Abstract
A common feature of some parasitic infections and allergic and atopic skin diseases is the involvement of Th2 lymphocytes and the dermal appearance of eosinophils (Eos). Because Th2 lymphocytes apparently do not release Eo attractants, we addressed the question of whether the Th2 cytokine IL-4 induces its production in dermal fibroblasts. We therefore stimulated fibroblasts with IL-4. HPLC investigation of supernatants revealed a single Eo chemotactic protein, which was purified to homogeneity giving a single 13-kDa band upon SDS-PAGE analyses. Peptide mapping with subsequent amino acid sequencing revealed an Eo-selective chemotaxin, which consists of a mixture of N-terminally truncated and O-glycosylated forms of the chemokine eotaxin. Other chemokines such as RANTES, MCP-3, MCP-4, or MIP-1α were not detected as Eo chemotaxins under these conditions. Using reverse transcriptase-PCR techniques, we found that IL-4 dose and time dependently induces eotaxin mRNA in dermal fibroblasts. Stimulation with IL-4 and TNF-α caused a 10- to 20-fold increase of the release of three biochemically different eotaxin forms, each consisting of a mixture of N-terminally truncated and O-glycosylated variants having the same backbone amino acid sequence but different specific activities. Our findings support the hypothesis that eosinophil recruitment seen in IL-4-mediated skin reactions, at least in part, may be due to Th2 cytokine-mediated induction of eotaxin in dermal fibroblasts.
Eosinophils (Eos)4 are effector cells having their role in host defense against helminthic parasites and may contribute to the acute and late manifestations, including tissue damage of allergic diseases such as bronchial asthma, allergic rhinitis, and atopic dermatitis (1, 2, 3).
For unknown reasons Eos infiltrate the dermis of patients with atopic dermatitis after patch test reactions with aeroallergens (4). In lesional atopic dermatitis skin, dermal deposits of the Eo major basic protein have been demonstrated, suggesting that Eos degranulate in the affected skin (5).
Because Eos predominantly infiltrate the dermis rather than epidermis in all the above-mentioned skin diseases, including parasite infection, it is intriguing to speculate that dermal cells may contribute to Eo infiltration through the production of Eo attractants.
Elevated IgE level, activation of mast cells, basophils, Th2-type CD4+ T lymphocytes, and Eo tissue infiltration are known to be strongly associated with helminth parasite infections (6, 7). Key mediators produced by Th2 lymphocytes for defense against helminths are IL-5 and IL-4.
Whereas IL-5 mainly promotes blood eosinophilia (8), IL-4 induces IgE synthesis (9) and endothelial vascular cell adhesion molecule-1 expression (10), which, together with β1 integrin very late antigen-4 expression on eosinophils (11), may increase Eo adherence to the vessels. It does not, however, explain Eo tissue accumulation. Therefore we speculate that, in addition, an Eo-specific chemoattractant such as eotaxin (12, 13, 14) is involved.
This raises the possibility that fibroblasts, which represent the major cellular component of the dermis, are capable of producing Eo chemotactic factors after treatment with a Th2 cytokine. In this study we investigated whether cultured human foreskin-derived fibroblasts have the capability to produce Eo attractants when stimulated with the Th2 cytokine IL-4.
Material and Methods
Culture of dermal fibroblasts
Human surgically removed skin (foreskin) was incubated with a mixture of 0.05% (w/v) trypsin and 0.02% (w/v) EDTA over night at 4°C, and epidermis was removed. Thereafter, dermis was cut into pieces of about 2 mm in diameter. These were transferred into 25-mm2 flasks and supplied with MEM (Life Technologies-AMIMED, Allschwil, Switzerland) supplemented with penicillin (10 U/ml), streptomycin (100 μg/ml), glutamine (2 mM), and 10% FCS. When dermal fibroblasts had started growth, dermal pieces were removed. Medium was changed once a week. Passage was performed by the use of a mixture of 0.05% (w/v) trypsin and 0.02% EDTA for 5 min at 37°C. Fibroblasts from culture of the 5th to 10th passages were transferred either into six-well plates (9 cm2/well) or, for preparative purposes, into 175-cm2 or 500-cm2 flasks (Nunc, Wiesbaden, Germany) and were used at confluence.
Isolation of Eos
Eosinophils were isolated from human peripheral blood of normal donors having mild eosinophilia using Percoll density gradient centrifugation as previously described (15). Eo preparations used for experiments contained more than 90% Eos.
Eo chemotaxis assay
Screening of cell culture supernatants and HPLC fractions for eosinophil chemotactic activity was assayed as previously described (15) using blind well Boyden chambers (Bio-Rad, Munich, Germany), which were filled with samples at appropriate dilutions and covered with a polyvinylpyrrolidone-containing filter (pore size: 3 μm) (Costar, Tübingen, Germany). Chemotactic activity was expressed as chemotactic index: 
For screening of Eo chemotactic activity in HPLC fractions, aliquots (usually 10 μl for preparative HPLC and 1 to 5 μl for micro-HPLC) were lyophilized in the presence of 10 μl 0.1% BSA in PBS and used at appropriate dilutions.
In some experiments, Eos were determined microscopically using a modification of Boyden’s method, as previously described in detail (15).
Production of Eo chemotactic proteins by human fibroblasts
Confluent growing cultures of dermal fibroblasts (∼5 × 107 cells/flask) in 500-cm2 flasks were washed twice with MEM and subsequently stimulated for 48 h with 20 ng/ml human rIL-4 (Pepro Tech, London, UK) dissolved in 50 ml MEM/flask containing the supplements but not FCS. Thereafter supernatants were harvested and centrifuged and stored below −70°C until further use.
High performance liquid chromatography
HPLC was performed at room temperature with a Spectra Physics liquid chromatography system equipped with a pump, a SP 8700 solvent delivery system, a Kratos spectroflow 783 spectrophotometer (Kratos, Westwood, NJ), and a SP 4270 integrator.
Purification of Eo chemotactic proteins
Pooled supernatants (500 ml/experiment, supernatants of 10 500-cm2 flasks) of IL-4 (20 ng/ml)-stimulated dermal fibroblasts (corresponding to approximately 5 × 108 cells) were concentrated and diafiltered against Tris/citrate buffer (10 mM/10 mM), pH 8.0, by using an Amicon YM-3 filter (cut off: 3kDa), and clarified by centrifugation.
Concentrated supernatants were then applied to a prepacked heparin-Sepharose column (Hi Trap, Pharmacia, Uppsala, Sweden) (1 ml volume) equilibrated with 10 mM Tris/10 mM citrate buffer, pH 8.0, using a HPLC pump and a flow rate of 1 ml/min. Thereafter the column was washed with equilibration buffer until no UV (280 nm)-absorbing material appeared in the effluent.
Bound material was eluted using 1 M NaCl in equilibration buffer and the pooled fractions were concentrated and diafiltered against 0.1% TFA in water until pH 2 was reached.
Heparin-bound proteins of stimulated fibroblasts were then applied to a preparative wide pore RP-8-HPLC column (300-7 C8 Nucleosil, 250 × 12.6 mm, Macherey and Nagel, Düren, Germany), and proteins were eluted from the column using a gradient of increasing concentrations of acetonitrile in 0.1% aqueous TFA (flow: 3 ml/min). During the HPLC separation of proteins, absorbance was monitored at 215 nm. Integration values obtained by the peak integrator were used to determine the amounts of proteins eluting in a given peak. To convert integration units to protein concentrations, known amounts of ubiquitin (Sigma, St. Louis, MO) were used for calibration. Nearly 106 integration units corresponded to 1 μg ubiquitin.
Fractions collected from preparative RP-8-HPLC, containing Eo chemotactic activity, were concentrated when necessary using YM-3 filters or were directly applied to a micro-Mono S HPLC column attached to a Smart Micro-HPLC apparatus (Pharmacia) previously equilibrated with 50 mmol/L ammonium formate, pH 4.0, containing 25% (v/v) acetonitrile. Proteins were eluted with a gradient of increasing concentrations of NaCl (maximum: 1 mol/L) in equilibration buffer using a flow rate of 100 μl/min. During the HPLC runs, absorbance at 215 nm, 254 nm, and 280 nm, as well as conductivity, were monitored.
Fractions containing Eo chemotactic activity were directly applied to a micro-RP-18-HPLC column (Pharmacia) previously equilibrated with 0.1% aqueous TFA. Proteins were eluted with a gradient of increasing concentrations of acetonitrile as shown in the chromatograms.
SDS-PAGE analyses
SDS-PAGE for polypeptides was performed with commercially available Tricin-buffer containing discontinuous gels (“high density gels,” Pharmacia) and a “Phast” electrophoresis system (Pharmacia).
Proteins were fixed with 0.5% (v/v) glutaraldehyde containing 0.1% (w/v) sodium thiosulfate in 30% aqueous (v/v) ethanol containing 0.4 mol/L sodium acetate at pH 6.0 and were visualized by the use of a silver-staining kit (Sigma).
Peptide mapping experiments
From 2 to 4 μg of purified eotaxin variants or recombinant eotaxin (Pepro Tech) were treated with 100 ng of different endoproteinases (Lys C, EC 3.4.21.50; Asp N, EC 3.4.24.33; Glu C, EC 3.4.21.19) at appropriate conditions according to the instructions of the manufacturer (Boehringer Mannheim, Mannheim, Germany) in a volume of 40 to 100 μl. Thereafter samples were separated by micro RP-18 HPLC (Smart System, Pharmacia) and peaks were analyzed for UV absorbance at 215, 254, and 280 nm or sequenced and investigated for molecular masses or for presence of carbohydrates.
Amino acid sequence determinations
Amino acid sequences were determined by Edman degradation in an Applied Biosystems 476 A-pulsed liquid protein sequencer using reversed-phase (RP)-HPLC for phenylthiohydantoin-derivative detection.
Cysteine residues were confirmed after on filter reduction with tributylphosphine and alkylation with 4-vinylpyridine.
Matrix-assisted laser desorption ionization (MALDI) and electrospray ionization (ESI) mass spectrometry (MS)
A MALDI instrument (TOF SPER, V6-Fison Instruments, Manchester, UK) was used for mass determinations. A 10-mg/ml solution of α-cyano-4-hydroxy-cinnamic acid (Sigma) in aqueous 70% acetonitrile containing 0.1% TFA was used as matrix. For ESI MS analysis, a Trio-2 quadruple (VG Biotech-Fisons Instruments, Manchester, UK) was used. Samples were preconcentrated using a reverse-phase trapping system (16).
Glycan ELISA
For determination of carbohydrate chains in eotaxin preparations, a digoxigenin glycan ELISA kit (Boehringer Mannheim) was used. Briefly, different amounts (0 to 100 ng) eotaxin were bound to ELISA plates and, by use of the kit, investigated for the presence of carbohydrate chains, as described by the manufacturer. The detection limit was near 2 to 5 ng glycoprotein. In other experiments, instead of eotaxin HPLC fractions, off peptide-mapping experiments were used to localize peptide fragments containing carbohydrate linkage.
Semiquantitative duplex RT-PCR
Total RNA from dermal fibroblast was isolated by acidic guanidinium thiocyanate-phenol-chloroform extraction (17). One microgram of total RNA was reverse transcripted using an oligo T18 primer and standard reagents (Life Technologies, Eggenstein, Germany). Intron spanning sets of primers specific for GAPDH and eotaxin (18) were used to differentiate between genomic and cDNA templates. cDNA corresponding to 50 ng RNA served as template in a duplex PCR reaction containing 0.8 μM eotaxin-specific primers (forward primer: 5′-CCCAACCACCTGCTGCTTTAACCTG-3′; reverse primer: 5′-TGGCTTTGGAGTTGGAGATTTTTGG-3′) and (as internal control for equal amounts of cDNA before PCR) 0.1 μM GAPDH-specific primer pair (18). Amplification and analysis of the PCR products were performed as described (18).
Results
IL-4 induces the release of an Eo selective chemotaxin
In initial experiments, we found that stimulation of cultivated dermal fibroblasts with 20 ng/ml IL-4 for 48 h resulted in the release of Eo chemotactic activity, whereas nonstimulated cells did not contain this activity. Biochemical characterization experiments revealed that Eo attractants completely bound to a heparin-Sepharose column. The nonbinding proteins, as well as lipids containing effluent, did not show any strong Eo chemotactic activity (data not shown).
Further purification of the heparin-binding Eo chemotactic proteins by preparative RP HPLC revealed early eluting fractions containing EO chemotactic activity (Fig. 1⇓). Other leukocytes, i.e., neutrophils or monocytes, did not respond to these HPLC fractions. This Eo chemotaxin was then further purified by the use of micro cation exchange HPLC followed by micro-RP HPLC (Fig. 2⇓). The purified material we termed eotaxin β showed a single band upon Tricine (Sigma) SDS-gel electrophoresis corresponding to a molecular mass of 13 kDa. Recombinant eotaxin as control protein gave a single band near 10 kDa. (Fig. 2⇓).
Fibroblasts release a single Eo-attractant upon IL-4 stimulation. Preparative RP HPLC of heparin-bound proteins secreted by IL-4 (20 ng/ml)-stimulated (48 h) fibroblasts. The shaded area represents fractions containing Eo chemotactic activity. RANTES ELISA determinations failed to show any RANTES immunoreactivity (not shown). A typical experiment is shown.
Purification of the Eo chemotaxin. Fractions of preparative RP-8 HPLC containing Eo chemotactic activity (Fig. 1⇑) were further purified by micro-Mono S HPLC (not shown). Fractions containing Eo chemotactic activity were finally purified by micro-RP-18 HPLC. SDS-PAGE analysis (inset) of the peak containing Eo chemotactic activity showed a single band near 13 kDa (lane 2), whereas recombinant eotaxin migrated as 10-kDa protein (lane 3). Lane 1 contained m.w. markers. A typical experiment is shown.
Structural characterization of eotaxin β
In order to obtain information about the structure of this chemotaxin, we performed N-terminal amino acid sequence analyses of the purified attractant and peptide-mapping experiments using different endoproteinases.
N-terminal sequence analyses revealed the presence of three N termini starting with Ala-Ser-Val (50%), Gly-Pro-Ala-Ser-Val (40%), and Ser-Val (10%), which correspond to the predicted N terminus of human eotaxin.
MALDI MS analyses, however, indicated the presence of a complex mixture of different proteins (Fig. 3⇓). Masses of the principal components were 9322 and 9020. Further analyses of the purified Eo attractant by peptide mapping with the endoproteinase Lys C and comparison of the pattern with that of recombinant eotaxin show some similarities as well as marked differences: natural eotaxin is missing the peaks 1 and 4 present in the recombinant material. Instead, four other peaks (numbers 1, 2, 4, and 5) can be seen in natural eotaxin (Fig. 4⇓). Sequence analyses of the peaks different in the patterns of natural and recombinant material revealed identical sequences for peaks having identical numbers and, in addition, peaks 2 and 5 in natural material have identical sequences as 1 and 4, respectively. The complete amino acid sequence of natural eotaxin produced by IL-4-stimulated dermal fibroblasts could be evaluated by comparing and aligning sequences obtained from different peptide-mapping experiments (Fig. 5⇓). Interestingly, in all experiments we were unable to determine residue 71. Peak 1 in recombinant eotaxin clearly represents a carboxyl-terminal fragment starting with serine, which contains the nonidentified residue 71 predicted to be threonine (Fig. 5⇓).
MALDI mass spectrum of purified eotaxin β. Purified Eo chemotaxin (Fig. 2⇑) was analyzed for molecular mass determination using MALDI MS: note the presence of numerous principal masses at 7882, 8478, 9020, and 9322.
Peptide mapping of natural and recombinant eotaxin. Purified eotaxin β (upper panel) and recombinant eotaxin (lower panel) were digested with endoproteinase Lys C and peptide fragments were separated by RP-18 HPLC. Peaks were sequenced and peaks showing identical sequences got the same number. Arrows indicate major differences in both patterns.
Complete amino acid sequence analysis of natural eotaxin β. Purified eotaxin β was digested with various endoproteinases (Lys C, L; Asp N, A, Glu C, G) and peptide fragments were sequenced and aligned. Numbers represent the peak numbers of the peptide-mapping experiments. Boxed residues show differences to the eotaxin cDNA. Percentages at residues 1, 3, and 4 indicate N-terminal variants starting with the respective residue. Ten percent at residue 28 means 10% of the residue identified represents serine.
In order to analyze whether glycosylation is the reason for both the higher molecular mass as well as the differences of the peptide-mapping pattern, we determined the presence of sugars by the use of a glycan ELISA in native material as well as in peptide-mapping fractions. As a result we found that natural eotaxin β, in contrast to recombinant material, gave a glycan-ELISA-positive response (data not shown). In addition, fractions of peaks 1 and 2 of natural eotaxin β (Fig. 4⇑) were positive, whereas none of the fractions of recombinant material showed any glycan-ELISA activity (data not shown).
MALDI MS analyses of peaks 1 and 2 revealed molecular masses of 649 and 1068 for both peaks. The calculated molecular mass of the peptide fragment predicted is 529.
Dose dependency and kinetics of IL-4-induced eotaxin mRNA expression
In order to analyze the conditions of eotaxin induction after stimulation with IL-4, we analyzed in semiquantitative reverse transcriptase (RT)-PCR eotaxin mRNA production in cultivated dermal fibroblasts.
Half-maximum eotaxin mRNA expression occurred after stimulation with 0.1 ng/ml IL-4, and maximum mRNA expression was seen at concentrations higher than 1 ng/ml (Fig. 6⇓A). Eotaxin mRNA was up-regulated within 1 h and showed maximum responses after 18 h (Fig. 6⇓B).
IL-4 induces eotaxin mRNA expression in dermal fibroblasts. Eotaxin mRNA expression relative to GAPDH mRNA content was determined by semiquantitative RT-PCR in human dermal fibroblasts treated for 6 h with various concentrations of IL-4 (A) or treated with 5 ng/ml IL-4 for various lengths of time (B). Relative eotaxin mRNA expression after treatment of dermal fibroblasts for 6 h using different conditions (medium alone, 30 ng/ml TNF-α, 0.1 ng/ml IL-4, 0.1 ng/ml IL-4 + 5 ng/ml TNA-α, or 0.1 ng/ml IL-4 + 30 ng/ml TNF-α) are compared with each other in C. GAPDH (359 bp) and eotaxin (207 bp)-specific amplification products are indicated. A 100-bp ladder was used as m.w. size marker (MW).
We further investigated whether conditions do exist that enhance IL-4-dependent eotaxin production. As shown in Figure 6⇑C, a combination of IL-4 and TNF-α seems to represent a very potent stimulus. We therefore analyzed whether release of bioactive eotaxin is also enhanced.
IL-4 in combination with TNF-α strongly increases release of different eotaxin variants
Heparin-bound proteins that were secreted from fibroblasts stimulated with IL-4 (20 ng/ml) together with TNF-α (20 ng/ml) were separated by preparative RP HPLC (Fig. 7⇓A). Two peaks of Eo-activity were identified in HPLC fractions. The later eluting Eo attractant is identical with RANTES (Fig. 7⇓A) as could be revealed by further purification and N-terminal sequencing (data not shown).
A combination of IL-4 and TNF-α strongly stimulates the release of eotaxin by dermal fibroblasts. Heparin-bound proteins secreted by fibroblasts previously stimulated with IL-4 (20 ng/ml) plus TNF-α (20 ng/ml) for 48 h were separated by preparative RP-8 HPLC (A). Note the presence of eosinophil chemotactic (shaded area) eotaxin as a single and early eluting protein peak fraction giving a single band at 13 kDa upon SDS-PAGE (A, insert, lane 2; lane 1 contains m.w. marker). The second peak of eosinophil chemotactic activity (shaded area, retention time near 24 min) could be identified as RANTES. A typical experiment with n > 14 is shown. Micro-Mono S HPLC of the eotaxin-containing fraction resulted in two protein peaks (peak I and peak II) containing eosinophil chemotactic activity (shaded area) (B), which were further separated upon micro-RP-18-HPLC. Peak I resulted in two bioactive protein peaks (eotaxin β and eotaxin γ) (C) and peak II led to one bioactive protein peak (eotaxin α) (D). Note the presence of eotaxin α as a quantitatively dominating variant produced under these conditions.
The early eluting Eo attractant represents a mixture of eotaxin isoforms (eotaxin α, β, and γ) that eluted together as a single and one of the major at 215-nm absorbing protein peaks (Fig. 7⇑A). SDS-PAGE analyses of this eotaxin preparation (Fig. 7⇑A, inset) revealed a single band of 13-kDa size. The amount of eotaxin reproducibly (n > 14) appeared to be 10- to 20-fold higher when compared with IL-4 alone as revealed by comparison of the peak integration units. Upon micro-Mono S-HPLC (Fig. 7⇑B), however, two Eo chemotactic 13-kDa proteins were separated. Whereas two isoforms (eotaxin β and eotaxin γ) (Fig. 7⇑C, left panel) were separated by micro-RP HPLC of the earlier eluting protein peak I (Fig. 7⇑B), the stronger and most abundant cationic protein peak II (termed eotaxin α) appeared to be homogeneous giving a single peak (Fig. 7⇑D). Comparison of the peptide-mapping patterns seen after digestion of eotaxins α, β, and γ, respectively, with Lys C, with the pattern of the eotaxin we isolated from IL-4-stimulated fibroblasts, (Fig. 4⇑) showed identity with eotaxin β.
N-terminal sequences of all eotaxins were found to be identical giving in each case a mixture of three N-terminal variants in a similar ratio as shown before. Furthermore, all variants gave positive results in a glycan assay indicating glycosylation. MALDI MS analyses revealed a size heterogeneity in all eotaxin variants with principal masses at 9320, 9020, and 8475. The intensities of some masses varied; eotaxin β also contains a component having a mass of 7882.
Natural eotaxin variants express different biologic activity
When the different purified variants of natural eotaxin were studied for eosinophil chemotactic activity, we found the major eotaxin α, which represents 60% of total eotaxin, showing similar potency and efficacy (percentage of input migrating cells) in an in vitro Eo chemotaxis assay as seen for recombinant eotaxin (Fig. 8⇓). Both eotaxin β and eotaxin γ were found to be slightly less active regarding potency, but showing identical efficacy as recombinant eotaxin.
Eosinophil chemotactic activity of natural eotaxin variants. Eotaxin variants were purified from supernatants of dermal fibroblasts stimulated with a combination of IL-4 and TNF-α and were tested for induction of chemotactic activity in human eosinophils using a Boyden chamber assay system (eotaxin α, •; eotaxin β, □; eotaxin γ, ▪). For comparison, the dose-response curve of recombinant human eotaxin purification of commercially available material (after micro-RP-HPLC) is included (○). PAF (10−7 M) served as a positive stimulus, whereas medium alone (−) served as a negative control. We estimated an ED50 ± SD of 18 ± 13, 50 ± 11, and 48 ± 42 ng/ml for eotaxin α, β, and γ, respectively. The mean of three experiments, each performed in duplicate, is shown.
Discussion
Immunoinflammatory responses to parasitic nematode infections and allergic diseases have some similarities, the most profound being increase of blood Eos and serum total IgE. Eos infiltrate affected tissue and—in the case of helminth infection—finally will adhere to the parasite possibly to kill and eliminate it (6).
Th2 cells are known to favor activation of mast cells and Eos and stimulation of B cell growth and isotype switching to cells producing IgE and thus are believed to have a protective role in helminth infection (7).
A body of evidence supports a role of Eos as effector cells in killing helminthic parasites, especially in the larval stage (19). Thus it is intriguing to speculate that there is a link between Th2 cells and tissue recruitment of Eos. Initial experiments in our laboratory revealed that blood- and tissue-derived Th2 cell clones produced almost no Eo-attracting RANTES or other Eo attractants (our unpublished results). This unexpected finding was supported by a recent report that Th1 cells, but not Th2 cells, produce RANTES (20). Furthermore, experiments failed to show that T lymphocyte preparations express eotaxin mRNA—even after stimulation (Ref. 12 and our unpublished results); however, eotaxin mRNA expression in Ag-induced pulmonary eosinophilia in rodents is T cell dependent (21, 22), which favored the hypothesis that Th2 cytokines may indirectly induce tissue infiltration of Eos.
In this study we could identify dermal fibroblasts as the first cellular source of secreted and biologically active eotaxin, which is released as a post-translational modified isoform of eotaxin—termed eotaxin β—when fibroblasts were stimulated with IL-4. This finding is of particular importance because it would explain why Eos usually appear in the dermis rather than epidermis in eosinophilic skin diseases (5). Originally RANTES appeared to be of high relevance as attractant involved in Eo tissue accumulation (23). Indeed dermal fibroblasts produce high amounts of biologically active RANTES when stimulated with TNA-α (24). However, in allergic lung inflammation of mice, eosinophilic infiltrates appeared to depend upon the presence of T lymphocytes but not RANTES (25). In this study, we did not find any RANTES as well as MCP-3 bioactive and immunoreactive protein after stimulation of dermal fibroblasts with IL-4 when solid phase ELISAs with specific mAbs were used (our unpublished observations). Furthermore, the discovery of only minute amounts of RANTES in lesional scales of atopic patients (26) and lack of RANTES mRNA correlation with Eo tissue infiltration in allergen-induced cutaneous late phase responses (27) point toward a comparatively minor role of RANTES as well as possibly MCP-3 in Th2 cytokine-mediated skin inflammation.
Although there are reports showing that eotaxin mRNA is constitutively expressed in the gut and skin or rodents (28) as well as the gut of humans (12, 13), our knowledge about regulation of human eotaxin gene expression and especially protein release is scarce. As yet there is no information whether bioactive eotaxin is released by cultured normal human cells although a recent report indicates the release of minute amounts of immunoreactive eotaxin in a lung epithelial cell line (29). Umbilical cord endothelial cells have been reported to express eotaxin mRNA, when stimulated with TNF-α, but not with IL-4 (13). In our experiments we were unable to detect any Eo chemotactic cytokine in culture supernatants of TNF-α- or IL-4-stimulated umbilical cord endothelial cells (our unpublished observations).
Furthermore, in our hands, cultured keratinocytes also failed to express eotaxin mRNA and to release Eo attractants (our unpublished observations). In skin, the property to respond toward IL-4 stimulation with a strong release of eotaxin appears to be restricted to dermal fibroblasts. Dermal fibroblasts express IL-4R (30) and produce collagen I upon stimulation with IL-4 (31), thus representing important cells mediating Th2-dependent effector mechanisms in skin.
An association between tissue eosinophilia and IL-4 production has been presumed because transgenic mice expressing IL-4 in the lung showed eosinophilic inflammation (32). Furthermore, in IL-4-deficient mice, tissue eosinophilia is strongly inhibited upon allergen-induced airway inflammation (33) and upon Onchocerca volvulus-mediated corneal inflammation (34), but not blood eosinophilia, which depends upon IL-5. The failure to completely abolish tissue eosinophilia by the use of anti-IL-4 points toward a role of other factors. A likely candidate would be IL-13, which binds to the α-chain of the IL-4R (35). This receptor is known to be expressed in fibroblasts (30) and thus IL-13 could act similarly as IL-4.
A rather indirect role of IL-4 together with IL-5 in mediating effector mechanisms is likely from experiments showing that CD4+ T lymphocytes are involved in Eo accumulation in airways of Ag-challenged allergic mice: depletion of T cells resulted in a decrease of lung eosinophilia together with a decrease of IL-4 and IL-5 mRNA expression in the lung (36).
Apart from T lymphocytes (mainly CD4+NK1.1+ T cells) (37) other cells such as mast cells (38) and basophils (39) are known to be major cellular sources of stored IL-4, which is released after stimulation.
It is well documented that mast cell and basophil activation in vivo is linked with tissue eosinophilia (40). As mechanism for this association the release of Eo attractants by mast cells and/or basophils has been discussed (41). Our studies would add another likely mechanism whereby activated mast cells and/or basophils release IL-4, which would activate in vivo fibroblasts for eotaxin production.
Our finding that a combination of IL-4 and TNF-α induces the release of large amounts of a mixture of biologically active eotaxin isoforms (termed eotaxins α, β, and γ) is of importance because mast cells and basophils contain both preformed IL-4 and TNF-α (42), which are released when stimulated through the IgE receptor. Thus it is conceivable that eosinophil tissue infiltration seen after mast cell activation is the result of strong eotaxin release in neighboring fibroblasts, which needs to be proved in in vivo experiments.
Eotaxin is the only Eo attractant that is secreted by IL-4-stimulated dermal fibroblasts. Thus other chemokines including MCP-4 (43, 44) apparently are not released as bioactive molecules or in sufficient amounts and thus appear to be of less importance for recruitment of Eos into human skin in IL-4-mediated reactions. It is interesting to note that MCP-3, which also expresses Eo chemotactic activities (45, 46), is not detected in supernatants of cultured dermal fibroblasts, despite strong mRNA expression (our unpublished observations). This may point toward a differential post-translational regulation of some chemokines and a careful interpretation of gene expression vs protein release.
The natural eotaxin isoform we have isolated from IL-4 stimulated fibroblasts (eotaxin β) in principle shows the predicted backbone protein sequence deduced from the eotaxin cDNA (12, 14, 18). There are, however, marked differences that include mobility upon SDS-PAGE analysis and peptide-mapping profiles. With respect to the previous identification of different eotaxin cDNAs (18), it was interesting to clarify whether the eotaxin variants are really released. In the case of the chemokine gro, different forms (gro α, gro β, and gro γ) are encoded by different genes (47). MIP-1α on the other hand shows a gene polymorphism with various, slightly different forms (48). To identify differences in the primary structure, we therefore performed peptide mapping and sequencing experiments with natural eotaxin purified from IL-4-stimulated fibroblasts (eotaxin β) as well as recombinant eotaxin.
The data from these experiments indicate that the eotaxin variant encoded by clone 53 (Lys58 is substituted by Arg) (18) represent 10% of the natural eotaxin (α, β, and γ) preparations. Furthermore, residue 71, predicted to be a Thr, is glycosylated as shown by the shift to early retention time upon RP HPLC of the peptide-mapping fragment when compared with recombinant material as well as positive glycan ELISA results. The polar shift, however, and duplication of the HPLC peak representing the most prominent UV-absorbing peak in the peptide-mapping experiment of recombinant eotaxin using the endoproteinase Lys C (see Fig. 4⇑) indicate derivatization in the area between residues 58 and 65 (see Fig. 5⇑). The lack of carbohydrate in these modified fragments (data not shown) and the clear evidence for a Ser as residue 63 upon Edman degradation indicate derivatization other than glycosylation, which needs further investigation.
There is clear evidence that residue 71, which is predicted to be a threonine (12, 13, 14) and which could not be determined by Edman degradation, is O-glycosylated. Instead, HPLC fraction variants containing C-terminal fragments of this natural eotaxin contained carbohydrates as revealed by a glycan-ELISA.
Different eotaxins have also been demonstrated in guinea pig bronchoalveolar lavage fluid. These differences were ascribed to differences in O-glycosylation and accompanied by an unidentified residue (49) also predicted to be a threonine (50). Additional biochemical variants of natural human eotaxin (termed eotaxin α and eotaxin γ) could also be purified from fibroblast supernatants previously stimulated with IL-4 plus TNF-α. Peptide-mapping patterns of eotaxin α and γ were different from that of eotaxin β (data not shown) and again these eotaxin forms seemed to be O-glycosylated at residue 71.
Glycosylation can affect biologic activity or cytokines (51). Therefore, it was interesting to investigate specific Eo chemoattractant activities of the three biochemically distinct forms of natural eotaxin. Indeed, eotaxin β, which is released upon stimulation of fibroblasts with IL-4, shows a slightly decreased potency in vitro, whereas the eotaxin α, which is released as major variant upon fibroblast stimulation with a mixture of IL-4 and TNF-α, reveals identical activity as seen with recombinant material.
Thus, eotaxin bioactivity, unlike that of other chemokines known to be N- or O-glycosylated (52), can be affected by glycosylation. Because both eotaxin variants are O-glycosylated at residue 71 (threonine), derivatization, possibly at residue 63 (serine), might be responsible for the decreased potency seen for eotaxin β.
In summary, we have shown that dermal fibroblasts represent a major cellular source of bioactive natural O-glycosylated eotaxin when stimulated with the Th2 cytokine IL-4 or—more efficiently—in combination with TNF-α. Our findings would, at least in part, explain selective recruitment of Eos into affected skin seen in Th2-mediated skin inflammation.
Acknowledgments
We greatly appreciate the expert technical assistance of Jutta Quitzau, Marlies Brandt, and Christine Gerbrecht-Gliessmann and would like to acknowledge Gabriele Tams for editorial help.
Footnotes
-
↵1 This work was supported by Deutsche Forschungsgemeinschaft, Grant Ch-38/7-2.
-
↵2 Both authors contributed equally to this study.
-
↵3 Address correspondence and reprint requests to Dr. J.-M. Schröder, Dept. of Dermatology, University of Kiel, D-24105 Kiel, Germany. E-mail address: jschroeder{at}dermatology.uni-kiel.de
-
↵4 Abbreviations used in this paper: Eos, eosinophils; MALDI, matrix-associated laser desorption ionization; MS, mass spectrometry; PAF, platelet-activating factor; RP, reversed phase; TFA, trifluoroacetic acid; GAPDH, glyceraldehyde phosphate dehydrogenase; ESI, electrospray ionization.
- Received June 2, 1997.
- Accepted September 16, 1997.
- Copyright © 1998 by The American Association of Immunologists
References
In this issue
