine receptor 3 expression in bronchial biopsies from atopic and nonatopic (intrinsic) asthmatics. J Immunol. 1999;163:6321–9. [PubMed]
Schnurr M, Then F, Galambos P, Scholz C, Siegmund B, Endres S, Eigler A. Extracellular ATP and TNF-alpha synergize in the activation and maturation of human dendritic cells. J Immunol. 2000;165:4704–9. [PubMed]
Mohanty JG, Raible DG, McDermott LJ, Pelleg A, Schulman ES. Effects of purine and pyrimidine nucleotides on intracellular Ca2+ in human eosinophils: activation of purinergic P2Y receptors. J Allergy Clin Immunol. 2001;107:849–55. [PubMed]
Costa JJ, Matossian K, Resnick MB, et al. Human eosinophils can express cytokines tumor necrosis factor-alpha macrophage inflammatory protein-1 alpha. J Clin Invest. 1993;91:2673–84. [PMC free article] [PubMed]
Carp H. Mitochondrial N-formylmethionyl proteins as chemoattractants for neutrophils. J Exp Med. 1982;155:264–75. [PMC free article] [PubMed]
Plotz SG, Lentschat A, Behrendt H, et al. The interaction of human peripheral blood eosinophils with bacterial lipopolysaccharide is CD14 dependent. Blood. 2001;97:235–41. [PubMed]
Nagase H, Okugawa S, Ota Y, et al. Expression and function of toll-like receptors in eosinophils: activation by Toll-like receptor 7 ligand. J Immunol. 2003;171:3977–82. [PubMed]

Articles from Immunology are provided here courtesy of British Society for Immunology
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...