Mitogen‐activated protein kinase (MAPK) activation controls diverse cellular functions including cellular survival, proliferation, and apoptosis. Tuning of MAPK activation is counter‐regulated by a family of dual‐specificity phosphatases (DUSPs). IL‐33 is a recently described cytokine that initiates Th2 immune responses through binding to a heterodimeric IL‐33Rα (ST2L)/IL‐1α accessory protein (IL‐1RAcP) receptor that coordinates activation of ERK and NF‐κB pathways. We demonstrate here that DUSP5 is expressed in eosinophils, is upregulated following IL‐33 stimulation and regulates IL‐33 signaling. Dusp5−/− mice have prolonged eosinophil survival and enhanced eosinophil effector functions following infection with the helminth Nippostrongylus brasiliensis. IL‐33‐activated Dusp5−/− eosinophils exhibit increased cellular ERK1/2 activation and BCL‐XL expression that results in enhanced eosinophil survival. In addition, Dusp5−/− eosinophils demonstrate enhanced IL‐33‐mediated activation and effector functions. Together, these data support a role for DUSP5 as a novel negative regulator of IL‐33‐dependent eosinophil function and survival.
IL‐33 plays an important role in allergic diseases and fibrosis through activation of NF‐κB and mitogen‐activated protein kinase (MAPK) pathways. DUSP5, a dual‐specificity phosphatase, is significantly upregulated following IL‐33 stimulation and attenuates eosinophil functions and survival.
DUSP5 is a negative regulator of IL‐33 signaling.
DUSP5 modulates ERK, but not NF‐κB, activation to regulate pro‐inflammatory cytokines and eosinophil survival, the latter through a MEK/BCL‐XL pathway.
Dusp5−/− eosinophils are hyperactivated and have prolonged survival following helminthic infection or following IL‐33 administration.
Eosinophils comprise a distinct innate immune subset that contributes to host homeostasis, helminth‐induced immunity, and allergic diseases. While the role of eosinophils in helminth‐induced immunity has been a long‐standing area of investigation and accumulation of eosinophils in end‐organs has been appreciated to be a common feature of many human diseases involving the respiratory tract (asthma and allergic rhinitis), gastrointestinal tract (gastroenteritis, eosinophilic esophagitis, and inflammatory bowel diseases), and skin (atopic dermatitis) (reviewed in Fulkerson & Rothenberg, 2013; Rosenberg et al, 2013), recent studies continue to reveal additional roles for eosinophils in controlling host homeostatic processes including glucose metabolism and tissue regeneration (Wu et al, 2011; Goh et al, 2013; Heredia et al, 2013).
Eosinophils arise in the bone marrow (BM) from CD34+ hematopoietic stem cells, and development is governed by the coordinated actions of GATA‐1, PU.1, and C/EBP transcription factors and regulated by a multitude of cytokines (Akuthota & Weller, 2012). GM‐CSF and IL‐3 direct the early differentiation of granulocytes that include eosinophils, neutrophils, basophils, and mast cells. The latter stages of eosinophil differentiation are controlled by IL‐5. Il5−/− or Il5ra−/− mice have a twofold reduction in eosinophil numbers under homeostatic conditions and are unable to increase BM, blood, or tissue eosinophils following infection with the metacestode parasite Mesocestoides corti (Kopf et al, 1996) or the nematode Nippostrongylus brasiliensis (Knott et al, 2007). A similar inability to induce eosinophilia following N. brasiliensis infection occurs with administration of a neutralizing anti‐IL‐5 mAb (Coffman et al, 1989). In addition, treatment of allergic asthma patients with anti‐IL‐5 antibodies reduced numbers of blood, BM, and sputum eosinophils, as well as BM eosinophil myelocytes and metamyelocytes (Menzies‐Gow et al, 2003). Treatment of patients with an anti‐IL‐5 mAb has demonstrated a reduction in disease exacerbations in patients with refractory eosinophilic asthma and reduction in corticosteroid use in patients with hypereosinophilic syndrome (Haldar et al, 2009; Bel et al, 2014; Ortega et al, 2014). Hence, IL‐5 plays an important role in eosinophil maturation, parasite‐induced eosinophilia in rodents, and human eosinophilic diseases.
IL‐33, a recently described cytokine of the IL‐1 family, has emerged as an important mediator of Th2 biology and fibrosis (Milovanovic et al, 2012). Single nucleotide polymorphisms (SNP) in IL33 and its cognate receptor, IL1RL1, are associated with the development of asthma in humans (Moffatt et al, 2007; Gudbjartsson et al, 2009; Torgerson et al, 2011). IL‐33 is highly expressed in the airway epithelium of patients with allergic asthma (Sakashita et al, 2008; Prefontaine et al, 2009) and in intestinal epithelial cells and myofibroblasts of patients with active ulcerative colitis (Kobori et al, 2010). Levels of the IL‐33 soluble ST2 decoy receptor are elevated in human asthma patients during clinical exacerbation (Oshikawa et al, 2001). In addition, Il33−/− mice have reduced airway inflammation, while mice overexpressing IL‐33 have enhanced airway inflammation (Oboki et al, 2010; Zhiguang et al, 2010). IL‐33 has pleiotropic effects and acts on multiple hematopoietic‐derived cell types, including CD4+ Th2 cells, dendritic cells, B1 B cells, mast cells, basophils, neutrophils, macrophages, innate lymphoid type 2 cells (ILC2s), eosinophils, and natural killer (NK) cells, as well as non‐hematopoietic epithelial and endothelial cells. The variety of cell types affected underscores the broad evolving biology of IL‐33 that includes modulation of cytokines, chemotaxis, cell differentiation, cell proliferation, and fibrosis.
IL‐33 has recently been demonstrated to exert a direct effect on eosinophils. IL‐33 augments eosinophil differentiation from CD117+ BM hematopoietic progenitor cells (Stolarski et al, 2010), triggers production of pro‐inflammatory cytokines by mature eosinophils, augments eosinophil‐mediated polarization of M2 macrophages (Kurowska‐Stolarska et al, 2009), and exacerbates eosinophil‐mediated airway inflammation. While Il33−/− mice have normal eosinophil development, they are unable to mount tissue eosinophilic responses following infection with Strongyloides venezuelensis (Yasuda et al, 2012) or N. brasiliensis (Hung et al, 2013).
IL‐33 signals through a heterodimeric receptor consisting of the IL‐33R α‐chain (ST2L) and IL‐1α accessory protein (IL‐1RAcP). Receptor ligation activates both NF‐κB and MAPK pathways (Chow et al, 2010) to upregulate cell surface expression of adhesion proteins β2‐integrin CD11b and ICAM‐1 (Suzukawa et al, 2008), induces secretion of pro‐inflammatory cytokines including IL‐13, IL‐6, and IL‐8, and provides a potent survival signal for eosinophils (Suzukawa et al, 2008).
DUSPs (also known as Mitogen‐activated Protein Kinase Phosphatases) include sixteen catalytically active enzymes in mammalian cells (Caunt & Keyse, 2013). Ten of these comprise the ‘typical’ DUSPs that share a common structure with an amino‐terminal domain that regulates binding of specific MAPK isoforms and DUSP subcellular localization and a carboxy‐terminal catalytic domain that removes phosphates on tyrosine and serine/threonine residues. These DUSPs are divided into three subfamilies: JNK/p38 selective DUSPs (DUSP8, DUSP10 [MKP‐5], and DUSP16 [MKP‐7]), ERK selective cytoplasmic MKPs (DUSP6 [MKP‐3], DUSP7 [MKP‐X], and DUSP9 [MKP‐4]), and inducible nuclear MKPs (DUSP1 [MKP‐1], DUSP2, DUSP4 [MKP‐2], and DUSP5). In immune cells, the functions of DUSPs 1, 2, 4, and 10 have been more extensively studied (reviewed in Lang et al, 2006; Jeffrey et al, 2007). These DUSPs regulate their cognate MAPK pathways, and overexpression and genetic ablation studies have demonstrated important roles in cytokine secretion and nitric oxide production in macrophages, mast cell apoptosis, and T‐cell function.
DUSP5 is a growth‐factor‐inducible phosphatase that targets and anchors ERK1 and ERK2, but not other MAP kinases, to the nucleus (Mandl et al, 2005). DUSP5 is preferentially expressed in immune cells, and overexpression studies have demonstrated a role for DUSP5 in M‐CSF signaling during macrophage development (Grasset et al, 2010), common gamma chain signaling (Kovanen et al, 2003) in T‐lymphocyte development/proliferation (Kovanen et al, 2008) and plasma cell differentiation (Rui et al, 2006). We report here the generation of mice deficient in Dusp5 and describe a novel mechanistic role for DUSP5 in IL‐33‐mediated activation of ERK1/2 in eosinophil survival and function.
DUSP5 regulates eosinophilia induction during helminth infection
To explore the functions of DUSP5, we analyzed Dusp5 mRNA from sorted splenic cells from mice. Dusp5 mRNA was highest in eosinophils and NK cells and, to a lesser extent, CD4+ T lymphocytes (Fig 1A). To better understand the physiologic functions of DUSP5, mice deficient in Dusp5 were generated (Supplementary Fig S1A). Southern blot analysis confirmed the predicted genomic incorporation (Supplementary Fig S1B). RT–PCR with primers spanning exons 2–4 downstream of the deleted region confirmed the absence of Dusp5 mRNA (Supplementary Fig S1C). Western blot analysis confirmed the absence of DUSP5 protein (Supplementary Fig S1D). Mice deficient in Dusp5 were developmentally normal, presented no gross developmental or growth abnormalities, and were fertile.
Since mice expressing a DUSP5 transgene under the H2‐Kb promoter and immunoglobulin heavy chain enhancer demonstrated a block in thymocyte development at the CD4+CD8+ double‐positive (DP) stage (Kovanen et al, 2008), we initially characterized the effects of Dusp5 deficiency on T‐cell development. Total thymocyte numbers were normal, though there were modest increases in CD4+ and CD8+ thymocytes in Dusp5−/− mice when compared to Dusp5+/+ mice (Supplementary Fig S2A). No differences in CD4+ or CD8+ T‐cell numbers were observed in spleen or lymph nodes (Supplementary Fig S2B and C). As overexpression of DUSP5 also decreased IL‐2‐augmented T‐cell proliferation (Kovanen et al, 2008), we examined the function of peripheral CD4+ T cells. Sorted naïve CD62LhiCD4+ T cells from Dusp5−/− mice proliferated to a greater degree following activation with anti‐CD3 and anti‐CD28 mAbs (Supplementary Fig S3A). In contrast, effector/memory CD62LloCD4+ T cells from Dusp5−/− mice proliferated at a rate similar to Dusp5+/+ cells (Supplementary Fig S3B). These modest differences observed in T cells are consistent with the previously described phenotypes observed with DUSP5 overexpression (Kovanen et al, 2008). In addition, we did not observe differences in innate or adaptive cell numbers in the BM, spleen, blood, and inguinal lymph nodes in Dusp5−/− mice compared to Dusp5+/+ mice (Supplementary Table S1 and Supplementary Fig S2B and C).
Given the higher level of Dusp5 expression in eosinophils, we focused on the effects of Dusp5 deficiency on eosinophil functions. Because eosinophils regulate host responses to helminthic infections, we analyzed the effects of Dusp5 deficiency in mice infected with N. brasiliensis. Helminth infection evokes a Th2 response characterized by increased systemic eosinophilia (Coffman et al, 1989). Since Dusp5−/− mice have a modest effect on T‐cell functions, we crossed Dusp5−/− mice onto a Rag2‐deficient background to eliminate potential contributions from Dusp5−/− T and B lymphocytes. Dusp5−/− Rag2−/− mice accumulated a greater percentage of circulating eosinophils at days 6 and 13 following N. brasiliensis infection when compared to Dusp5+/+ Rag2−/− mice (Fig 1B). In addition, increased eosinophils were observed in the blood, bronchoalveolar lavage fluid (BALF), spleen, and BM 14 days following N. brasiliensis infection (Fig 1C–F). This increased systemic eosinophilia in Dusp5−/− Rag2−/− mice was only observed following helminth infection, since BM and splenic eosinophil numbers are equivalent in uninfected Dusp5+/+ Rag2−/− and Dusp5−/− Rag2−/− mice (Supplementary Table S1). No differences in neutrophil, monocyte, NK, or ILC2 cell numbers were observed (Supplementary Fig S4A–C). Corresponding with increased eosinophils, Dusp5−/− Rag2−/− mice had a lower worm burden compared to Dusp5+/+ Rag2−/− mice (Fig 1G). A similar increase in circulating and BALF eosinophils was observed in germline Dusp5−/− Rag2+/+ mice (Supplementary Fig S5A–C). Together, these data suggest that DUSP5 plays a critical role in regulating the eosinophilic response to N. brasiliensis infection.
As Dusp5 expression is also increased in NK cells (Fig 1A), we analyzed whether NK cells contributed to the lower worm burden observed in Dusp5−/− Rag2−/− mice. Dusp5−/− Rag2−/− mice treated with an anti‐Asialo GM1 antibody (Ab) to deplete NK cells (Supplementary Fig S4E) still maintained a lower worm burden compared to anti‐Asialo GM1 Ab‐treated Dusp5+/+ Rag2−/− mice (Supplementary Fig S4D). In addition, eosinophils remained higher in anti‐Asialo GM1 Ab‐treated Dusp5−/− Rag2−/− mice when compared to anti‐Asialo GM1 Ab‐treated Dusp5+/+ Rag2−/− mice (Supplementary Fig S4F). Hence, despite high levels of DUSP5 expression, NK cells appear dispensable for eosinophilia or worm burden phenotypes observed in Dusp5−/− Rag2−/− mice.
DUSP5 functions as a negative regulator of eosinophil survival and activation
To investigate the basis for the increased numbers of Dusp5−/− eosinophils following N. brasiliensis infection, we analyzed the effects of Dusp5 deficiency on eosinophil progenitors. In non‐infected mice, numbers of BM eosinophil progenitors (defined by Lin− CD34+ cKitint Sca‐1− IL‐5Rα+ staining) were not affected by Dusp5 deficiency (Supplementary Fig S6A and B). To analyze the differentiation of eosinophils during N. brasiliensis infection, mice were injected intraperitoneally with 1 mg BrdU every 12 h for 24 h and eosinophils analyzed 36 h post‐BrdU injection (12 h chase). Similar numbers of BrdU+ eosinophils were observed in the BM, blood, and spleen from Dusp5+/+Rag2−/− and Dusp5−/− Rag2−/− mice, consistent with equivalent rates of precursor proliferation (Fig 2A). Since the half‐life of mature eosinophils during helminth infection is > 36 h (Ohnmacht et al, 2007), we extended the chase period to 60 h to measure the relative turnover of eosinophils during infection. At 84 h post‐BrdU injection (60 h chase), a two‐ to threefold increase in BrdU+ eosinophils was observed in BM, blood, and spleen of N. brasiliensis‐infected Dusp5−/− Rag2−/− mice compared to Dusp5+/+ Rag2−/− mice (Fig 2B). We also observed a significant decrease in apoptotic annexin V+ propidium iodide (PI)− eosinophils in BM and spleen of infected Dusp5−/− Rag2−/− mice (Fig 2C). A reduction in annexin V+ PI− eosinophils was observed in blood, but was not statistically significant (P = 0.08). These findings suggest that the absence of DUSP5 does not affect the proliferation of eosinophil precursors, but prolongs survival of eosinophils following N. brasiliensis infection.
To test whether the increased eosinophil survival advantage in Dusp5−/− mice was intrinsic to the hematopoietic compartment, we established mixed BM chimeras by reconstituting Dusp5+/+ or Dusp5−/− BM competitively with host BM in a 1:1 ratio (Supplementary Fig S7A). Donor CD45.2+ Dusp5−/− and Dusp5+/+ BM reconstituted at comparable rates under homeostatic conditions (Supplementary Fig S7B). Chimeric mice were then infected with N. brasiliensis and the relative eosinophil percentage in the blood analyzed 9 days following infection. Donor‐derived CD45.2+ Dusp5−/− eosinophil percentage was increased compared to CD45.2+ Dusp5+/+ eosinophils (Fig 2D). In contrast to the increased Dusp5−/− eosinophils, no differences in neutrophil numbers were observed between donor‐derived (CD45.2+) Dusp5+/+ and Dusp5−/− BM (Supplementary Fig S7C). Together, these data suggest that the survival advantage of Dusp5−/− eosinophils is intrinsic to the hematopoietic compartment.
We next explored the effect of DUSP5 deficiency on eosinophil function during N. brasiliensis infection. Eosinophils are generated in the BM from eosinophil progenitors and form a pool of cells capable of replenishing the peripheral population. Following egress from the BM, eosinophils upregulate SiglecF and CD11b, markers associated with functional activation following tissue deposition or response to infection (Voehringer et al, 2007). Eosinophils harvested from blood, spleen, and, to a lesser extent, BM of N. brasiliensis‐infected Dusp5−/− Rag2−/− mice had higher surface expression of CD11b compared to eosinophils from infected Dusp5+/+ Rag2−/− mice (Fig 3A and B). Increased eosinophil activation was intrinsic to the hematopoietic cell compartment, since we also observed increased CD11b surface expression on donor‐derived (CD45.2+) Dusp5−/− blood eosinophils from the mixed BM chimeric mouse experiments following N. brasiliensis infection described above (Fig 3C and D), while CD11b surface levels on host‐derived (CD45.1+) blood eosinophils were similar (Supplementary Fig S7D). These observations are consistent with enhanced eosinophil activation in mice lacking DUSP5.
DUSP5 negatively regulates IL‐33‐dependent survival in eosinophils
The coordinated actions of IL‐5 and IL‐33 regulate eosinophil hematopoiesis (Le et al, 2013). Both Il5−/− and Il33−/− mice develop normally and have minimally reduced and normal eosinophil numbers, respectively, under homeostatic conditions (Kopf et al, 1996). However, IL‐5 and IL‐33 are critical regulators of peripheral eosinophilia following helminthic infection, as both Il5−/− and Il33−/− mice fail to induce eosinophilia (Knott et al, 2007; Hung et al, 2013). To test whether Dusp5 is upregulated in response to these cytokines, BM‐derived eosinophils were treated with IL‐5 or IL‐33 and Dusp5 mRNA levels analyzed. Dusp5 mRNA and DUSP5 protein were significantly elevated following treatment with IL‐33 (Fig 4A). In contrast, Dusp5 mRNA levels were unchanged with IL‐5 treatment in mouse BM‐derived eosinophils (Supplementary Fig S8).
To examine the functional consequences of Dusp5 deficiency on IL‐33‐mediated eosinophil survival, we examined the degree of apoptosis in BM‐derived eosinophils in the presence of a suboptimal dose of IL‐33. IL‐33‐treated cultures of Dusp5−/− eosinophils had less apoptotic (annexin V+ 7AAD−) and necrotic (annexin V+ 7AAD+) cells when compared to Dusp5+/+ eosinophils (Fig 4B). Decreased apoptosis of Dusp5−/− BM‐derived eosinophils was associated with a reduction in mitochondrial membrane depolarization (Fig 4C) as assessed by the cationic dye JC‐1, which selectively enters the mitochondria in response to mitochondrial membrane potential decreases associated with apoptosis. No differences between Dusp5+/+ and Dusp5−/− BM‐derived eosinophils were observed when cells were treated with titrating amounts of IL‐5 (Supplementary Fig S9A). Dusp5−/− BM‐derived eosinophils also had an IL‐33‐dependent survival advantage in long‐term cultures, while IL‐5‐treated Dusp5−/− and Dusp5+/+ eosinophils demonstrated similar cell survival (Fig 4D). Finally, to confirm the role of DUSP5 in IL‐33‐mediated survival of eosinophils in vivo, recombinant IL‐33 was administered intraperitoneally to Dusp5+/+ Rag2−/− and Dusp5−/− Rag2−/− mice. Consistent with observations in vitro in BM‐derived eosinophils (Fig 4B–D) and in vivo following N. brasiliensis infection (Fig 1B–F), IL‐33 induced twice as many peritoneal eosinophils in Dusp5−/− Rag2−/− mice compared to Dusp5+/+ Rag2−/− mice (Fig 4E). Together, these data support a role for DUSP5 in IL‐33 regulation of eosinophil survival.
IL‐5‐dependent eosinophil survival is mediated through the upregulation of pro‐survival factors such as BCL‐XL, MCL‐1, and, to a lesser extent, BCL‐2 (reviewed in Kankaanranta et al, 2005). To date, the mechanism of IL‐33‐mediated survival in eosinophils has not been studied. Hence, we examined the effects of IL‐33 and the consequences of Dusp5 deficiency on BCL family members. While IL‐5‐ and IL‐33‐treated BM‐derived eosinophils had increased levels of cytoplasmic BCL‐2 compared to cytokine‐starved cells, no differences were observed between Dusp5+/+ and Dusp5−/− BM‐derived eosinophils (Fig 5A, middle row, and 5B). In contrast, IL‐33, but not IL‐5, increased cytoplasmic BCL‐XL (Fig 5A, top row). Moreover, Dusp5−/− BM‐derived eosinophils stimulated with IL‐33 demonstrated a twofold greater increase in cytoplasmic BCL‐XL than Dusp5+/+ eosinophils (Fig 5A and B). Increased cytoplasmic BCL‐XL was, in part, due to transcriptional upregulation of Bclxl, as Dusp5−/− BM‐derived eosinophils had higher levels of Bclxl mRNA than Dusp5+/+ (Fig 5C). Consistent with the lack of detectable cytoplasmic BCL‐XL protein following treatment with IL‐5 (Fig 5A), Bclxl mRNA was not induced by IL‐5 (Supplementary Fig S9B).
Membrane fraction‐associated BCL‐XL and MCL‐1 were similar in Dusp5+/+ and Dusp5−/− BM‐derived eosinophils treated with IL‐5 or IL‐33 (Supplementary Fig S9C). Membrane fraction‐associated BCL‐2 was not detected in unstimulated or cytokine‐treated BM‐derived eosinophils (Supplementary Fig S9C). Membrane fraction‐associated BID was upregulated following IL‐5 or IL‐33 treatment but did not significantly differ between Dusp5+/+ and Dusp5−/− BM‐derived eosinophils (Supplementary Fig S9D, second row). Membrane‐associated BAX levels were comparable between Dusp5+/+ and Dusp5−/− BM‐derived eosinophils (Supplementary Fig S9D, top row). Pro‐apoptotic members BIM, BAK, PUMA, and BAD were undetectable in membrane fractions (Supplementary Fig S9D).
Consistent with the growth survival functions of IL‐5 and IL‐33, cytoplasmic cytochrome c (cyt c) levels and cleaved caspase‐3 levels were decreased with IL‐5 or IL‐33 treatment compared to cytokine‐starved cells (Fig 5D and E). However, cytoplasmic cyt c levels differed between Dusp5+/+ and Dusp5−/− BM‐derived eosinophils only when treated with IL‐33 (Fig 5E).
Dusp5 negatively regulates IL‐33 signaling in eosinophils
To further characterize the functional consequence of Dusp5 deficiency on IL‐33‐treated BM‐derived eosinophils, we analyzed global gene expression changes by microarray analysis. A number of genes important in eosinophil function were differentially regulated in IL‐33‐treated Dusp5−/− BM‐derived eosinophils compared to Dusp5+/+ eosinophils (Fig 6A). Genes known to function in eosinophil activation (Cd69, Spred2), effector functions (Il13, Csf2, Il1b), and transcriptional regulation (Fosl1) were significantly upregulated in Dusp5−/− eosinophils. Similar differences in Il4 and Il13 mRNA, but not Tnfa, levels were observed in freshly isolated peritoneal eosinophils elicited with recombinant IL‐33‐treated Dusp5−/− mice (Fig 6B). Consistent with the in vitro and in vivo mRNA analysis (Fig 6A and B), IL‐33‐treated Dusp5−/− BM‐derived eosinophils secreted greater amounts of IL‐4 and IL‐13 proteins compared to Dusp5+/+BM‐derived eosinophils, while TNF‐α secretion was unchanged (Fig 6C). In contrast, no differences in IL‐4, IL‐13, or TNF‐α levels were observed between supernatants of Dusp5+/+ and Dusp5−/− BM‐derived eosinophils following IL‐5 treatment (Supplementary Fig S10C). Similarly, RT–PCR analysis of IL‐33‐treated Dusp5−/− BM‐derived eosinophils confirmed higher levels of Fosl1 mRNA compared to Dusp5+/+ eosinophils (Fig 6D). Finally, surface expression of CD11b and CD69 was higher on IL‐33‐stimulated Dusp5−/− BM‐derived eosinophils compared to Dusp5+/+ eosinophils (Fig 6E and F). In contrast, no difference in CD11b was observed following IL‐5 treatment and CD69 was not induced by IL‐5 treatment (Supplementary Fig S10A and B). Together, these data are consistent with a negative regulatory role of DUSP5 in IL‐33‐, but not IL‐5‐, mediated eosinophil activation.
DUSP5 inhibits IL‐33‐dependent ERK activation to regulate eosinophil survival and function
As DUSP5 interacts with and regulates ERK1/2, but not JNK or p38, in fibroblasts and T cells (Mandl et al, 2005; Kovanen et al, 2008), we assessed the activation of ERK1/2 following IL‐33 stimulation in BM‐derived eosinophils. Phospho‐ERK1 (p‐ERK1) and phospho‐ERK2 (p‐ERK2) were increased in IL‐33‐stimulated Dusp5−/− BM‐derived eosinophils relative to Dusp5+/+ eosinophils (Fig 7A), while no difference in p‐ERK1 was observed in IL‐5‐stimulated Dusp5+/+ and Dusp5−/− eosinophils (Supplementary Fig S11A). We did not detect p38 and JNK phosphorylation following IL‐33 stimulation (Supplementary Fig S11B), and phosphorylation of the p65 subunit of NF‐κB was similar in IL‐33‐stimulated Dusp5+/+ and Dusp5−/− BM‐derived eosinophils (Fig 7B). Consistent with this observation, analysis of microarray data from IL‐33‐treated BM‐derived eosinophils showed upregulation of multiple ERK, but not NF‐κB, targeted genes in Dusp5−/− eosinophils compared to Dusp5+/+ eosinophils (Supplementary Table S2).
To link the requirement of ERK signaling to IL‐33‐mediated survival, we analyzed the effects of ERK inhibition through use of the MEK1/2 inhibitor U0126 (Favata et al, 1998). IL‐33‐mediated upregulation of CD11b was partially inhibited with the addition of U0126 (Fig 7C). Inhibition of MEK1/2 also decreased cytoplasmic BCL‐XL but not BCL‐2 following IL‐33 treatment (Fig 7D and E). Finally, MEK inhibition reduced IL‐33‐, but not IL‐5‐, mediated eosinophil viability (Fig 7F and Supplementary Fig S11C). Together, these data provide a mechanistic link between IL‐33‐mediated cellular ERK1/2 activation with eosinophil activation and survival.
Our studies here provide the first evidence for a role for DUSPs as a regulator of pro‐survival signaling in immune cells. DUSP5 curtails IL‐33‐mediated survival signals by attenuating the activation of ERK1/2 and controlling BCL‐XL expression and function in eosinophils. DUSPs 1, 2, and 16 have been implicated in controlling cell death processes, but primarily through regulation of pro‐apoptotic pathways. Overexpression of DUSP1 in a U937 pro‐monocytic cell line inhibits caspase‐3‐mediated apoptosis (Franklin et al, 1998). Overexpression of DUSP16 (MKP7) in Ba/F3 cells or T cells protects from cytokine withdrawal and TCR‐mediated apoptosis, respectively. Conversely, knock down of DUSP16 in T cells increases TCR‐mediated apoptosis (Hoornaert et al, 2003; Kiessling et al, 2010). Dusp2−/− mast cells upregulate pro‐apoptotic genes and undergo increased spontaneous apoptosis in long‐term cultures (Jeffrey et al, 2007). In addition to DUSPs that serve as negative regulators of apoptosis, DUSP4 is required for TGF‐β‐induced apoptosis (Ramesh et al, 2008). TGF‐β induces Dusp4 expression in a SMAD3‐dependent manner and, in turn, DUSP4 curtails ERK1/2 signaling, increases BIM stability, and promotes apoptosis. Our studies here broaden the mechanisms by which DUSPs can regulate cellular survival.
ERK1/2 activation has been critically linked to survival signals in tumorigenesis through transcriptional and post‐translational regulation of both pro‐ and anti‐apoptotic BCL family members (Balmanno & Cook, 2009). ERK activation downregulates transcription of pro‐apoptotic members Bim and Bad and can regulate complex assembly of BCL‐XL and MCL‐1. Conversely, ERK1/2 activation can induce the transcription of Mcl1, Bcl2, and Bclxl mRNAs through direct binding of AP‐1 and CREB to their promoter regions (Ballif & Blenis, 2001; Sevilla et al, 2001; McCubrey et al, 2007), as well as phosphorylate and stabilize MCL‐1 and BCL‐2 to promote cell survival (Domina et al, 2004). BAX is highly expressed in eosinophils, while other pro‐apoptotic (BAK, BID, BIK, and BAD) proteins are absent or expressed at low levels. ERK inactivates BAX in human eosinophils by phosphorylating Thr167 following GM‐CSF treatment, retaining it in an ‘inactive’ form by facilitating interaction with the peptidyl‐prolyl isomerase PIN1 (Shen et al, 2009). In our studies, we detected high levels of BAX expression, although no differences in expression levels or cellular distribution were observed between Dusp5+/+ and Dusp5−/− eosinophils. Whether PIN1 or its association with BAX is regulated following IL‐33 treatment or whether there are any alterations in BAX phosphorylation or BAX association with BCL‐XL in Dusp5−/− eosinophils will require further investigation.
IL‐33 has recently been shown to provide pro‐survival signals in eosinophils and other cells, though the mechanisms have not been elucidated (Cherry et al, 2008; Suzukawa et al, 2008). Microarray analysis of murine eosinophils has demonstrated that IL‐33, but not IL‐4, upregulated Dusp5 mRNA expression (Bouffi et al, 2013). In our studies, IL‐33, but not IL‐5, increased Dusp5 mRNA expression. In addition, IL‐33 induced BCL‐XL but not BCL‐2 expression in a MEK‐dependent fashion (Fig 7D and E). Consistent with this differential control of BCL‐2 and BCL‐XL, Dusp5−/− eosinophils with increased cellular ERK activation had increased BCL‐XL expression and increased accumulation of cytoplasmic BCL‐XL, while BCL‐2 expression or localization was unaffected. In ILC2 cells, IL‐33 induces expression of BCL‐2 and BCL‐XL (Kabata et al, 2013). A role for BCL‐XL has recently been demonstrated in the pro‐survival effects of IL‐33 in human mast cells (Wang et al, 2014). In non‐immune cells, IL‐33 can also provide survival signals for hepatocytes during liver ischemia/reperfusion and Con A‐induced hepatitis. In both cases, cytoprotection was associated with BCL‐2 upregulation, though BCL‐XL levels were not analyzed (Sakai et al, 2012; Volarevic et al, 2012). In light of our observations in eosinophils, whether IL‐33 also facilitates cell survival in other cell types in a DUSP5/ERK‐dependent manner warrants additional investigation.
While IL‐5 also induces ERK activation, IL‐5‐mediated eosinophil survival or activation was not affected by Dusp5 deficiency. In addition, MEK inhibition did not affect IL‐5‐mediated survival. We did not observe appreciable levels of cytoplasmic‐associated BCL‐XL in IL‐5‐differentiated BM‐derived eosinophils, and treatment with high doses of IL‐5 did not induce Bclxl expression (Fig 5A and Supplementary Fig S9B). A previous study reported that BCL‐XL expression was downregulated upon cytokine withdrawal and restored with IL‐5 (Dibbert et al, 1998), though this finding has not been consistently observed (Dewson et al, 1999). Our finding that IL‐5 does not regulate Dusp5 mRNA levels in BM‐derived eosinophils supports the notion that DUSP5 is a selective regulator of IL‐33‐dependent functions in mouse eosinophils.
In addition to cell survival, Dusp5−/− eosinophils also demonstrated increased cellular activation (CD11b and CD69) and cytokine production in response to IL‐33. These augmented IL‐33‐mediated functions are also MEK dependent and consistent with a previous study demonstrating MEK dependency of IL‐33‐mediated upregulation of cytokines, chemokines and surface expression of integrins on human eosinophils (Chow et al, 2010). Other DUSP family members have also demonstrated requisite or inhibitory roles in immune cell effector functions. Dusp1−/− and Dusp4−/− macrophages secrete more IL‐6 and TNF‐α in response to Toll‐like receptor (TLR) activation when compared to wild‐type cells (Hammer et al, 2006; Al‐Mutairi et al, 2010). A similar heightened response for Dusp1−/− macrophages has also been described for glucocorticoid receptor activation (Abraham et al, 2006). In contrast to these inhibitory DUSP functions, DUSP2 plays a requisite role in TLR and IgE receptor functions in macrophages and mast cells, respectively (Jeffrey et al, 2006). Dusp2−/− macrophages and mast cells secrete less TNF‐α, IL‐6, PGE2, and nitric oxide in response to TLR stimulation and decreased IL‐6 and TNF‐α production following IgE receptor activation, respectively. Finally, DUSP10 plays both positive and inhibitory roles in cellular functions. Dusp10−/− macrophages and T cells demonstrate enhanced production of pro‐inflammatory cytokines, while Dusp10−/− T cells exhibit a profound defect in T‐cell proliferation (Zhang et al, 2004).
Dusp5−/− Rag2−/− mice were able to more efficiently clear parasite burden following infection with N. brasiliensis. Consistent with our working hypothesis that DUSP5 regulates IL‐33‐mediated effector functions, IL‐33−/− mice are compromised in their ability to clear N. brasiliensis (Hung et al, 2013). While Dusp5 is also expressed in NK cells, these cells do not play a requisite role in parasite clearance (Supplementary Fig S4D). However, since IL‐33 has broad effects on many cell types beyond eosinophils, the greater ability of Dusp5−/− Rag2−/− mice to clear N. brasiliensis may still have additional contributing cellular components.
Given the association of SNPs in IL33 and its cognate receptor IL1RL1 in human asthma patients, our findings have interesting implications for human diseases. Dusp5 mRNA is also expressed in human eosinophils and its levels are also increased by IL‐33 (Supplementary Fig S13A and B), though there may be differences in Dusp5 regulation with IL‐5 (Temple et al, 2001). Investigation of Dusp5 regulation following treatment of asthma patients with anti‐IL‐5, anti‐IL‐13, or anti‐IL‐33R antibodies should provide additional insights.
In summary, our studies reveal a previously unknown role for DUSP5 in the regulation of IL‐33‐mediated eosinophil survival and function. While DUSPs have been described to control cell survival through regulation of pro‐apoptotic pathways, our studies define a novel mechanism by which DUSP5 regulates IL‐33‐mediated cell survival through the pro‐survival BCL‐XL axis. Given the established and emerging roles of eosinophils and IL‐33 in resolution of inflammation and mucosal homeostasis, respectively (Isobe et al, 2012; Lopetuso et al, 2012), defining how dysregulation of IL‐33 or DUSP5‐mediated pathways may provide insights into host maintenance and disease pathogenesis.
Materials and Methods
Generation of Dusp5−/− mice
The construct for targeting the C57BL/6 DUSP5 locus in C57BL/6 ES C2 cells was made using a combination of recombineering and standard molecular cloning techniques. C57BL/6 ES cells were transfected by electroporation with a linearized DUSP5 targeting vector and selected in media containing G418 (200 ng/ml). The targeted ES cells were transfected with a Cre‐containing plasmid TNLOX1‐3 to remove the neomycin resistance cassette. ES cells were injected into blastocysts using standard techniques, and germline transmission was obtained after crossing resulting chimaeras with C57BL/6 females. Genotyping with primer sets 5′‐(WT) CAGCTGCAGAATCTGCAAGGGTGG, 5′‐(KO) AAGCTATGCTGGTGCAGCCAGTCC, 3′‐TCATTGGTGTTGCTTCTGGGGAGG was used to confirm the generation of Dusp5−/− mice.
B6.SJL‐Ptprca Pepcb/BoyJ mice were purchased from Jackson Laboratories. Rag2−/− were purchased from Jackson Laboratories for breeding with Dusp5+/+ and Dusp5−/− mice. All mice were maintained under specific pathogen‐free conditions. All animal experimentation protocols were approved by the Laboratory Animal Resources Committee at Genentech, Inc. (S San Francisco, CA).
Bone marrow chimeras
Eight‐week‐old (CD45.1+) Dusp5+/+ (B6.SJL‐Ptprca Pepcb/BoyJ) mice were lethally irradiated (2 doses of 500 cGy with 137Cs γ‐irradiator) at 3‐h intervals ~3 h prior to injection. BM from Dusp5+/+ (CD45.1+) were isolated and mixed at a 1:1 ratio with either Dusp5+/+ or Dusp5−/− (CD45.2+) BM in Dulbecco's modification of Eagle's medium. Irradiated mice (CD45.1+) were injected in the lateral tail vein with approximately 1 × 106 total BM cells. All irradiated mice were given antibiotic‐supplemented water (1.1 g/l neomycin, 2 g/l glucose, and 110 mg/l polymyxin B, Sigma‐Aldrich) for 2 weeks.
Nippostrongylus brasiliensis infection
Dusp5+/+ or Dusp5−/− mice were placed under anesthesia and infected subcutaneously on the flank with 500 N. brasiliensis L3 larvae in 200 μl of saline. Control and infected animals were placed on polymyxin B and neomycin‐supplemented water for 5 days post‐infection. Blood, spleen, BM, BALF, and serum were harvested 8 days (Rag2+/+ mice) or 13 days (Rag2−/− mice) after infection. Tissues were harvested for FACS analysis, and worm burden was assessed on day 8–9 (Rag2+/+ mice and BM chimeric mice) or day 13–14 (Rag2−/− mice). Worm burden was determined by counting the numbers of adult worms under a dissection microscope. In vivo BrdU labeling was performed by injecting 1 mg of BrdU (BD Biosciences) into the peritoneum. For 36‐h pulse/chase experiments, mice were injected on day 10 post‐infection three times at 0, 12, and 24 h. Mice were sacrificed, and spleen, blood, and BM were harvested 36 h after initial pulse (12‐h chase). For 84‐h pulse/chase experiments, mice were injected on day 10 post‐infection three times at 0, 12, and 24 h. Spleen, blood, and BM were harvested 84 h after initial pulse (60‐h chase) for analysis. BrdU incorporation was analyzed by FACS.
Single‐cell suspensions from spleen, blood, BM, or BALF were washed in phosphate‐buffered saline (PBS) containing 5 mM ethylenediaminetetraacetic acid and 0.5% bovine serum albumin (BSA). Cells were incubated for 15 min with Fc receptor block (clone 2.4G2, BD Pharmingen) prior to antibody staining. The following antibodies (Ab) were used for FACS staining and sorting: anti‐CD3 (145‐2C11), CD4 (GK1.5), and Ly6c (HK1.4) Abs were purchased from EBioscience; anti‐CD8 (53–6.7), CD43 (S7), B220 (RA3‐6B2), NK1.1 (PK136), CD11b (M1/70), SiglecF (E50‐2440), CD69 (H1.2f3), and Ly6g (1A8) Abs were purchased from BD Pharmingen; anti‐F4/80 (BM8) Ab was purchased from BioLegend. Aqua Fluorescent reactive dye (Life Technologies) was used to discriminate viable from dead cells. Intracellular BrdU was analyzed using the FITC BrdU Flow Kit (BD Biosciences) per manufacturer's protocol. Mitochondrial membrane depolarization was analyzed with the JC‐1 dye (Life Technologies) per manufacturer's protocol. Apoptosis was assessed using the Apoptosis Kit – Pacific Blue annexin V/Sytox AADvanced (Invitrogen/Molecular Probes). Cells were sorted for RNA analysis on a BD FACSAria and stained cells visualized on a BD LSR Fortessa (BD Pharmingen). Flow cytometry data were analyzed using FlowJo software (Treestar).
In vivo depletion of NK cells during Nippostrongylus brasiliensis infection
Anti‐Asialo GM1 antibody was reconstituted in 1 ml water per manufacturer's recommendations (Wako chemicals). 24 h prior to N. brasiliensis infection, mice were administered 20 μl antibody or 200 μg isotype control antibody by intraperitoneal injection followed by intraperitoneal injections every 3 days. Worm burden and extent of NK cell depletion was analyzed on day 13 post‐infection.
In vivo administration of IL‐33
500 ng of recombinant murine IL‐33 (Genentech) was injected into the peritoneum of Dusp5+/+ and Dusp5−/− mice for six consecutive days. Peritoneal lavage fluid was harvested and analyzed by FACS or sorted for eosinophils for gene expression analysis on day 7.
Western blot analysis
BM‐derived eosinophils (5 × 106) were stimulated for indicated times and washed with PBS on ice. For whole‐cell lysates, cells were lysed in RIPA lysis buffer (Thermo Scientific) with protease and phosphatase inhibitors. Cells were fractionated using the Pierce cellular fractionation kit according to manufacturer's protocol. Total protein from each fraction was determined by BSA quantification method (Pierce). Nuclear and cytoplasmic fractionation purity was assessed by immunoblotting for HDAC2 (nuclear protein) and HSP90 (cytoplasmic protein) (Supplementary Fig S12). Cell lysates were loaded on a NuPAGE 4–12% Bis‐Tris gel (Life Technologies) and transferred to PVDF membrane using an iBlot Gel transfer device (Invitrogen). Membranes were blocked with 5% milk, 2% Tween in PBS. Abs specific for actin, BCL‐XL, BCL‐2, MCL‐1, BAX, BIM, BAD, PUMA, BAK, BID, caspase‐3, cytochrome c, ERK, p‐ERK, p38, p‐p38, JNK, p‐JNK, HSP90, HDAC2, p65, and p‐p65 were purchased from Cell Signaling. Anti‐DUSP5 monoclonal Ab (ab53217) was purchased from Abcam. All Abs were diluted in blocking buffer and incubated overnight at 4°C. Secondary Abs included HRP (Millipore)‐conjugated goat anti‐rabbit or goat anti‐mouse Abs in conjunction with Amersham ECL prime Western blotting detection reagent (GE Healthcare Life Sciences) to visualize bands using X‐ray film (Kodak). Bands were measured by densitometry analysis using ImageJ analysis software.
Cytochrome c ELISA
BM‐derived eosinophils were cytokine‐starved or stimulated with the indicated cytokines (25 ng/ml) for 3 days. Cells were washed, and cytoplasmic fraction was isolated (see Western blot method). 20 μl of total cytoplasmic protein was diluted and cyt c levels were analyzed with a cyt c ELISA kit (Invitrogen) per manufacturer's instructions. Total protein content was determined by BSA method, and cyt c levels were normalized to protein levels to derive a relative ratio.
Eosinophils were cultured from total BM as previously described (Dyer et al, 2008). Briefly, total BM cells were flushed from femurs from Dusp5+/+ and Dusp5−/− mice. Red blood cells (RBCs) were removed by lysis using ACK Lysing buffer (Gibco). PBS washed cells were cultured at 1 × 107 cells/ml in RPMI 1640 media supplemented with 20% fetal bovine serum (Hyclone), Pen/Strep, Glutamax, β‐mercaptoethanol, and non‐essential amino acids (Gibco). Cells were stimulated for 4 days in the presence of rmSCF (R&D) and rmFlt3L (R&D) at 100 ng/ml final concentration. After 4 days in culture, cells were washed and resuspended in complete media supplemented with rmIL‐5 (Peprotech) at a concentration of 10 ng/ml for 4 days. On culture day 8, complete media and IL‐5 were replenished and cultured until day 11 when cells were > 95% SiglecF+ and SSChi by FACS analysis. For long‐term cultures, day 11–13 BM‐derived eosinophils were washed and stimulated with 25–50 ng/ml of IL‐5 or IL‐33 for 1 day unless specified otherwise and total cell count determined by Guava Viacount Assay on a Guava easyCyte instrument (Millipore). Data are represented in total cell count or relative viability where percent viability of cultures was normalized to control‐treated cultures.
Real‐time quantitative PCR
Total RNA was extracted from cells with RNeasy Mini Plus Kits (Qiagen) and reverse‐transcribed with SuperScript III First‐Strand synthesis system (Life Technologies), in accordance with the manufacturer's instructions. Primers used to evaluate Dusp5 expression in activated CD4+ T cells were (forward) 5′‐GTCACCACTTGCGGGAGTAT‐3′ and (reverse) 5′‐GGAGCAGTCACAGGAAATGAC‐3′. Taqman probe sets for human DUSP5 and GAPDH or murine Gapdh, Fosl1, Il4, Il13, Tnfa and Dusp5 were mixed with Taqman Universal PCR mastermix (Applied Biosystems), and gene expression was measured on a 7900HT Fast Real‐Time PCR System (Applied Biosystems). Sybr Green analysis of gene expression with Quantitect Sybr Green PCR Kit (Qiagen) was used to analyze Bclxl normalized to Actinb. Primers use for Sybr Green analysis were Bcl‐xl forward primer, 5′‐TGG AGT CAG TTT ACT GAT GTC GAA G‐3′, reverse primer, 5′‐AGT TTA CTC CAT CCC GAA AGA GTT C‐3′; Actinb forward primer 5′‐TAT TGG CAA CGA GCG GTT C‐3′ and reverse primer 5′‐CAA TAC CCA AGA AGG AAG GCT‐3′. Data were analyzed using the standard 2ΔCt method, and target genes normalized to housekeeping genes GAPDH or Actinb.
Human cell culture
Buffy coats were obtained through Genentech Health Services from human donors. 50 ml of blood was diluted 1:1 in PBS, layered onto Ficoll, and centrifuged at 800 × g at room temperature. Leukocytes were harvested, and B cells, CD4 and CD8 T cells, NK cells, and monocytes were isolated using their respective Miltenyi human isolation kits. RBCs were removed using with Ammonium‐Chloride‐Potassium lysis buffer (Gibco). Eosinophils were purified using a Miltenyi eosinophil isolation kit per manufacturer's recommendations. Isolated eosinophils were stimulated with 100 ng/ml hIL‐33 (Genentech) in RPMI medium (Gibco).
Microarray gene expression profiling
Total RNA was extracted from cells using RNeasy Mini Kits (Qiagen) according to the manufacturer's protocol. RNA samples were quantitated with a NanoDrop ND‐1000 UV spectrophotometer (Thermo Scientific, West Palm Beach, FL), and RNA quality was assessed with an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). The quantity of total RNA used in a two‐round amplification protocol ranged from 10 ng to 50 ng per sample. First‐round amplification and second‐round cDNA syntheses were done with the Message Amp II mRNA Amplification Kit (Applied Biosystems, Foster City, CA). Cyanine‐5 dye was incorporated with the Quick Amp Labeling kit (Agilent Technologies). Each cyanine‐5‐labeled test sample (750 ng) was pooled with cyanine‐3‐labeled Universal Mouse Reference RNA (Stratagene, La Jolla, CA) and hybridized onto Affymetrix Mouse genome 430 2.0 (MOE430V2) arrays as described in the manufacturer's protocol. Data are deposited in the GEO database under accession number GSE62999 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE62999). Statistical calculations for analysis of gene expression microarray data were performed with the R Project software package, version 2.15.2.
DAH designed experiments, generated and interpreted data, and prepared the manuscript. ACC designed experiments, interpreted data, and prepared the manuscript. DY and MX performed the helminth infection, worm burden analysis and assisted in mouse necropsy. J‐HY designed and generated Dusp5−/− mice.
Conflict of interest
The authors are employees of Genentech, Inc.
Supplementary Figure S1
Supplementary Figure S2
Supplementary Figure S3
Supplementary Figure S4
Supplementary Figure S5
Supplementary Figure S6
Supplementary Figure S7
Supplementary Figure S8
Supplementary Figure S9
Supplementary Figure S10
Supplementary Figure S11
Supplementary Figure S12
Supplementary Figure S13
Supplementary Table S1
Supplementary Table S2
Source Data for Supplementary Figure S1
Source Data for Supplementary Figure S9
Source Data for Supplementary Figure S11
Source Data for Supplementary Figure S12
Legends for Supplementary Figures, Tables
The authors thank Tim Behrens, Eric Brown, Nico Ghilardi, Rajita Pappu, John Monroe, and Menno van Lookeren Campagne for advice and critical review of this manuscript.
- © 2014 Genentech Inc.