IL-5 is a major therapeutic target to reduce eosinophilia. However, all of the eosinophil-activating cytokines, such as IL-5, IL-3, and GM-CSF, are typically present in atopic diseases, including allergic asthma. As a result of the functional redundancy of these three cytokines on eosinophils and the loss of IL-5R on airway eosinophils, it is important to take IL-3 and GM-CSF into account to efficiently reduce tissue eosinophil functions. Moreover, these three cytokines signal through a common β-chain receptor but yet differentially affect protein production in eosinophils. Notably, the increased ability of IL-3 to induce the production of proteins, such as semaphorin-7A, without affecting mRNA levels suggests a unique influence of IL-3 on translation. The purpose of this study was to identify the mechanisms by which IL-3 distinctively affects eosinophil function compared with IL-5 and GM-CSF, with a focus on protein translation. Peripheral blood eosinophils were used to study intracellular signaling and protein translation in cells activated with IL-3, GM-CSF, or IL-5. We establish that, unlike GM-CSF or IL-5, IL-3 triggers prolonged signaling through activation of ribosomal protein S6 (RPS6) and the upstream kinase 90-kDa ribosomal S6 kinase (p90S6K). Blockade of p90S6K activation inhibited phosphorylation of RPS6 and IL-3–enhanced semaphorin-7A translation. Furthermore, in an allergen-challenged environment, in vivo phosphorylation of RPS6 and p90S6K was enhanced in human airway compared with circulating eosinophils. Our findings provide new insights into the mechanisms underlying differential activation of eosinophils by IL-3, GM-CSF, and IL-5. These observations identify IL-3 and its downstream intracellular signals as novel targets that should be considered to modulate eosinophil functions.
Mature eosinophils are nondividing, innate immune cells with a limited (∼3 d) circulating life-span. Eosinophilia is found in a variety of diseases, including the hypereosinophilic syndromes, eosinophilic esophagitis, atopic dermatitis, and eosinophilic asthma. Numerous in vivo and in vitro studies suggested that eosinophils trigger tissue damage, influence the immune response (1), and drive tissue fibrosis by release of toxic granule proteins, leukotrienes, cytokines, and chemokines (2). In asthma, disease severity, chronicity, and exacerbations are frequently associated with airway eosinophilia (3). Depletion of eosinophils in allergen-challenged animals reduced collagen deposition, mucus production, and airway hyperresponsiveness (4, 5). In aggregate, these studies establish a critical role for eosinophils in asthma pathogenesis.
IL-5 regulates the differentiation, survival, and function of eosinophils. The nearly exclusive presence of IL-5RA on eosinophils made IL-5 an ideal drug target to reduce eosinophilia. Several therapeutic mAbs (anti–IL-5 or anti–IL-5RA) are currently in phase III clinical trials (6). In asthma, these Abs dramatically decreased peripheral blood eosinophils but had much less effect on reducing airway eosinophil numbers (∼50%) (7). Yet, this decrease in tissue eosinophils by anti–IL-5 therapy reduced asthma exacerbations by ∼50% and decreased the use of corticosteroids in severe asthmatic subjects with previously demonstrated persistent airway eosinophilia (8, 9). However, in a more general asthma population, anti–IL-5 failed to improve symptoms and pulmonary functions (10). We showed previously that airway eosinophils from bronchoalveolar lavage (BAL) lose their IL-5RA and do not degranulate in response to IL-5 (11), as well as that the upregulation and activation of β2 integrins on airway eosinophils are not affected by treatment with anti–IL-5 (12). These observations support the idea that other factors, in addition to IL-5, are important for airway eosinophil presence and activation.
IL-5, GM-CSF, and IL-3 initiate signaling via a common β-chain receptor and are known as βcR signaling cytokines. IL-3 and GM-CSF are far more pleiotropic than IL-5, but all three are thought to have largely redundant functions on eosinophils. Mounting evidence suggests that this is an oversimplification. We (13, 14) and other investigators (15) showed that IL-3 alone or associated with TNF-α was more potent than IL-5 or GM-CSF for inducing the production and release of proteins from eosinophils. These studies demonstrate that IL-3, IL-5, and GM-CSF have unappreciated distinct roles in eosinophil biology and, by inference, asthma and allergy.
IL-3 is relevant in asthma, and allergy in general, because it is released by activated Th2 lymphocytes and by mast cells or basophils following IgE cross-linking (16). Serum IL-3 levels are significantly elevated in poorly controlled asthmatics, and plasma levels are elevated in patients with asthma or airway allergy (17–19), supporting a role for IL-3 in asthma. Moreover, IL-3+ cells are more abundant in BAL cells or ex vivo–activated T cells from subjects with asthma compared with control subjects, and their numbers increase with asthma severity (19). Finally, airway allergen challenge in subjects with mild asthma leads to increased levels of IL-3 in BAL fluid (1, 20).
Recently, we (13) reported that human airway eosinophils expressed more of the profibrotic membrane protein semaphorin-7A than did blood eosinophils following an airway allergen challenge. Blood eosinophils treated ex vivo with IL-3, but not GM-CSF or IL-5, showed an important increase in semaphorin-7A protein without changes in its mRNA levels. Consistent with an effect on translation, we found more semaphorin-7A mRNA associated with polyribosomes after IL-3 signaling (13). To test the hypothesis that IL-3 regulates translation, in this study we examined the phosphorylation of ribosomal protein S6 (RPS6). RPS6 is one of the two ribosomal proteins susceptible to phosphorylation following cellular stimulation by cytokines (21, 22). In stromal cells, RPS6 phosphorylation is controlled by the mammalian target of rapamycin (mTOR) and downstream by the kinases p70S6K1 and p70S6K2 (23). In genetically modified RPS6 (knock-in) mice with alanine substitutions at all five phosphorylatable serine residues, global protein synthesis was decreased in liver and mouse embryonic fibroblasts (24). These results are consistent with other studies showing a positive role for RPS6 phosphorylation in the initiation of translation, probably through more efficient 40S ribosomal subunit assembly (25). Phosphorylated RPS6 is located near the mRNA- and tRNA-binding sites at the interface between the small and the large ribosomal subunits (26), and polyribosomes display a higher level of RPS6 phosphorylation than do the ribosomes (21).
The correlation of RPS6 phosphorylation with cell division during mitogenic activation suggests that RPS6 controls mRNA translation in dividing cells (27). Phosphorylation of RPS6 via mTORC1 occurs in tumor cells, and a lack of phosphorylation is associated with reduced development of pancreatic cancer in mice (28). However, the role of phospho-RPS6 in nondividing cells, such as eosinophils, is unknown. The aim of the current study was to analyze the possible role of RPS6 in IL-3–induced semaphorin-7A mRNA translation in eosinophils and to identify intracellular signaling pathways upstream of RPS6 to understand the differential induction of eosinophil semaphorin-7A by IL-3 versus IL-5 and GM-CSF.
Materials and Methods
Subjects, cell preparations, and cultures
The study protocol was approved by the University of Wisconsin-Madison Health Sciences Institutional Review Board. Informed written consent was obtained from subjects prior to participation.
Sixty subjects were used to study peripheral blood eosinophils. Among them, 48 had both mild asthma and allergy, 11 had allergy only, and 1 had asthma only. Four of the 60 subjects had prescriptions for low doses of inhaled corticosteroids, which were not used on the day of the blood draw. Peripheral blood eosinophils were purified by negative selection, as previously described (14). Eosinophil viability was determined using trypan blue exclusion. Eosinophil preparations with purity and viability > 99% were used. Eosinophils (1 × 106/ml) were maintained in complete medium (RPMI 1640 plus 10% FBS) at 37°C, with or without recombinant human (rh)IL-5, rhGM-CSF, or rhIL-3 (2 ng/ml, unless indicated otherwise). rhGM-CSF was purchased from R&D Systems (Minneapolis, MN), and rhIL-3 and rhIL-5 were purchased from BD Biosciences (San Jose, CA). In a previous study, rhGM-CSF, rhIL-3, and rhIL-5 prepared in Escherichia coli and insect cells showed uniformities among vendors with regard to eosinophil activation and protein release (13).
Bronchoscopy and BAL were performed before and 48 h after segmental bronchoprovocation with an allergen (SBP-Ag) in subjects allergic to ragweed, pollen, or cat and with a history of mild asthma (14). Eosinophils were purified from the BAL cell preparation (eosinophils from BAL after an allergen challenge [BAL EOS (AG)]) and from peripheral blood (eosinophils from peripheral blood after an allergen challenge [PB EOS (AG)]) of the same allergen-challenged patient. On the same day, eosinophils were also purified from the peripheral blood of a control unchallenged subject (PB EOS from a control unchallenged subject [PB EOS (CTRL)]). Methods for preparation of airway eosinophils were described previously (14). Immediately after eosinophil preparation, cell pellets were stained for flow cytometry analysis or were snap-frozen and stored at −80°C for further analysis by Western blot.
Cytokine (IL-3, IL-5, GM-CSF) consumption and ELISA
Cytokines (2 ng/ml) were added to peripheral blood eosinophils in RPMI 1640 plus 10% FBS at the beginning of the culture. Cytokines left in culture over time were measured by two-step sandwich ELISAs, as previously described (29) Unlabeled (coating) and biotinylated (detecting) anti–IL-3 and anti–IL-5 Abs and corresponding recombinant protein standards for ELISA were from BD Biosciences. Unlabeled and biotinylated anti–GM-CSF Abs and recombinant protein standard for ELISA were from R&D Systems. The assay sensitivities were <3 pg/ml for GM-CSF and IL-5 and <12 pg/ml for IL-3.
Cells were lysed in RIPA buffer (Cell Signaling, Danvers, MA) plus 0.2% SDS and protease inhibitors or directly in Laemmli buffer (10% SDS) before boiling and loading onto 10–12% SDS-PAGE gels. Immunoblot analysis was performed as previously described (13). Rabbit mAbs anti-RPS6, anti–RPS6-S235/236, anti–Akt-S473, anti–p70S6K-T389, anti–eIF2α-S51, anti–eIF4E-S209, anti–4E-BP1-T37/T46, anti-RSK1/2/3, anti-RSK1 (90-kDa ribosomal S6 kinase [p90S6K1])-380, anti–RSK1-T573, anti–RSK1-T359/S363 and anti–eIF4B-S422 were from Cell Signaling. Secondary HRP-conjugated anti-rabbit IgG Abs were from Cell Signaling or Pierce/Thermo Fisher Scientific (Rockford, IL). Mouse monoclonal anti–β-actin was purchased from Sigma (St. Louis, MO). Goat polyclonal anti–semaphorin-7A Ab was from R&D Systems. Donkey anti-goat Ab was from Santa Cruz Biotechnology. Immunoreactive bands were visualized with Super Signal West Femto chemiluminescent substrate (Pierce/Thermo Fisher Scientific). Bands were quantified using the FluorChem Q Imaging System (Alpha Innotech/ProteinSimple, Santa Clara, CA).
LY294002, rapamycin, calyculin A, and okadaic acid were purchased from Cell Signaling. BI-D1870 and PF-4708671 were purchased from Selleckchem (Houston, TX). U0126, its inactive analog, U0124, and SB203580 were from Calbiochem/EMD Millipore. Cycloheximide was bought from Calbiochem. The neutralizing anti–IL-3 Ab and the control goat IgG were purchased from R&D Systems.
ER-Tracker dyes (Molecular Probes, Invitrogen Detection Technologies, Eugene, OR) were used on living cells during the last 30 min of culture. Cytospins were prepared, and cells were fixed in 4% paraformaldehyde and permeabilized in 0.1% Triton X-100. Following incubation with anti–RPS6-S235/236 (Cell Signaling), cells were treated with 0.2% chromotrope-2R to quench nonspecific autofluorescence and incubated with FITC-conjugated anti-rabbit Ab (A10530; Invitrogen). Images were collected by fluorescence microscopy and analyzed with ImageJ software (http://rsb.info.nih.gov/ij/index.html). Fluorescence of cells was counted in a “blinded” fashion in random visual fields. Pearson correlation coefficients were used to objectively quantify colocalization of phospho-RPS6 (S235/236) to the endoplasmic reticulum (ER) (30). Scatter plots representing pixel intensities were generated, and the Pearson correlation coefficients between red and green pixels were calculated using ImageJ software with a colocalization plugin (Colo2). A coefficient close to 1 indicates good colocalization, whereas a coefficient close to 0 indicates poor colocalization.
Measurement of translation
The determination of translation was performed using Click-iT Labeling Technology (Life Technologies). Fourteen hours after the beginning of activation, eosinophils were centrifuged and recultured in methionine-free RPMI 1640 plus 10% FBS and cytokine (2 ng/ml) for 1 h before adding a nonradioactive amino acid (l-azidohomoalanine, 50 μM) for 4 h. After the amino acid pulse, cells were pelleted, washed twice with PBS, and snap-frozen at −80°C. Cells pellets were lysed in Tris-HCl buffer containing 1% SDS and phosphatase and protease inhibitors. The l-azidohomoalanine present in the cell lysate was chemically ligated (“clicked”) to an alkyne-modified biotin. Proteins were precipitated with methanol/chloroform and suspended in Laemmli buffer for separation in a 10% SDS-PAGE gel. After transfer onto a polyvinylidene difluoride membrane, the newly synthesized proteins (biotin labeled) were visualized using streptavidin-HRP. Signals were quantified, as described above, using Western blot. When indicated, the p90S6K inhibitor BI-D1870 or the p70 ribosomal S6 kinase (p70S6K) inhibitor PF-4708671 was added 15 min before IL-3. For semaphorin-7A immunoprecipitation following the Click-iT labeling procedure, proteins precipitated with methanol/chloroform were suspended in RIPA buffer and immunoprecipitated using 4 μg of a goat anti–semaphorin-7A Ab (R&D Systems) and protein G beads (Millipore). After an overnight incubation at 4°C, beads were washed four times, suspended in the Laemmli loading buffer, and run on a 10% SDS-PAGE gel. Immunoblotting for total semaphorin-7A was performed using a goat anti–semaphorin-7A Ab from R&D Systems and a donkey anti-goat Ab from Santa Cruz Biotechnology. Newly synthetized immunoprecipitated semaphorin-7A protein was visualized using streptavidin-HRP.
Purified blood eosinophils were stained with PE-conjugated anti-CD208 (semaphorin-7A), FITC-conjugated anti–GM-CSFRα (CD116), PE-conjugated anti–IL-3Rα (CD123), or PE-conjugated anti–IL-5Rα (CD125) and a corresponding isotype control (BD Biosciences, San Jose, CA). PB EOS (AG) or BAL EOS (AG) were stained from whole blood or total BAL cells, as previously described (14). Ten thousand cells were acquired on a FACSCalibur (BD Biosciences). Dead cells were excluded using propidium iodide. Data were analyzed with FlowJo software (TreeStar, Ashland, OR) and expressed as the geometric mean or median channel fluorescence.
Blood eosinophils were activated with IL-3 or GM-CSF (2 ng/ml) for 14 h. Cells (6 × 106) were washed with PBS, and polyribosomes were prepared as previously described (14). The polyribosome pellet was washed once with lysis buffer without detergent and lysed in TRIzol reagent (Sigma) for RNA extraction.
Total RNA was extracted from eosinophils or the polyribosome preparation using an RNeasy Mini Kit (QIAGEN, Valencia, CA) or TRIzol reagent (Sigma), respectively. A reverse-transcription reaction was performed using the Superscript III system (Invitrogen/Life Technologies, Grand Island, NY). mRNA expression was determined by real-time quantitative PCR using SYBR Green Master Mix (SABiosciences, Frederick, MD), as previously described. Specific primers shown in Supplemental Table I were designed using Primer Express 3.0 (Applied Biosystems, Carlsbad, CA) and blasted against the human genome to determine specificity (http://www.ncbi.nlm.nih.gov/tools/primer-blast). The reference gene, β-glucuronidase forward: 5′-CAGGACCTGCGCACAAGAG-3′ and reverse: 5′-TCGCACAGCTGGGGTAAG-3′), was used to normalize the samples. Polyribosomes were quantified by 18S real-time quantitative PCR using the TaqMan human 18S rRNA endogenous control primers (reference sequence: X03205.1) and hydrolysis probe (VICR/MGB probe; Life Technologies). An Applied Biosystems 7500 Sequence detector was used. Standard curves were performed, and efficiencies were determined for each set of primers. Efficiencies ranged between 93 and 96%. Data are expressed as fold change using the comparative cycle threshold (∆∆CT) method, as previously described (1, 14).
Eosinophils were lysed with Tris-HCl (50 mM), NaCl (150 mM), 0.05% SDS, and protease inhibitors. Phosphatase activity in cell lysates was measured using the EnzChek Phosphatase Assay Kit, as recommended by the provider (Molecular Probes, Eugene, OR). The fluorescence was measured by a plate reader (Synergy HT; BioTek Instruments, Winooski, VT) using excitation at 350 nm and emission at 450 nm.
Differences between groups were analyzed using the unpaired or paired Student t test and the SigmaPlot 11.0 software package. In Fig. 4A, the ratios p90S6K-S380/Total p90S6K and RPS6 S235-S236/Total RPS6 were compared by source [PB EOS (CTRL), PB EOS (AG), BAL EOS (AG)] using mixed-effect linear models with a fixed effect for source and a random effect for experiment to account for within-experiment and within-donor correlation.
Eosinophils consume a large amount of IL-3
Exposure to >6-pM concentrations of IL-5, GM-CSF, or IL-3 increases eosinophil viability after a 1-d culture. However, concentrations between 250 and 700 pM are often used to evoke changes in eosinophil morphology, gene expression, or protein secretion. In the current study, unless indicated otherwise, the three βcR signaling cytokines were used at 2 ng/ml (∼130 pM). This concentration was chosen because the three cytokines have a very similar effect on eosinophil survival during the first 6 d in culture (Supplemental Fig. 1). Longer-term cultures (12 d) demonstrated that GM-CSF was significantly superior to IL-5 or IL-3 in promoting survival (Supplemental Fig. 1).
The cell surface levels of the ligand-binding α-chains of the IL-5, GM-CSF, and IL-3 receptors are differentially controlled by these cytokines. As previously described, IL-5 induced rapid loss of IL-5RA on eosinophils (31) (Fig. 1A). Conversely, over the same time period, IL-3 increased IL-3RA, whereas GM-CSF had no effect on its own α-chain receptor (CSF2RA; also called GM-CSFR) (Fig. 1A). In accordance with the receptor expression, cytokine consumption over a period of 48 h was minimal when 1 million eosinophils were cultured with rhIL-5 (2 ng/ml) (Fig. 1B). In contrast, as much as 2 ng of rhIL-3 was consumed by eosinophils, whereas only ∼200 pg/ml of the 2 ng/ml of rhGM-CSF was used (Fig. 1B). Of note, rhIL-3 did not degrade when incubated at 37°C for 48 h in cell-free medium or conditioned medium from GM-CSF–activated eosinophils (data not shown). These results suggest that the βcR signaling cytokines drive different outcomes in eosinophils, including continuous signaling downstream of the IL-3/IL-3R complex.
IL-3 induces and prolongs RPS6 phosphorylation
Among the ribosomal proteins, RPS6 has been characterized for its ability to be phosphorylated following cell activation. Thus, we asked whether βcR signaling cytokines altered the RPS6 phosphorylation state in eosinophils. Indeed, IL-3 triggered the phosphorylation of RPS6 (Fig. 2A). Phosphorylation was not detected in IL-5–activated eosinophils, and only a modest phosphorylation of RPS6 was observed in cells treated with GM-CSF for 14 h (Fig. 2A). These results were confirmed by fluorescence microscopy (Fig. 2B), which revealed that phospho-RPS6 was perinuclear and associated with the ER. A comparison of Pearson correlation coefficients of red and green pixel intensities demonstrated that colocalization of phospho-RPS6 and ER was greater in IL-3–activated versus GM-CSF–activated eosinophils (Fig. 2C). Notably, Akt (Fig. 2A) or p70S6K phosphorylation was marginally detectable, and phosphorylation of the eukaryotic initiation factors (eIF2α and eIF4E) was undetectable. eIF4E-binding protein (4E-BP1), a repressor of translation in its nonphosphorylated state, was strongly phosphorylated in both resting and cytokine-activated eosinophils (Fig. 2A), suggesting that 4E-BP1 does not regulate translation in eosinophils under these conditions. Kinetic studies revealed that GM-CSF induced rapid and transitory phosphorylation of RPS6 that was lost between 4 and 6 h (Fig. 2D). However, phosphorylation was maintained in IL-3–activated eosinophils over a period of 14 h (Fig. 2D). Dose-response studies (Fig. 2E) showed that significant and prolonged RPS6 phosphorylation was detectable even at low doses of IL-3 (0.2 ng/ml), and it was not significantly increased at concentrations > 2 ng/ml. In contrast, even at 10 ng/ml, GM-CSF had little effect on RPS6 protein phosphorylation (Fig. 2E). These data show differential signaling mediated by βcR signaling cytokines that culminate in RPS6 phosphorylation status.
IL-3 induces and prolongs p90S6K phosphorylation upstream of RPS6
RPS6 phosphorylation is usually driven by PI3K/mTOR/p70S6K. However, in eosinophils, blockade of this pathway (PI3K with LY294002 or mTOR with rapamycin) did not prevent IL-3–induced phosphorylation of RPS6 at 14 h (Fig. 3A). Because RPS6 phosphorylation was independent of this pathway, we investigated a possible role for p90S6K (also called RSK), which is phosphorylated during cell activation and was reported to phosphorylate RPS6 (32, 33). A p90S6K inhibitor (BI-D1870), but not an inhibitor of p70S6K activity (PF-4708671), blocked IL-3–induced RPS6 phosphorylation at 14 h (Fig. 3B). Moreover, in eosinophils, p90S6K was continuously phosphorylated by IL-3 for 16 h (Fig. 3C). In contradistinction, GM-CSF induced a strong, but relatively brief (between 10 min and 1 h), phosphorylation of p90S6K (Fig. 3C). Of note, similarly to GM-CSF, IL-5 did not maintain phosphorylation of p90S6K (data not shown). The prolonged phosphorylation of p90S6K in IL-3–activated eosinophils was confirmed using two other Abs directed against different phospho-sites (T573-S363-p90S6K and T359-S363-p90S6K, Supplemental Fig. 2). p90S6K phosphorylation is known to be regulated by the MAPKs and particularly by ERK1/2 (34). Consistent with these data, a selective inhibitor of both MEK1 and MEK2 (U0126), added 3 h after IL-3, blocked the phosphorylation of p90S6K on Ser380 in eosinophils for 4 h (Fig. 4A). Another MAPK, p38, also was implicated as a potential activator of p90S6K in dendritic cells (35). However, a p38 inhibitor (SB203580) did not affect p90S6K phosphorylation in IL-3–activated eosinophils (Fig. 4A). The phosphorylation of RPS6 also was reduced by U0126 added either before or 3 h after IL-3 treatment (Fig. 4B). Altogether, these data indicate that, unlike IL-5 and GM-CSF, IL-3R activation leads to prolonged RPS6 phosphorylation via ERK and p90S6K but not PI3K/mTOR/p70S6K.
eIF4B is another element of the translation machinery, and it is a target of p90S6K (36). In eosinophils, IL-3 induced only a slight phosphorylation of eIF4B (data not shown), providing further evidence for a novel and selective effect of IL-3 on RPS6.
p90S6K and RPS6 are phosphorylated in vivo in airway eosinophils
To determine the relevance of our in vitro data to allergic in vivo human disease, we analyzed p90S6K1 and RPS6 phosphorylation in eosinophils isolated from BAL [BAL EOS (AG)] 48 h after a localized (segmental) airway allergen challenge. This model allows acquisition of large numbers of human eosinophils from a site of allergic inflammation (11). Fig. 5A shows robust phosphorylation of both p90S6K and RPS6 in BAL EOS compared with PB EOS (AG) or PB EOS (CTRL). Phospho-eIF4B was not detectable in BAL EOS (AG) or PB EOS (AG).
The implications of IL-3 signaling through a p90S6K/RPS6 pathway in vivo remain uncertain. Using the same in vivo model (i.e., obtaining human eosinophils 48 h after allergen challenge), BAL eosinophils displayed lower levels of IL-5RA and higher levels of IL-3RA and GM-CSFRA compared with blood eosinophils obtained from the same allergen-challenged subjects (Fig. 5B). Because both IL-3 and GM-CSF are elevated in BAL fluids 48 h after airway allergen challenge (14, 20), we asked whether GM-CSF could potentially block IL-3–induced prolongation of RPS6 phosphorylation. To the contrary, in vitro stimulation of blood eosinophils with GM-CSF plus IL-3 induced greater RPS6 phosphorylation than did IL-3 alone (Fig. 6). These data indicate that GM-CSF amplifies the response to IL-3 and suggests that, in vivo, the presence of both cytokines may additively maintain the phosphorylation of RPS6 and p90S6K in airway eosinophils.
Direct effect of IL-3 on sustained intracellular signaling.
The rapid decrease in IL-3R during the first 4 h after the initiation of blood eosinophil activation with IL-3, combined with the consumption of IL-3 (Fig. 1), suggests that the IL-3/IL-3R complex might be quickly internalized. The internalization could then trigger an IL-3–independent second signal responsible for the sustained signaling observed in eosinophils. To test the constant requirement of IL-3 for the maintenance of p90S6K and RPS6 phosphorylation, a neutralizing anti–IL-3 Ab was added to eosinophils at different times after the initiation of cell activation. Fig. 7 indicates that p90S6K and RPS6 phosphorylation was inhibited, even when the anti–IL-3 Ab was added 4 h after the initiation of the culture. The lack of complete inhibition of RPS6 phosphorylation might be caused by a slower turnover of phosphorylation compared with p90S6K. Fig. 7 indicates that the constant presence and direct effect of IL-3 are required to maintain signaling via p90S6K. This suggests that a second signal may not be required to maintain activation of this pathway.
IL-3 and p90S6K control semaphorin-7A translation
To evaluate the function of IL-3 and the activation of the p90S6K/RPS6 pathway on translation, a nonradioactive, modified amino acid (l-azidohomoalanine) was pulsed into eosinophils for 4 h, 14 h after activation with either IL-3 or GM-CSF. After cell lysis, a chemoselective ligation was performed between the incorporated amino acid and a biotin–alkyne complex. Fig. 8A shows that global translation was increased by 40% in IL-3–activated eosinophils compared with GM-CSF–activated eosinophils and was significantly reduced by the p90S6K inhibitor BI-D1870 but not by the p70S6K inhibitor PF-4708671 (Fig. 8B). Phospho-RPS6 was implicated as a regulator of translatability of 5′ terminal oligopyrimidine tract (TOP) mRNAs (reviewed in Ref. 37). Thus, polyribosome preparations from eosinophils from two donors were assessed for TOP mRNA loading onto polyribosomes from IL-3–activated versus GM-CSF–activated eosinophils. Unlike semaphorin-7A mRNA (14), TOP mRNAs, PABP and EEF1A1, were not enriched in polyribosomes from IL-3–activated eosinophils (Supplemental Fig. 3). In addition, a recent study showed that hypophosphorylated RPS6 in liver impaired the expression of numerous transcripts coding for proteins involved in ribosomal biogenesis (37, 38). Among these transcripts, IL-3 had little or no effect on the expression of RRP12, NOP56, and GARE1 (data not shown).
To analyze more specifically newly synthesized semaphorin-7A, semaphorin-7A protein was immunoprecipitated from whole-cell lysates following the chemoselective ligation described above. Although almost no detectable newly synthetized semaphorin-7A was present in GM-CSF–activated eosinophils, there was an ∼14-fold increase in semaphorin-7A protein from IL-3–activated eosinophils (Fig. 8C), which was dependent on p90S6K (Fig. 8D). This difference in semaphorin-7A synthesis between IL-3 and GM-CSF activation was not due to semaphorin-7A protein degradation, because semaphorin-7A protein stability was unchanged after cycloheximide treatment (data not shown). Also, coding mRNA levels were very similar in IL-3–activated eosinophils that were treated or not with the p90S6K inhibitor, eliminating transcription or mRNA stability as a cause (data not shown).
Therefore, our data demonstrate that, compared with GM-CSF, IL-3 globally increases protein synthesis, and select mRNAs, including semaphorin-7A, are preferentially translated in a phospho-p90S6K–dependent manner.
In a previous study, we showed that IL-3 was more potent than GM-CSF or IL-5 in increasing cell surface semaphorin-7A on eosinophils (14). Consistent with this previous study and the data described above, we found in this study that membrane-associated semaphorin-7A protein was reduced by p90S6K inhibition, whereas p70S6K blockade had no effect on membrane semaphorin-7A (Fig. 9).
Phosphatase 1 dephosphorylates RPS6 in GM-CSF–activated eosinophils
To examine the mechanisms leading to sustained signaling in IL-3–activated cells, we sought to determine the phosphatase(s) responsible for RPS6 dephosphorylation in GM-CSF–activated eosinophils. Eosinophils were activated with GM-CSF for 5 h and treated with low doses of the phosphatase inhibitors okadaic acid or calyculin during the last 3.5 h of the culture. Fig. 10A shows that calyculin was a potent inhibitor of RPS6 dephosphorylation, whereas okadaic acid had no effect. Although both compounds can inhibit phosphatase 1 (PP1) and phosphatase 2A at low doses, calyculin and okadaic acid preferably inhibit PP1 and phosphatase 2A, respectively (39, 40). In accordance with previous studies (41), our data suggest that RPS6 dephosphorylation is regulated by PP1. PP1 activity was measured in resting or activated eosinophils, and it was not significantly different in resting eosinophils, IL-3–activated eosinophils, and GM-CSF–activated eosinophils (Fig. 10B), raising the possibility that specific PP1-induced RPS6 dephosphorylation is negatively regulated in IL-3–activated, but not GM-CSF–activated, eosinophils.
We showed that, among the βcR signaling cytokines, IL-3 is unique in prolonging the activation of p90S6K and RPS6 in human eosinophils. Although phosphorylation of RPS6 by the PI3K/mTOR/p70S6K pathway is thought to regulate translation in dividing cells, the ERK/p90S6K pathway was crucial for RPS6 activation and translation of the profibrotic protein semaphorin-7A in the nondividing eosinophils. Our data elucidate key differences in the mechanisms underlying the induction of protein translation by IL-3 versus IL-5 or GM-CSF and suggest the importance of IL-3 signaling in mediating eosinophil function.
IL-3–activated eosinophils were unique in that p90S6K, rather than the canonical PI3K/mTOR/p70S6K pathway (23), drove RPS6 phosphorylation. p90S6K was the first RPS6-phosphorylating kinase described in Xenopus eggs (42) and was since shown to be involved in human cell proliferation and survival (43). p90S6K includes three isoforms (RSK1, RSK2, and RSK3), all with inducible activity and similar functions. p90S6K is a well-known substrate for ERK and was reported to be associated with polyribosomes in neurons (44). p90S6K has four phosphorylation sites that are critical for its function. ERK1/2 initially phosphorylates threonine 573 and sequentially threonine 359, serine 363, and, finally, serine 380 in the linker and hydrophobic motifs of p90S6K. Ultimately, 3′-phosphoinositol-dependent kinase-1 phosphorylates serine 221, leading to full p90S6K activation (45). We detected all four phospho-sites initially targeted by ERK in IL-3–activated eosinophils in this study.
It does not appear that dynamic turnover of the α-chain of IL-3 and GM-CSF receptors is a contributing factor for the important consumption of IL-3, and the prolonged maintenance of p90S6K phosphorylation by IL-3 activation versus GM-CSF activation or IL-5 activation of eosinophils. Indeed, we and other investigators (46) did not detect downmodulation of GM-CSFRA following eosinophil activation with GM-CSF. In addition, there are incongruent effects of PI3K activation on receptor expression and RPS6 activation. Upregulation of IL-3RA by the βcR signaling cytokines is reported to be PI3K dependent (46), whereas we showed that RPS6 phosphorylation in IL-3–activated eosinophils was PI3K independent. Also, decreased IL-3R levels and high consumption of IL-3 could reflect IL-3/IL-3R internalization, which, in turn, can trigger a signal into the nucleus. The translocation of ligand/receptor complexes into the nucleus was described for IFN-γ and IL-5 (47, 48). For instance, the IFN-γ/IFNGR/JAK/STAT complex can enter the nucleus where it stimulates the transcription of specific genes in a noncanonical way (47). However, the use of a neutralizing anti–IL-3 Ab indicates that the constant presence of IL-3, rather than a secondary signal, accounts for sustained intracellular signaling in IL-3–activated eosinophils. In addition, a possible explanation for the GM-CSF–mediated attenuation of p90S6K and RPS6 phosphorylations is that GM-CSF triggers a negative-feedback response on its own receptor by activating an accessory receptor carrying an ITIM (49). Finally, the level of the common β-chain receptor could be downregulated by IL-3 compared with GM-CSF or IL-5. Yet, Gregory et al. (46) showed that GM-CSF or IL-5 did not differentially change the β-chain receptor level compared with IL-3 1 h after activation. However, 19 h after activation, the β-chain receptor level was lower in IL-3–activated eosinophils compared with those activated by GM-CSF or IL-5 (46), indicating that changes in the β-chain receptor level cannot account for the sustained signaling in IL-3–stimulated cells compared with GM-CSF–stimulated cells.
The function of RPS6 phosphorylation in nondividing cells, such as eosinophils and neutrophils, is unexplored. RPS6 phosphorylation was described in LPS-stimulated neutrophils; it was p38 and ERK dependent (50), but the possible function of p90S6K on RPS6, as well as translation regulation, was not analyzed. In this study, we showed that, compared with GM-CSF, IL-3 prolonged p90S6K activity and significantly increased global translation by ∼40% and the translation of semaphorin-7A by >10-fold. These data suggest that phosphorylation of RPS6 enhances protein synthesis from a subset of mRNAs, stimulating their binding to the 40S ribosomal subunit. This was suggested previously in proliferating cells in which RPS6 activation mediates ribosome biogenesis (51). Also, Thomas et al. (52) demonstrated that cell activation significantly increased the translation of a group of unidentified transcripts independently of transcription. This mechanism allows rapid translation of pre-existing transcripts. It is noteworthy that semaphorin-7A mRNA is abundant in resting eosinophils (14) but is modestly translated in the absence of IL-3 activation and prolonged RPS6 phosphorylation. Other ribosomal proteins, such as RPL13a and RPL26, can enhance the translation of specific mRNAs (53, 54). RPS6 phospho-target mRNAs may be those with a pyrimidine consensus sequence in their 5′-untranslated region (TOP mRNAs) and coding for proteins implicated in translation (55). Because our data suggest that TOP mRNAs (PABP or EEF1A1) were not further translated in IL-3–activated versus GM-CSF–activated eosinophils, other features must also be involved in the selection of the group of mRNAs affected by p90S6K/RPS6 phosphorylation in IL-3–activated eosinophils. Analysis by gene-wide microarrays of polyribosomes could yield important information, but it is technically challenging as a result of the meager recovery of polyribosome-associated RNA from eosinophils.
In addition to RPS6, eIF4B is a potential ribosomal p90S6K target (36). Phospho-eIF4B interacts with eIF3A, enhancing translation initiation (56). However, because of the fact that little or no phosphorylation of eIF4B in eosinophils was observed in vitro or in vivo, eIF4B is probably not responsible for the increased translation of semaphorin-7A. However, we cannot rule out that eIF4B might be involved in the modest increase in global translation (∼40%) in IL-3–activated versus GM-CSF–activated eosinophils. p90S6K can also target transcription factors, such as CREB and IκBα and, thus, increase transcription (57). However, inhibition of p90S6K activity had little effect on semaphorin-7A mRNA accumulation in IL-3–activated eosinophils.
We provide evidence that p90S6K and RPS6 are phosphorylated in vivo following airway allergen challenge of atopic individuals. Airway eosinophils lose most of their IL-5Rs (11), suggesting that GM-CSF and IL-3 drive eosinophil function after eosinophils migrate to the airway. This is supported by our data showing that activation of p90S6K/RPS6 signaling in airway eosinophils does not require stimulation ex vivo. Because GM-CSF and IL-3 have synergistic functions on RPS6 phosphorylation in cultured blood eosinophils, and both of these cytokines and their receptors are present in the airway (14, 20), it is reasonable to postulate that both cytokines participate in this signaling in airway eosinophils. Also, because we showed recently that airway eosinophils express higher semaphorin-7A levels compared with blood eosinophils (13), we propose that p90S6K/RPS6 signaling increased semaphorin-7A translation in airway eosinophils (15). Although the function of semaphorin-7A in eosinophils is largely unknown, it is profibrotic in lung and liver (14, 58, 59). We found previously that IL-3–activated eosinophils adhere to plexin-C1 (14), whose only known ligand is semaphorin-7A (60). Via an interaction with plexin-C1, semaphorin-7A could be involved in eosinophil migration, as demonstrated for neutrophils (61). In addition, beyond semaphorin-7A, a potential translational increase in a variety of cytokines, chemokines, and growth factors by eosinophils would have a strong immunomodulatory impact during an allergic disease (reviewed in Ref. 62).
In conclusion, our study shows important differences among the βcR signaling cytokines with regard to signaling in eosinophils. Through p90S6K, IL-3 has profound and distinctive effects on global and mRNA-specific translation. By enhancing p90S6K and RPS6 activity, IL-3 likely increases many proteins. Their identification is critical to appreciate the potential influence and value that blocking this pathway would have on eosinophil function. Further studies characterizing the underlying mechanisms, as well as the regulated proteins, may yield novel therapeutic opportunities to treat allergic diseases.
We thank the subject volunteers who participated in this study, the staff of the Eosinophil Core facility (University of Wisconsin) for blood and airway eosinophil purification, our research nurse coordinators for subject recruitment and screening, and our pulmonologists for assistance with bronchoscopies. We thank Larissa DeLain for laboratory technical support and Mike Evans for statistical analyses.
This work was supported by a Program Project Grant (National Institutes of Health HL088594) and by the University of Wisconsin Institute for Clinical and Translational Research (National Center for Research Resources/National Institutes of Health 1UL1RR025011).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- BAL EOS (AG)
eosinophils from BAL after an allergen challenge
mammalian target of rapamycin
- PB EOS (AG)
eosinophils from peripheral blood after an allergen challenge
- PB EOS (CTRL)
PB EOS from a control unchallenged subject
p70 ribosomal S6 kinase
90-kDa ribosomal S6 kinase
ribosomal protein S6
segmental bronchoprovocation with an allergen
terminal oligopyrimidine tract.
The authors have no financial conflicts of interest.