Activation of Eotaxin Gene Transcription by NF-κB and STAT6 in Human Airway Epithelial Cells¹

Allergic diseases, such as asthma, are characterized by the infiltration of eosinophils and lymphocytes into the sites of inflammation. Many recent studies have implicated C-C chemokines in this process. Eotaxin is a C-C chemokine originally discovered as the predominant eosinophil chemoattractant in the airways of Ag-challenged guinea pigs (1). It mediates eosinophil migration via the CCR3 receptor and also has chemotactic activity for basophils and T cells, especially Th2 cells (2, 3). An important role for eotaxin in allergic disease has been suggested based on studies in animal models (1, 4, 5) and humans (6, 7) in vivo. Eotaxin is overexpressed in the airway epithelium of asthmatics. As has previously been found for RANTES (8, 9), histological studies have shown that a primary cell type expressing eotaxin is the airway epithelial cell (6, 7). Investigators have reported up-regulation of the expression of the eotaxin gene by proinflammatory cytokines, such as TNF-α and IL-1 (10). Antigen-induced up-regulation has been attributed to T cells, especially Th2 cells (11, 12). Cultured airway epithelial cells have been reported to express eotaxin (10, 13). Based upon the preceding, an analysis of the regulation of eotaxin expression in airway epithelial cells is likely to yield information relevant to the pathogenesis of allergic inflammation.

Mochizuki et al. (14) reported that TNF-α and IL-4 synergistically stimulated eotaxin expression in fibroblasts. We have found a similar synergistic effect of TNF-α and IL-4 in the induction of eotaxin mRNA and protein expression in airway epithelial cells (15). Up-regulation of eotaxin expression by TNF-α and IL-4 might be expected to occur in asthma, because TNF-α and IL-4 are both implicated in the pathogenesis of asthma. TNF-α is known to be produced by basophils, mast cells, monocytes, and macrophages (16, 17). IL-4 is a pivotal cytokine associated with allergic disease that supports Th2 development and is produced by Th2 cells, basophils, and mast cells (18, 19, 20).

The mechanisms regulating eotaxin expression in airway epithelial cells are presently unknown. Many putative binding sites for transcription factors are identified in the eotaxin promoter. Putative binding sites for NF-κB and STAT6 coexist within the proximal region of the eotaxin promoter. NF-κB is a key regulator of immune and inflammatory genes (21). NF-κB is comprised of a homodimer or heterodimer of Rel family proteins, including NF-κB1(p50/p105), NF-κB2(p52/p100), RelA(p65), RelB, and c-Rel. It is sequestered in an inactive form in the cytoplasm by the inhibitory molecules known as IκB. TNF-α activates NF-κB by inducing the phosphorylation and degradation of IκB and leads to the translocation of NF-κB to the nucleus where it can bind to specific promoter binding sites (21). Many cell types are reported to express NF-κB, including airway epithelial cells, and NF-κB has been implicated in epithelial cell responses (22, 23). IL-4 has been shown to trigger tyrosine phosphorylation of a latent cytoplasmic transcription factor termed STAT6 via receptor-associated Janus kinase-1 (JAK1)³ and JAK3. Phosphorylated STAT6 assembles in a dimeric form, translocates into the nucleus, and trans-activates genes containing specific regulatory elements (24, 25). Several cell types have been shown to express STAT6; however, minimal data are available concerning airway epithelial cells (26).

In the present study we investigated the possible roles of NF-κB and STAT6 in the transcriptional regulation of the eotaxin gene in airway epithelial cells. We show that overlapping elements for both factors within the proximal eotaxin promoter mediate the transcriptional induction of reporter gene constructs by TNF-α and IL-4. We also demonstrate that TNF-α activated NF-κB and IL-4 activated STAT6 in airway epithelial cells. These findings provide insight into the mechanisms of epithelial eotaxin expression in allergic diseases.

BEAS-2B is a human airway epithelial cell line transformed with adenovirus 12-SV40 hybrid virus (a gift from Dr. Curtis Harris) (27). BEAS-2B cells were cultured in Hanks’ F12/DMEM (Biofluids, Rockville, MD) with 10% heat-inactivated FBS, 100 U/ml penicillin, and 100 ng/ml streptomycin (Life Technologies-BRL, Gaithersburg, MD) at 37°C with 5% CO₂ in humidified air.

The promoter region of eotaxin was amplified by the method of Siebert et al. using the Genome Walker kit (Clontech, Palo Alto, CA) (28). The first PCR product was amplified from adaptor-ligated genomic DNA fragments using 5′ adaptor primer and a primer derived from the eotaxin exon I sequence (5′-TAGCAGCTGCCTTCAGCCCCCAGGGG-3) (29). The primary PCR reaction mixture was diluted and used as a template for a nested PCR reaction using the nested adaptor primer and the nested primer derived from the eotaxin 5′-upstream sequence (5-ACTTCTGTGGCTGCTGCTCATAG-3). The derived PCR product was cloned using the TA cloning kit (Invitrogen, San Diego, CA), and the construct is referred to as TA-Eotax. Recombinant clones were selected based on PCR screens of derived colonies. Plasmids were isolated using the QIAGEN plasmid purification kit (Qiagen, Santa Clarita, CA). Cloned DNA sequences were determined by the dideoxy method at Johns Hopkins DNA Analysis Facility.

A 1363-bp fragment of the promoter region of the eotaxin gene (site −1363 to −1) was amplified by PCR using a sense primer containing a restriction site for MluI and an antisense primer containing a restriction site for BglII. The amplified PCR product was cloned into a TA cloning vector and digested with MluI and BglII (New England Biolabs, Beverly, MA). The resulting fragment was cloned into the MluI and BglII sites of a promoterless luciferase reporter pGL3-Basic vector (Promega, Madison, WI), and the construct is referred to as pEotx.1363. The pEotx.478 plasmid was constructed by digesting TA-Eotx.1363 with SacI and BglII and cloning the resulting 478-bp eotaxin promoter fragment into the SacI and BglII sites of the pGL3-Basic vector. The pEotax.300 plasmid was constructed by 5′ end digestion of the pEotax.1363 plasmid using exonuclease III and S1 nuclease (Promega). Digested linear vector was circularized by DNA ligase after Klenow DNA polymerase treatment (Promega). The sequence of each plasmid insert DNA was determined by the dideoxy method using the fmol DNA Sequencing System (Promega).

Putative transcription factor binding sites in pEotx.1360 were mutated using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA), using pEotx.1363 as a template. Constructs pEotx.M-1, -2, and -3 were synthesized by temperature cycling using site-mutated primers and PfuTurbo DNA polymerase (Stratagene). Nonmutated template plasmid was digested with DpnI endonuclease (Stratagene). The mutated plasmid was transformed into XL1-Blue cells (Stratagene) and purified using the Qiagen plasmid purification kit. Mutations were selected based on the results of EMSA with mutant probes for each transcription factor (Table I). The sequence of each plasmid was determined by the method mentioned above.

BEAS-2B cells were seeded into six-well plates and allowed to grow to 50–70% confluence. Cells were transfected with 3 μl of Fugene 6 transfection reagent (Roche Molecular Biochemicals, Indianapolis, IN) and 1 μg of the plasmids indicated in the text and incubated in 2 ml of medium. Cytokine was added 24 h after transfection. Six hours after cytokine treatment, cells were washed twice with Ca²⁺- and Mg²⁺-free HBSS (Life Technologies) and solubilized by incubation in 300 μl of reporter lysis buffer for 15 min (Promega). Cells were transferred to microtubes and centrifuged to pellet cellular debris, and the supernatant was stored at −70°C. Luciferase activity was measured using the Luciferase Assay System (Promega) and a luminometer (Analytic Luminescence Laboratories, Sparks, MD). The protein concentration of the samples was measured using the Bradford protein dye reagent (Bio-Rad, Hercules, CA), and the relative luciferase activity was normalized to the protein concentration.

Nuclear extracts were prepared by the modified method of Schreiber et al. (30). The BEAS-2B cells were stimulated as indicated and then harvested with a scraper after washing twice with HBSS. The cells were washed with buffer A (10 mM HEPES, 15 mM KCl, 2 mM MgCl₂, 0.1 mM EDTA, 1 mM DTT, 0.5 mM PMSF, and 10 μg/ml leupeptin; Life Technologies). The cell pellet was resuspended in buffer B (buffer A containing 0.2% Nonidet P-40) and incubated for 2 min. Nuclei were pelleted by centrifugation and resuspended in buffer C (buffer A containing 0.25 M sucrose). Nuclei were again pelleted, then resuspended in buffer D (50 mM HEPES, 400 mM KCl, 10% glycerol, 0.1 mM EDTA, 1 mM DTT, 0.5 mM PMSF, and 10 μg/ml leupeptin) and incubated with shaking for 30 min. All procedures were performed on ice. The mixture was centrifuged, and the supernatant was stored at −70°C.

Whole cell extracts were prepared by modification of a method previously reported (26, 31). Cells were harvested as described above. The cells were resuspended in lysis buffer (20 mM Tris-HCl, 1 mM EDTA, 1 mM DTT, 1% Nonidet P-40, 10% glycerol, 1 mM PMSF, 0.2 mg/ml leupeptin, 1 mM Na₃VO₄, and 50 mM NaF). Cells were sonicated for 10 s with an ultrasonic cell disrupter (Heat System Ultrasonics, Farmingdale, NY) and then incubated for 5 min. All procedures were performed on ice. Lysates were centrifuged, and the supernatant was collected and stored at −70°C.

DNA-protein binding assays were based on the modified methods previously reported (31, 32). Aliquots of 5 μg of nuclear extracts or whole cell extracts were incubated in 10 μl of total reaction volume containing 10 mM Tris-HCl, 1 mM DTT, 1 mM EDTA, 10% glycerol, 0.1% Triton X-100, 50 μg/ml poly(dI-dC), 0.1 mg/ml BSA, and 50 mM KCl (Life Technologies) for 5 min with or without unlabeled oligonucleotide probe. The ³²P-labeled oligonucleotide probe was added to the reaction mixture and then incubated for 20 min. In some experiments, Abs against p50(NF-κB1), p65(RelA), RelB, c-Rel, p52(NF-κB2), STAT1, STAT6 (Santa Cruz Biotechnology, Santa Cruz, CA), or isotype-matched antisera were incubated with the mixture for 30 min after incubation with labeled probe. The reaction products were analyzed by electrophoresis in a 5% polyacrylamide gel with 0.5 × TBE buffer. The gels were dried and analyzed by autoradiography. The sequences of the oligonucleotide probes used are shown in Table I.

Western blot of nuclear extracts was performed to assess translocation of STAT6 into nuclei as reported previously (33). Fifty micrograms of nuclear extracts or 2 μg of recombinant STAT6 protein (a gift from Dr. Ulrike Schindler) were incubated with 1 μg of anti-STAT6 Ab prebound to protein G-Sepharose beads (Pharmacia Biotech, Piscataway, NJ) for 2 h at 4°C with gentle rotation. After washing three times with the lysis buffer, the immunoprecipitated proteins were boiled for 5 min in the sample buffer containing 0.25 M Tris-HCl, 10% glycerol, 5% SDS, and 0.025% 2-ME (Sigma, St. Louis, MO). The samples were then subjected to 10% Tris-glycine gradient gel electrophoresis (NOVEX, San Diego, CA) and transferred to a nitrocellulose membrane (Amersham, Arlington Heights, IL). The membrane was blocked with 5% nonfat milk powder in TBST (50 mM Tris, 0.15 M NaCl, and 0.05% Tween-20), incubated with 1 μg/ml rabbit anti-STAT6 Ab (Santa Cruz Biotechnology) in TBST for 3 h, washed with TBST, and then incubated with anti-rabbit Ig Ab (Amersham) for 1 h. After extensive washing with TBST, chemiluminescent substrate was added (ECL Western blot detection system, Amersham), and the membrane was subjected to autoradiography.

Analyses of data from transfection/luciferase assays were performed using StatView II (Abacus Concepts, Berkeley, CA). Data are expressed as the mean ± SE. Statistical differences were determined by ANOVA with Fisher’s protected least significant difference test.

To study the transcriptional regulation of the eotaxin gene, we cloned a region of the gene extending to bp −1363 upstream of the transcription start site using a genomic library. The sequence of this region is shown in Fig. 1. It contains a TATA box (at bp −43 to −39) and consensus binding sites for NF-κB (at bp −68 to −59), STAT6 (at bp −74 to −65), AP-1 (at bp −1082 to −1076), and C/EBP (at bp −1278 to −1270). The overall sequence is similar, but not identical, to those published by Garcia-Zepeda et al. and Hein et al. (34, 35). This fragment was cloned into a luciferase reporter plasmid for further functional analysis.

TNF-α (0.1, 1, 10, and 100 ng/ml) dose dependently increased luciferase activity in BEAS-2B cells transfected with the pEotx.1363 plasmid (Fig. 2,A). Maximum induction was observed using 100 ng/ml of TNF-α (2.5 ± 0.1-fold; p < 0.05 compared with control). IL-4 (0.5, 5, and 50 ng/ml) also increased luciferase activity dose dependently, although modestly compared with TNF-α (Fig. 2,B). Maximum induction was observed using 50 ng/ml of IL-4 (1.5 ± 0.04-fold; p < 0.05 compared with control). The combination of 100 ng/ml TNF-α and 50 ng/ml IL-4 additively activated the eotaxin promoter (3.6 ± 0.2-fold; p < 0.05 compared with control; Fig. 3,A). Similar results were obtained using cells transfected with reporter plasmids containing either a 478-bp fragment or a 300-bp fragment of the eotaxin promoter (Fig. 3, B and C).

Many effects of TNF-α are known to be mediated via activation of NF-κB, while IL-4 is known to activate STAT6 via JAK1 and JAK3. We therefore used EMSA to determine whether stimulation with these cytokines results in formation of active NF-κB or STAT6. EMSA with nuclear extracts from BEAS-2B cells stimulated with 10 ng/ml TNF-α resulted in the formation of a strong binding complex (complex 1) using a DNA probe spanning the putative binding sites for STAT6 and NF-κB in the eotaxin promoter (Fig. 4,A). This binding complex was blocked or supershifted by Abs to p50 and p65 NF-κB Rel family members. Abs against other Rel family members (RelB, c-Rel, and p52) and STAT6 did not affect formation of the binding complex. Preimmune rabbit and goat IgG did not influence complex formation (data not shown). The binding specificity of the NF-κB complex was further assessed by competition with unlabeled oligonucleotides. Unlabeled wild-type oligonucleotide and an oligonucleotide only mutated within the putative STAT6 binding site (referred to as m1, Table I) competed for DNA-protein complex formation (Fig. 4,B), while two oligonucleotides carrying a mutation within the putative NF-κB binding site (referred to as m2 and m3) were ineffective. Taken together, these results suggest that TNF-α induces the formation of a sequence-specific NF-κB complex composed of p65/p50 heterodimers. IL-4 did not induce NF-κB and the combination of IL-4 and TNF-α did not alter the binding complex compared with TNF-α alone (data not shown). An unknown complex (complex 2) was formed equally in unstimulated cells or cells stimulated with TNF-α (Fig. 4, A and B). Although this constitutive complex was not inhibited by any of the Ab mentioned above (Fig. 4,A), the complex formation was inhibited by excess of wild-type oligonucleotide and m1 (Fig. 4 B).

The STAT6-DNA binding complex was barely detectable using nuclear extract (data not shown), presumably due to the lower efficiency of EMSA from nuclei compared with NF-κB. Therefore, we confirmed the ability of DNA-STAT6 binding by following experiments based on the method described by Guo et al. (26), because activated STAT6 in cytoplasm can be bound to DNA probe. EMSA with whole cell extracts from BEAS-2B cells stimulated with 10 ng/ml IL-4 were performed. A binding complex (complex 1) was induced by IL-4 and detectable in whole cell extracts using a wild-type DNA probe (Fig. 5). This binding complex was confirmed to contain STAT6 by inhibition using anti-STAT6 Ab. Abs against Rel family members and STAT1 did not influence formation of the binding complex (Fig. 5), nor did rabbit IgG (data not shown). Unlabeled wild-type oligonucleotide and the oligonucleotide mutated at only the NF-κB binding site (m2) inhibited DNA-protein binding, while oligonucleotides mutated at the STAT6 binding site (m1 and m3) were inactive. TNF-α did not induce STAT6, and the combination of IL-4 and TNF-α did not alter the STAT6-DNA binding complex compared with IL-4 alone (data not shown). Unknown complexes 2, 3, and 4 seem to be comprised of constitutively expressed proteins that bound to the eotaxin promoter. Taken together, these results indicate that IL-4 induces activation of functional STAT6 in airway epithelial cells.

Western blot of nuclear extracts, immunoprecipitated with anti-STAT6 Ab, was used to assess the translocation of STAT6 into the nucleus following stimulation of 10 ng/ml IL-4. Recombinant STAT6 protein was used as positive control. A minimal level of STAT6 was detected from the nuclei of unstimulated BEAS-2B cells (Fig. 6). The amount of STAT6 in the nucleus increased after stimulation with IL-4.

To test the roles of NF-κB and STAT6 binding sites in transcription of the eotaxin gene, luciferase reporter plasmids with mutated proximal promoters were used. The sequences of sites in the mutant plasmids are shown in Table I. A reporter using the eotaxin promoter mutated at the binding site for STAT6 (referred to as pEotx.M1) showed total loss of IL-4 stimulation (Fig. 7,B) and an additive effect with TNF-α, but still responded to stimulation with TNF-α. A reporter plasmid, pEotx.M2, which was mutated at the binding site of NF-κB, showed total loss of TNF-α stimulation and additive activation with IL-4 (Fig. 7,C), but still responded to stimulation with IL-4. Neither TNF-α nor IL-4 activated the reporter, pEotx.M3, which was mutated at the overlapping binding sites for STAT6 and NF-κB (Fig. 7 D).

Several recent studies indicate that airway epithelial cells are a rich source of eotaxin in humans in individuals with allergic airways disease and/or after experimental allergen challenge in animals (1, 4, 5, 6, 7). To study the regulation of expression of the eotaxin gene in epithelial cells, we isolated 1363 bp of the proximal promoter. The sequence of the derived eotaxin 5′ upstream region is compatible to the sequences that have recently been reported by two independent groups of investigators (34, 35). Luciferase reporter constructs containing this promoter region were activated by both TNF-α and IL-4. Combining these two cytokines caused an additive effect. Alone, TNF-α was a stronger stimulus than IL-4. Because a strong synergy between these two stimuli is observed at the protein level, and an intermediate synergy is observed in Northern blot studies, we suspect that post-transcriptional and/or translational mechanisms may be involved (15). Based upon the presence of a single site containing overlapping consensus response elements for both NF-κB and STAT6, we used EMSA to probe for activation of these two transcription factors. NF-κB has been known to regulate expression of several inflammatory proteins, including RANTES, IL-8, monocyte chemoattractant protein-1, inducible nitric oxide synthase, VCAM-1, and ICAM-1 (23, 36, 37, 38, 39, 40). STAT6 has been shown to be essential for most responses to IL-4. It activates transcription of several IL-4-induced genes, such as CD23 (the low affinity IgE receptor), MHC class II, and IgE germline and mediates Th2 development (24, 41, 42, 43).

TNF-α induced the activation of NF-κB, which bound to a DNA probe derived from the eotaxin proximal promoter in airway epithelial cells. The sequence of this site is slightly different from a consensus sequence of the NF-κB binding site (the consensus sequence is GGGRNNYYCC; the eotaxin promoter sequence is GGAATCTCCC; positions bp −68 to −59; N indicates any of nucleotides, R indicates A or G, and Y indicates C or T; see Table I) (21, 44). The proteins bound to the DNA probe were confirmed to be p50 and p65 of the NF-κB family, however, using specific Abs.

Activation of STAT6 by IL-4 was detected by EMSA using a DNA probe derived from the eotaxin promoter that contains a putative STAT6 binding site (the consensus sequence is TTCN3/4 GAA; the eotaxin sequence is TTCCCTGGAA; positions bp −74 to −65; see Table I) (45) and also overlaps with the 5′-end of the putative NF-κB binding site mentioned above. The specificity of TNF-α-induced NF-κB and IL-4-induced STAT6 was further established by the ability of unlabeled consensus probes to inhibit respective complex formation as well as by the use of unlabeled eotaxin probe mutated at either site. Thus, we have identified a unique site in the eotaxin promoter that binds to both NF-κB and STAT6. Other investigators have reported that TNF-α can induce or enhance STAT6 activation, and IL-4 can induce activation of NF-κB (46, 47, 48); however, our results suggest that TNF-α activated NF-κB and not STAT6, while IL-4 activated STAT6 and not NF-κB. Both these transcription factors independently bound to the eotaxin promoter element in our studies.

Having established that TNF-α and IL-4 can activate eotaxin promoter reporter plasmids, and that TNF-α activated NF-κB and IL-4 activated STAT6, we next tested whether NF-κB and STAT6 were responsible for activating the reporter genes. Promoter assays using mutant plasmids showed that the putative NF-κB site at −68 to −59 is necessary for TNF-α activation of the eotaxin promoter. Likewise, results with the pEotx.M1 mutant indicate that the putative STAT6 binding site was necessary for IL-4 activation of the promoter. In both cases, the additive effects were lost when plasmids mutated at either site were used. These findings add support to the data from EMSA, suggesting that TNF-α and IL-4 activate the eotaxin promoter via NF-κB and STAT6, respectively.

TNF-α has been shown to activate NF-IL-6 (C/EBP family) and AP-1 (49, 50). Our eotaxin reporter plasmid contains consensus binding sites for both these factors as has been indicated by other investigators (34, 35). We did not detect any clear complexes from EMSA using DNA probes derived from these sequences after TNF-α treatment, however (data not shown). Although IL-4 has also been reported to activate the C/EBP family (51), no clear DNA-protein binding was observed in EMSA with a probe derived from the C/EBP consensus element contained in the eotaxin promoter (data not shown). Reporter gene assays using further truncated forms of the promoter (pEotx.478 and pEotx.300) recapitulated the findings with the 1363-bp promoter fragment. Because activation by TNF-α and IL-4 was preserved in the transfections with these plasmids that do not contain the putative binding sites for C/EBP or AP-1, these transcription factors are not likely to be essential for activation of the eotaxin promoter fragments used by TNF-α and IL-4.

The results of this study suggest that NF-κB and STAT6 act positively at a site on the eotaxin promoter containing overlapping binding sites. This contrasts with the results of Bennet et al., who reported suppression of TNF-stimulated E-selectin expression by STAT6 antagonism of NF-κB (52). They suggested that overlapping binding sites of NF-κB and STAT6 in the E-selectin promoter resulted in competition between STAT6 and NF-κB for binding. On the E-selectin promoter, there are two adjacent, but nonoverlapping, binding sites for NF-κB that are straddled by a STAT6 binding site that shares 5 bp with each NF-κB site. In the eotaxin promoter, STAT6 and NF-κB share only 4 bp for both factors; this structure may make it possible for factors to bind simultaneously to the overlapping site on the eotaxin promoter. Shen et al. and Iciek et al. have reported that NF-κB and STAT6 synergistically activate the mouse germline Cε promoter (48, 53). A similar mechanism has been reported in the human IgE germline promoter (54). In these cases, a STAT6 binding site was adjacent to a NF-κB binding site, although these sites were separated by several nucleotides. There are other reports that support an cooperation between NF-κB and STAT6 on overlapping binding sites. Isalan et al. demonstrated that zinc finger-containing proteins can generally recognize overlapping DNA binding domains based upon analysis of the crystal structure of these binding proteins (55). Although the three-dimensional crystal structures of STATs and NF-κB have been analyzed (44, 56), there are few data concerning such interactions of these two transcription factor families. Wotton et al. reported cooperative binding of Ets-1 and core binding factor to the individual sites, which are directly adjacent on the TCR β gene enhancer region of the promoter (57). Our mutagenesis experiments clearly establish independent roles for NF-κB and STAT6 in regulating the eotaxin promoter despite the intimate association of binding sites for these two factors. These reports and our data suggest that the differences in DNA motif and structure of promoter could result in different patterns of protein binding to DNA and promoter activity. Further study will be necessary to better understand the mechanism of interactions of NF-κB and STAT6 on the eotaxin promoter, especially in mediation of the synergy between TNF-α and IL-4.

Both NF-κB and STAT6 have been implicated in allergic inflammation. Yang et al. have shown that mice deficient in the p50 subunit of NF-κB protein do not mount eosinophilic lung inflammation and that eotaxin expression was inhibited compared with that in wild-type mice (58). Their data are in agreement with our findings indicating an important role for NF-κB in the regulation of eotaxin in airway epithelium. Airway eosinophilia and hypersensitivity were severely reduced in STAT6-deficient mice (59, 60). This could result from impaired Th2 development or IgE synthesis, which are mainly regulated by IL-4/STAT6. Our studies suggest that impaired eotaxin synthesis in STAT6-deficient mice may be expected to contribute to reduced eosinophil infiltration.

We thank Dr. Ulrike Schindler, Bruce S. Bochner, Lisa A. Beck, David Proud, Jean Kim, Mio Kawaguchi, and Shau-Ku Huang for excellent assistance and helpful discussions, and John E. Cumberland, John W. Schmidt, Eva Ehrlich-Kautzky, Eri Matsukura, and Bonnie Hebden for skillful assistance.