Eosinophil differentiation in the bone marrow is promoted by protein tyrosine phosphatase SHP2

SHP2 participates in multiple signaling events by mediating T-cell development and function, and regulates cytokine-dependent granulopoiesis. To explore whether and how SHP2 can regulate bone-marrow eosinophil differentiation, we investigate the contribution of SHP2 in the bone-marrow eosinophil development in allergic mice. Blockade of SHP2 function by SHP2 inhibitor PHPS-1 or conditional shp2 knockdown by adenovirus-inhibited bone-marrow-derived eosinophil differentiation in vitro, with no detectable effects on the apoptosis of eosinophils. Furthermore, SHP2 induced eosinophil differentiation via regulation of the extracellular signal-regulated kinase pathway. Myeloid shp2 conditional knockout mice (LysMcreshp2flox/flox) failed to induce eosinophilia as well as airway hyper-responsiveness. The SHP2 inhibitor PHPS-1 also alleviated eosinophilic airway inflammation and airway hyper-responsiveness, accompanied by significantly reduced levels of systemic eosinophils and eosinophil lineage-committed progenitors in allergic mice. We demonstrate that inhibition of eosinophil development is SHP2-dependent and SHP2 is sufficient to promote eosinophil formation in vivo. Our data reveal SHP2 as a critical regulator of eosinophil differentiation, and inhibition of SHP2 specifically in myeloid cells alleviates allergic airway inflammation.

The protein tyrosine phosphatase SHP2 is a ubiquitously expressed intracellular enzyme that contains two Src homology 2 domains and one catalytic protein tyrosine phosphatase domain. [10][11][12] SHP2 integrates multiple signaling events and mediates a variety of physiological functions. [13][14][15] Studies have shown that SHP2 participates in multiple signaling events by mediating T-cell development and function, and stimulating CEBPA gene expression to regulate cytokinedependent granulopoiesis. 16,17 Normal SHP2 function is critical for the initial step of embryonic stem (ES) cell differentiation to mesoderm and to hemangioblasts. It acts within the LIF-gp130-Stat3 pathway to keep a proper balance of ES cell differentiation, pluripotency and apoptosis, thereby maintaining a functional hematopoietic stem cell/progenitor pool. 13,[18][19][20] Pazdrak et al. [21][22][23] demonstrated that the physical association of SHP2 with the phosphorylated α common chain of the IL-5 receptor (IL-5αcR) and Grb2, and its early activation, are required for coupling of the receptor to the Ras-Raf-MAP/Erk2 pathway and for the prevention of eosinophil death by IL-5, and shp2 may also act as both a positive effector to downstream GM-CSF-and ICAM-1-dependent ERK1/2 activation in human eosinophils.
In a previous study, we have shown that specific deletion of shp2 expression in mouse airway epithelia reduces TGF-β1 production and attenuates allergic airway remodeling. 24 However, as SHP2 is expressed ubiquitously, it is of interest to know its possible roles in other cells or tissues during the development of asthma. As eosinophils are crucial for the pathogenesis of asthma, it is important to understand how eosinophil differentiation is regulated. Therefore, we investigated whether and how SHP2 affects eosinophil development. By using the SHP2 inhibitor PHPS-1 and specifically deleting shp2 in myeloid cells, we found that inactivation or loss of SHP2 in these cells decreased the level of eosinophil recruitment to the airway, resulting in alleviation of lung inflammation and reduction of airway hyper-responsiveness (AHR), which were most likely through a direct inhibition of eosinophil differentiation. These results suggest that SHP2 may be a key regulator of eosinophil differentiation and thus can serve as a potential therapeutic target for the treatment of asthma.

Results
SHP2 is required for eosinophil differentiation in vitro without influence on the apoptosis of eosinophils. To begin to test the function of SHP2 in eosinophil differentiation, we first analyzed the effect of phenylhydrazonopyrazolone sulfonate, PHPS-1, 25 as a cell-permeable compound, which is highly specific for SHP2 over the closely related tyrosine phosphatases Shp1 and PTP1B, on the outgrowth of eosinophils from purified bone-marrow cells. Non-adherent mononuclear cells (NAMNCs) were first cultured for 4 days with recombinant mouse FLT3 ligand (rmFlt3-L; 100 ng/ml) and recombinant mouse stem cell factor (rmSCF; 100 ng/ml), and then cultured for 6 days with rmIL-5 (10 ng/ml) for eosinophil differentiation (Figure 1a; detailed in Materials and Methods). This induced the development of eosinophils that contained eosinophilic granules and a characteristic donutshaped nucleus, as observed by Wright-Giemsa staining ( Figure 1b). Flow cytometric analysis confirmed the development of eosinophils, as these cells were SSC hi SiglecF + (Figure 1c). PHPS-1 (20 μM) administration together with IL-5 dramatically reduced the production of eosinophils (Figures 1d-h). However, there was no change in the production of eosinophils when PHPS-1 was given in the first 4 days before IL-5 was given ( Supplementary Figures  1a-c), suggesting that the decreased eosinophil generation was not due to alterations in the numbers of very early progenitors (likely GMPs, before IL-5 stimulation to EoPs). We further detected the effect of PHPS-1 on the cell apoptosis during bone-marrow eosinophil (bmEo) differentiation in vitro. Analysis of the percentage and total SiglecF + Annexin V − cells demonstrated that there was markedly increased percentage and total viable eosinophils in the control group in comparison with PHPS-1-treated group  Figure 1h). Thus, the decreased number of bmEos by PHPS-1 may partly due to the increased apoptosis of EoPs. These data together suggest that PHPS-1 inhibits the differentiation of eosinophils without affecting their survival, while increasing the apoptosis of EoPs.
We next harvested the bone-marrow NAMNCs of Shp2 flox/flox mice, and infected them with Ad-Cre-GFP to induce shp2 knockdown in vitro (Figure 1n). Western blot analysis confirmed that the level of the SHP2 protein was indeed significantly reduced in these cells (Figures 1o and p). Consistent with the effect of PHPS-1, bone-marrow NAMNCs in which shp2 was deleted showed remarkably reduced eosinophil percentages compared with controls (Figures 1q and r).
SHP2 is required for IL-5-induced colony formation. To assess the effect of SHP2 to IL-5-induced differentiation of eosinophils, we subjected cells from wild-type (WT) and allergic mice to colony-forming unit (CFU) assay using IL-5 with the treatment of PHPS-1 or not in vitro. The PHPS-1 group displayed a specific decrease in CFUs induced by IL-5 ( Figure 2a). Furthermore, the number of eosinophil CFU (Eos-CFU) was remarkably reduced when NAMNCs were  (Figure 2a), in which IL-5Rα was highly expressed. Consistent with the effect of PHPS-1, bone-marrow NAMNCs in which shp2 was deleted showed remarkably reduced numbers of Eos-CFU compared with controls ( Figure 2b). Furthermore, in response to IL-5, the Eos-CFUs were smaller in NAMNCs treated with Ad-Cre-GFP in comparison with the Ad-GFP group (Figure 2c). These data further suggest that SHP2 indeed has a key role in the development of eosinophils.
Genetic knockdown of shp2 decreases eosinophil percentage in the bone marrow. To further investigate the in vivo and in vitro function of SHP2 in the bone marrow, we generated LysM cre Shp2 flox/flox mice in which Cre expression was induced, and then inactivated the shp2 gene in myeloid cells (Figure 3a). Analysis of the genomic DNA from tails indicated expression of the floxed shp2 and the Cre genes ( Figure 3b). We first detected the base level of eosinophils in the bone marrow using flow cytometry and found that they were remarkably decreased in the LysM cre Shp2 flox/flox mice (Figures 3c and d), whereas they yielded normal numbers of macrophages (gated as SiglecF − F4/80 + ; Figures 3e and f) and neutrophils (gated as Gr-1 int CD11b int , Gr-1 + CD11b lo and Gr-1 + CD11b + , likely representing pro/mye and immature and mature neutrophils, respectively; Figures 3g and h). Furthermore, eosinophils were dramatically decreased in the bone marrow of LysM cre Shp2 flox/flox mice during the eosinophil differentiation in vitro (Figures 3i-k), whereas the decreased level of eosinophils was not due to alterations in the number of eosinophil progenitors (defined as Lin − Sca-1 − CD34 + c-Kit lo IL-5Rα + ; Supplementary Figure 1i). In addition, lin − cells were sorted and cultured with G-CSF and PHPS-1 in vitro to find whether SHP2 is required in neutrophil development. Although the percentage of mature neutrophils (defined as Gr-1 + CD11b + ) was decreased and that of immature neutrophils (defined as Gr-1 int CD11b + ) was increased in the PHPS-1 group, the total numbers of mature and immature neutrophils were decreased significantly ( Supplementary Figures 1j-l), which suggested that PHPS-1 inhibited the production of neutrophils. Collectively, these data indicate that SHP2 is required for eosinophil differentiation and can also affect on the development of neutrophils.
SHP2 regulates IL-5-induced eosinophil differentiation via p-Erk activation. To explore the possible mechanisms of eosinophilopoiesis, we first measured the dynamic change of SHP2 and found that it was induced in a time-dependent manner during eosinophil differentiation (Figure 4a). Realtime PCR analysis also revealed that the IL-5-induced mRNA levels of Gata-1 and Mbp were markedly reduced by PHPS-1 (Figures 4b and c). We also measured the level of Gata-1 and Mbp mRNA on day 10 of eosinophil differentiation of Shp2 flox/flox mice, and found that both were significantly reduced when NAMNCs were transfected with Ad-Cre-GFP (Figures 4d and e). We next examined the downstream signaling events of SHP2 in the regulation of eosinophil differentiation and found that SHP2 protein levels were markedly decreased in the bmEos of LysM cre Shp2 flox/flox mice compared with Shp2 flox/flox mice ( Figure 4f).   Figures 2g and h), in which most were eosinophil progenitor cells, the decreased number of bmEos was partly due to the apoptosis of SiglecF − cells.
Total AnnV + cells were also increased in the U0126treated group ( Supplementary Figures 2i and j). These data together suggest that U0126 inhibits the differentiation of eosinophils without affecting their survival while increasing the apoptosis of eosinophil progenitors. Moreover, the deficiency of SHP2 results in reduced activation of the Erk pathway, which then inhibits eosinophil differentiation in the bone marrow.   (Figures 5b and m). Moreover, AHR in LysM cre Shp2 flox/flox allergic mice was significantly reduced compared with OVAtreated Shp2 flox/flox controls (Figure 5c). Furthermore, myeloid cell-specific shp2 knockdown resulted in a remarkable reduction in the mRNA levels of IL-4, IL-5 and IL-13, whereas there was no appreciable difference in the IFN-γ mRNA level (Figures 5d-g). There was no obvious change in the level of serum eotaxin between OVA-treated LysM cre Shp2 flox/flox and Shp2 flox/flox mice (Supplementary Figure 2k). The decreased pro-inflammatory cell infiltrates in mice were further demonstrated by a corresponding alleviation in the histopathology observed in these mice as reflected quantitatively by the inflammatory and PAS scores (Figures 5h-k). These results reveal that myeloid cell-specific shp2 knockdown in an acute allergic model inhibits airway inflammation and reduces airway responsiveness, and the mechanism is probably through inhibition of the production of eosinophils.
PHPS-1 alleviates airway inflammation and decreases AHR in allergic mice. As myeloid shp2 knockdown alleviates airway inflammation, we suggest that SHP2 may be a previously underappreciated therapeutic target for intervention. As such, PHPS-1 was used to explore whether it can alter allergen-induced pulmonary inflammation and become the potential therapeutic drug. We sensitized and challenged WT mice with chicken egg ovalbumin with PHPS-1 or not (Figure 6a). We quantified total and differential cell counts from BALF. As expected, PHPS-1 led to a significant reduction in total cell number, and, specifically, there was a remarkable decrease in eosinophil number in OVA/PHPS-1 mice compared with the OVA/PBS group (Figure 6b). PHPS-1 also decreased AHR in allergic mice, reducing the levels to those in the PBS-treated control group (Figure 6c). AHR to methacholine (Mch) was similar in the Control/ PHPS-1 and Control/PBS groups. In addition, we assessed the effects of PHPS-1 on the mRNA levels of IL-4, IL-5, IL-13 and IFN-γ in the lung. Pharmacological inhibition of SHP2 resulted in a marked reduction of the mRNA levels of IL-4, IL-5 and IL-13; however, no detectable change was found in the IFN-γ mRNA level (Figures 6d-g). The protein levels of IL-4 and IL-13 in lung tissue were consistent with the mRNA levels (Figures 6h and i). The protein level of eotaxin was also reduced in the serum of OVA/PHPS-1 mice (Supplementary Figure 2l). Lung histopathology showed considerably fewer inflammatory cells in the lungs of OVA/PHPS-1 mice than OVA/PBS mice (Figures 6j and k). Periodic acid-Schiff (PAS) staining demonstrated less mucus production and fewer mucus-producing cells in the bronchioles and lungs of OVA/ PHPS-1 mice than in OVA/PBS mice (Figures 6l and m). These findings reveal a potential role of PHPS-1 in protection against allergic airway inflammation.
PHPS-1 inhibits the production of eosinophils and EoPs in allergic mice. To determine whether PHPS-1 inhibits the production of eosinophils and then eases lung inflammation, we measured the numbers of eosinophils in both blood and bone marrow. The numbers of eosinophils in blood and bone marrow decreased when allergic mice were treated with PHPS-1 (Figures 7a and b). We also analyzed the protein level of IL-5 in blood, and interestingly found no significant difference between OVA/PHPS-1 and OVA/PBS mice (Figure 7c). The apoptosis of eosinophils in the bone marrow was also similar in these two groups (Figures 7d and e).
These data indicate that PHPS-1 inhibits the formation of eosinophils independent of IL-5, without influencing the apoptosis of eosinophils. Next, we evaluated the EoPs in the bone marrow and found a dramatic decrease in their number in allergic mice treated with PHPS-1 (Figures 7f and  g), although there was no alteration in IL-5Rα surface expression in eosinophil progenitors (Supplementary Figure 2m). It indicates that the decreased eosinophil numbers was possibly because of the decreased numbers of eosinophil progenitors. An ex vivo Eos-CFU assay further confirmed this observation, as the number of Eos-CFU from bone-marrow NAMNCs was considerably reduced in OVA/ PHPS-1 mice relative to the OVA/PBS group (Figure 7h). This was supported with quantitative real-time PCR analysis of the key transcription factors in NAMNCs, as the mRNA level of Gata-1 was inhibited in OVA/PHPS-1 mice compared with OVA/PBS mice (Figure 7i). However, the absolute number of GMPs (Lineage − c-Kit + CD16/32 hi CD34 + ) was not significantly different in OVA/PHPS-1 and OVA/PBS mice (Figures 7j and k). These results suggest that PHPS-1 inhibits the differentiation of EoPs from GMPs. These data together indicated that PHPS-1 reduces the number of EoPs and represses the expression of transcription factor Gata-1 for eosinophil development.

Discussion
Allergic asthma is characterized by the infiltration of eosinophils into the airway and lung tissues with progressive tissue damage. 6 Eosinophils developed from the bone marrow and migrate into lung tissue after exposure to pulmonary allergens, [27][28][29] and thus the blockade of eosinophil production should be an efficacious approach to the treatment or prevention of asthma at the source. However, to the best of our knowledge, fewer efforts have targeted the inhibition of eosinophil differentiation for the prevention of allergic airway diseases. Recently, studies have already shown that Shp2 has an important role in lung diseases, 30,31 especially in chronic asthma. 24 In this study, we clearly demonstrated that SHP2 is critically involved in eosinophil differentiation. Genetic deletion of shp2 in myeloid cells effectively reduced the number of eosinophils in the bone marrow and eventually protected mice from OVA-induced allergic airway inflammation. Pharmacological inhibition of SHP2 with PHPS-1 also markedly attenuated eosinophil differentiation in vitro, and decreased the numbers of eosinophils and EoPs in allergic mice. More interestingly, we also clearly demonstrated that the inhibition of SHP2 had little effect on eosinophil apoptosis. Altogether, these data strongly suggest that blockade of eosinophil differentiation, such as through SHP2 inhibition, could be a potential therapy for asthma and other consequences of eosinophilic inflammation. The phenylhydrazonopyrazolone sulfonate PHPS-1 is a potent and cell-permeable inhibitor, which is specific for SHP2 and inhibits SHP2-dependent cellular events and downstream signaling. 25 Recent studies 30 further suggest that PHPS-1 inhibited the activation of Erk1/2 by CS and PHPS-1 was given 10 μM in vitro and 3 mg/kg in vivo 30 min before CS exposure. We established OVA-induced allergic model with PHPS-1 according to their method with minor modification. In our work, although PHPS-1 effectively attenuated eosinophil differentiation in vitro, its overall protective role in vivo cannot rule out other possible mechanisms such as the inhibition of functions in the epithelium, neutrophils and T cells. We have recently shown that SHP2 expression is induced locally in the airway during asthma development, and genetic knockdown of shp2 in airway epithelia significantly attenuates OVAinduced airway remodeling and lung dysfunction, 24 suggesting an important role of SHP2 in the regulation of local airway (h and i) Lung IL-4 and IL-13 protein levels measured using ELISA. (j-m) Effects of PHPS-1 on airway inflammation (H&E staining) and goblet cell metaplasia (PAS staining; × 400 magnification; scale bar = 10 μm). Total lung inflammation and mucus levels were defined as the average of the peribronchial inflammation scores or PAS scores. Results are expressed as means ± S.E.M. of two independent experiments (n = 8 per group). n.s, not significant; *Po0.05, **Po0.01, OVA-exposed control group versus OVA-exposed PHPS-1 group damage. Thus, the protection by PHPS-1 against allergic airway inflammation may also be due in part to its possible function in the prevention of airway damage.
Previous studies have shown that SHP2 is a positive regulator of growth-factor signaling. 17,20 Similarly, we also found that the SHP2 protein level was increased significantly, and this in turn positively regulated eosinophil development, depending on Erk activity. Adachi et al. 32   Shown are two to three experiments with six to eight mice per group per experiment. n.s, not significant. *Po0.05, **Po0.01, OVA-exposed control group versus OVA-exposed PHPS-1 group transcription 3 (Stat3) necessary to maintain the proper balance of stem cell differentiation, self-renewal and apoptosis. Kano et al. 36 showed that Erk activation is critical in delivering pro-apoptotic signals via Siglec-8 in IL-5-activated eosinophils. In our data, the phosphatase activity of Erk1/2 was remarkably reduced when shp2 was conditionally deleted in LysM cre Shp2 flox/flox mice during bmEo differentiation. Therefore, we assumed that SHP2 regulates eosinophil differentiation at least partially through the Erk1/2 MAPK pathway. To prove this hypothesis, a pharmacological inhibitor of Erk1/2, U0126, was used in the experiment. We found U0126 also significantly reduced the eosinophil differentiation. Thus, SHP2 has a positive role in IL-5-induced activation of the Ras-Erk kinase pathway and leads to bmEo differentiation in the bone marrow.
It might be noteworthy that PHPS-1 markedly decreased the OVA-induced EoPs while having no appreciable effects on GMPs. These data suggested that SHP2 is probably not involved in OVA-induced GMP production. Moreover, the in vitro study also showed that there was no change in eosinophils when PHPS-1 was treated before IL-5 was given, suggesting that the decreased eosinophil was not due to alterations in the numbers of GMPs. However, whether SHP2 directly regulates the development of EoPs from GMPs is not clear. If SHP2 is involved in EoP production from GMPs, PHPS-1 could have a direct regulatory effect on this process. Our in vitro study proved that the apoptosis of SiglecF − cells were increased, in which most cells were likely EoPs. These data indicated that PHPS-1 increases the apoptosis of EoPs. However, SHP2 knockdown mediated by the M-lysozyme gene (LysM) should not change the SHP2 levels during the development of EoPs from GMPs. The LysM gene is strongly expressed in mature macrophages and myeloid cells, and is a marker of myeloid differentiation. It is progressively turned on during differentiation from myeloid precursor cells to mature eosinophils, in which it is fully active. 37,38 In such a case, the decreased eosinophils in the OVA-challenged LysM cre Shp2 flox/flox mice might be due to the decreased differentiation of eosinophils from EoPs in the bone marrow. Furthermore, PHPS-1 has no effect on serum IL-5 production in allergic mice, which may indicate that PHPS-1 can directly inhibit eosinophil development both in vitro and in vivo.
As the LysM gene is expressed in mature macrophages and myeloid cells, the next question is whether myeloid SHP2 knockdown alters the development of neutrophils or macrophages. We found that neutrophils were significantly decreased in the BALF of LysM cre Shp2 flox/flox allergic mice and OVA-challenged mice treated with PHPS-1, although there was no considerable difference in the base level of bonemarrow neutrophils. The in vitro neutrophil number also decreased when lineage-negative (Lin − ) mouse marrow cells were cultured with G-CSF and PHPS-1. Jack et al. 39 found that G-CSF activate STAT3 and SHP2, and potentially shifts the balance to granulopoiesis via the effects of SHP2 on regulators. These data suggest that SHP2 also has a critical role in neutrophil development. However, we found that monocytes were similar in the BALF of allergic mice either with myeloid SHP2 knockdown or with PHPS-1 treatment. Jack et al. 39 have also shown that M-CSF activates ERK independently of SHP2 in monopoiesis. Thus, SHP2 may have no effects on the development of monocytes or macrophages.
The detailed mechanisms by which SHP2 regulates eosinophil differentiation remain unclear. Several lines of evidence have demonstrated unequivocally that GATA-1 has an integral role in the differentiation of myeloid progenitors to the eosinophil lineage and the regulation of eosinophil-specific genes. 40 − 42 Disruption of a high-affinity double-palindromic GATA site in the mouse GATA-1 promoter results in the selective loss of the eosinophil lineage in vivo. 37,38 We showed that inhibition of SHP2 activity in the WT NAMNCs by PHPS-1, or reduction in SHP2 levels in NAMNCs from Shp2 flox/flox mice by Ade-Cre effectively decreased the expression of Gata-1. These data suggested the possibility that SHP2 regulates eosinophil differentiation through downregulation of Gata-1 expression. However, these data do not rule out the possibility that SHP2 regulates eosinophil differentiation through other pathways that eventually results in decreased numbers of eosinophils, and the decreased Gata-1 expression is just a marker of eosinophil downregulation in this case. Nonetheless, the detailed molecular pathways that mediate SHP2 function in eosinophil differentiation warrant further study.
In summary, we have demonstrated that SHP2 is a potential positive regulator of eosinophil differentiation, and inhibition of SHP2 specifically in myeloid cells eventually alleviates eosinophilic airway inflammation. Our study also represents an initial effort to demonstrate that targeting eosinophil differentiation could be an effective therapeutic approach to asthma.

Materials and Methods
Mice. All animal procedures conformed to the Guide for the Care and Use of Laboratory Animals and were approved by the Zhejiang University Medical Laboratory Animal Care and Use Committee. WT C57BL/6 mice were purchased from the Laboratory Animal Center of the Zhejiang University (Hangzhou, China). Shp2 flox/flox and LysM cre mice on the C57BL/6 background were a generous gift from Dr. Gen-Sheng Fen (University of California at San Diego, CA, USA). LysM cre Shp2 flox/flox mice and littermate controls (Shp2 flox/flox ) were used for the experiments. All protocols were approved by the Ethics Committee for Animal Studies at the Zhejiang University. The primers used for gene typing were as follows: shp2: forward, 5′-ACGTCATGATCCGCTGTCAG-3′; reverse, 5′-ATGGGAG GGACAGTGCAGTG-3′; Cre: common primer, 5′-CCCAGAAATGCCAGATTACG-3′; mutant primer, 5′-CTTGGGCTGCCAGAATTTCTC-3′.
Isolation and ex vivo culture of mouse bmEos. The method of ex vivo culture of bmEos was as described previously with slight modification. 41 Briefly, bone-marrow NAMNC cells were cultured at 10 6 /ml in Iscove's modified Dulbecco's medium (IMDM; Invitrogen, Waltham, MA, USA) with 20% FBS (Invitrogen), 100 IU/ ml penicillin and 10 mg/ml streptomycin, 2 mM glutamine, 25 mM HEPES, 1 × nonessential amino acids, 1 mM sodium pyruvate and 0.006‰ β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 100 ng/ml rmSCF (PeproTech, Rocky Hill, NJ, USA) and 100 ng/ml rmFLT3-L (PeproTech) from days 0 to 4. On day 4, the medium was replaced with medium containing 10 ng/ml rmIL-5 (R&D Systems, Minneapolis, MN, USA). On day 8, the cells were provided with fresh medium supplemented with rmIL-5. In most experiments, 20 μM PHPS-1 (Sigma-Aldrich) was treated on days 4 and 8 when the medium was replaced. While in some experiments 20 μM PHPS-1 (Sigma-Aldrich) was treated on day 0 and washed off on day 4 when the medium was replaced. U0126 (20 μM; Cell Signaling Technology, Danvers, MA, USA) was treated on days 4 and 8 when the medium was replaced. Cells were enumerated and harvested for analysis of the mRNA levels of Gata-1 and Mbp. Apoptosis was detected and cells were stained with SiglecF for anylysis of eosinophils. Cells were lysed and analyzed by western blotting with polyclonal ERK1/2, p-ERK1/2, SHP2 and β-actin. Antibodies against ERK and p-ERK were from Cell Signaling Technology, and antibodies against SHP2 and