CRKL but not CRKII Inhibits Erythropoiesis and Megakaryopoiesis of CML via Inactivating Raf/MEK/ERK/Elk-1 Pathway

Background: As members of the CT10 regulation of kinase (CRK) adaptor protein family, CRK-like (CRKL) and CRKII are involved in cell proliferation, survival, adhesion, migration and differentiation. However, the exact role and underlying mechanism of CRKL and CRKII in leukemic cell differentiation are still unknown. Methods: Quantitative real-time qPCR (qRT-PCR) was used to detect the expression levels of CRKL and CRKII in chronic myeloid leukemia (CML) patients and complete remission (CR) patients; Western blotting (WB) was used to measure the expression levels of CRKL and CRKII during hemin-induced erythroid differentiation of K562 cells; Benzidine staining, isobaric tags for relative and absolute quantitation (iTRAQ) proteomic analysis, cDNA microarray assay, qRT-PCR and WB were used to examine the effects of CRKL and CRKII deregulation on erythroid and megakaryocyte differentiation of K562 cells; PD98059 was used to investigate the underlying mechanism of CRKL in erythropoiesis and megakaryopoiesis. Results: CRKL was found to be overexpressed in chronic myeloid leukemia (CML) patients compared with normal samples, while its expression level was lower in CR patients than in corresponding CML patients. The CRKL expression level was signicantly decreased during the erythroid differentiation of K562 cells following hemin treatment. Moreover, CRKL downregulation promoted erythroid and megakaryocyte differentiation of K562 cells accompanied by increased expression level of the erythroid differentiation markers γ-globin, glycophorin (GPA) and the megakaryocyte differentiation markers CD41, CD61. Furthermore, gene microarray and iTRAQ quantitative proteomic analysis showed that CRKL downregulation increased hemoglobin (HB) molecules HBD, HBA1, HBA2, HBZ, HBE1, HBG1 and globin transcription factor 1 (GATA1), high-mobility group protein (HMGB2) expression levels. Mechanistically, CRKL inhibited erythroid and megakaryocyte differentiation of K562 cell via inactivating Raf/MEK/ERK/Elk-1 pathway. Conversely, CRKII was only slightly overexpressed in CML patients and had no effect on erythroid differentiation of K562 cells. Conclusions: Taken together, our results demonstrate that CRKL but not CRKII contributes to erythropoiesis and megakaryopoiesis of CML, providing novel insights into effective diagnosis and for CML patients.


Background
Hematopoiesis is a precisely modulated multi-step process including hematopoietic stem cell (HSC) selfrenewal and hematopoietic stem/progenitor cell differentiation [1,2]. Erythropoiesis and megakaryopoiesis are important parts of hematopoiesis [3,4]. Normal erythropoiesis produces about 10 11 new red blood cells (RBCs) every day in an adult human through the commitment of hematopoietic stem cells into erythroid progenitors, which subsequently differentiate into mature RBCs [5,6].
Megakaryopoiesis is responsible for blood platelets [7]. Destruction of erythropoiesis and megakaryopoiesis processes leads to various diseases, including thrombocytopenia, anemia and leukemia. Understanding the regulatory mechanisms of erythropoiesis and megakaryopoiesis can lead to characterizing novel modulators and developing new methods for treatment of blood related diseases.
Chronic myeloid leukemia (CML) is a clonal myeloproliferative pluripotent hematopoietic stem cell malignancy disorder characterized by the expression of BCR/ABL1 (B-cell receptor/v-abl Abelson murine leukemia viral oncogene) fusion gene [8,9], which is generated from the Philadelphia chromosome translocation of chromosome 9 to 22 [10,11]. BCR-ABL is the molecular hallmark of CML with tyrosine kinase activity that can potentially activate multiple signal transduction pathways, resulting in abnormal cell proliferation, apoptosis, migration, invasion and differentiation [12][13][14]. The delay of differentiation and maturation is considered to be a characteristic of leukemia, and inducing leukemia cell differentiation and breaking through the barrier of differentiation and maturation haa become a research hotspot in basic medical research and its clinical translation [15]. The K562 cell is human leukemia cell derived from the pleural effusion of a CML woman patient in terminal blast crisis [16,17]. K562 cells behave more like undifferentiated early pluripotent hematopoietic progenitors, and have been widely used as a model for studying hematological cell differentiation due to its ability to express speci c markers of granulocytic, monocytic, erythroid and megakaryocytic lineages [18,19].
The CT10 regulation of kinase (CRK) adapter protein family is involved in intracellular signal transduction. Members of the CRK family were the rst identi ed adaptor proteins, which connect with upstream molecules through their C-terminal SH2 domain and with downstream molecues through their N-terminal SH3 domain [20]. CRK consists of cellular homologs CRKI, CRKII and CRK-like (CRKL) which are ubiquitously expressed and conserved across eukaryotic organisms [21]. CRKI and CRKII were originally described as splice variants, while CRKL is encoded by another homologous gene. CRKI is composed of one SH2 domain and one SH3 domain, CRKII and CRKL are composed of one SH2 domain, one SH3N and one SH3C domains [21][22][23]. CRKII and CRKL are highly similar in sequence and both possess tyrosine phosphorylation sites that can be phosphorylated by BCR-ABL to activate signaling pathways [21,22], suggesting CRKII and CRKL have overlapping functions. CRKII and CRKL contains a variety of linkages for docking BCR-ABL, p130Cas, Dock180, GAB, ABL-1, Pax, GEF, C3G and SOS to form localized complexes critical for cell proliferation, survival, adhesion, migration and invasion [24][25][26]. CRKII and CRKL deregulation has been proved to be involved in the development and progression of a variety of cancers [27,28]. Nevertheless, a few studies report the association of CRK with differentiation: CRK could induce pheochromocytoma PC12 cell differentiation [29,30]; CRK and CRKL could synergistically increase RANKL-induced osteoclast differentiation [31]. However, the precise roles and underlying mechanisms of CRKL and CRK in leukemic cell differentiation are still not reported.
In our study, we investigated the effect of CRKL and CRK on erythroid and megakaryocyte differentiation using the K562 cell lines as a model system. Interestingly, we found CRKL was overexpressed in CML patients compared with normal samples, while its expression level was lower in complete remission (CR) patients than in corresponding CML patients. Moreover, CRKL was down-regulated in hemin-induced erythroid differentiation of K562 cells. Furthermore, we have demonstrated, for the rst time, that CRKL downregulation promoted K562 cell erythroid and megakaryocyte differentiation via the Raf/MEK/ERK/Elk-1 pathway. Conversely, CRKII was slightly overexpressed in CML patients and had no effect on erythroid differentiation of K562 cells. Our results provide novel insights into CRKL regulates erythroid and megakaryocyte differentiation through the Raf/MEK/ERK/Elk-1 pathway, and suggest that CRKL may serve as a potential target for therapeutic treatment and prognosis of CML disease.

Materials And Methods
Patients and blood samples A total of 33 CML patient samples, 5 CR patient samples and 13 healthy subject normal samples were collected from the Department of Hematology, The Second A liated Hospital of Dalian Medical University, Dalian, China. The mononuclear bone marrow (BM) and peripheral blood (PB) cells were separated from the CML patients and healthy subject normal samples, respectively. All mononuclear cell specimens were frozen in liquid nitrogen immediately after separation and stored at -80 ºC prior to RNA isolation. The study protocol was approved by the Medical Ethics Committee of Dalian Medical University and informed consent was obtained from all patients. All experiment methods were performed in accordance with the relevant guidelines and regulations.
Benzidine staining assay 1 × 10 5 K562 cells in 2 ml RPMI-1640 medium supplemented with 15% FBS were seeded into a well of a 6-well plate, then treated with 20 µM hemin (Sigma-Aldrich, Japan) in a humidi ed incubator with 5% CO 2 at 37 o C for 0, 1 and 2 d before harvesting and washing once with PBS. Benzidine dihydrochloride solution (Beyotime, China) was prepared with 0.5% acetic acid containing 0.1% H 2 O 2 , then 9 µl benzidine dihydrochloride solution was directly added to 81 µl cell suspension, incubated at room temperature (RT) for 5 min, and immediately imaged by an upright light microscope (Olympus, Japan) with 100 × magni cation. Benzidine-positive cells were stained blue, while benzidine-negative cells were light yellow.
Isobaric tags for relative and absolute quantitation (iTRAQ) proteomic analysis 3 × 10 7 shRNA-NC-K562 and shRNA-CRKL-K562 group cells were harvested in at least three independent experiments, and centrifugated at 1000 rpm for 5 min and then the cell pellets were washed with ice-cold PBS. Total protein was extracted from each group of cells using SDT buffer. The SDT buffer was added to the sample and boiled for 15 min, then the supernatant was collected by centrifugation at12000g for Quantitative real-time RT-PCR (qRT-PCR) assay Total RNA was extracted from patient samples and K562 cells using Trizol™ reagent (Invitrogen, USA) and reversely transcribed into cDNA using PrimeScript™ RT Kit with gDNA Eraser (Takara, Japan). qRT-PCR was then performed using FastStart universal SYBR Green Master (ROX) (Roche, USA) with an Mx3005P Real-time PCR System (Agilent, USA). ACTB was used as internal reference. The relative expression levels of CRKL, CRK , γ-globin, GPA (glycophorin), CD41, CD61, Elk-1, GATA-1, HMGB2 in different groups of cells and in CML patient samples were analyzed using the 2 −△△CT method. The speci c primers of CRKL, CRK , γ-globin, GPA, CD41, CD61, Elk-1, GATA-1, HMGB2 and ACTB are provided in Table 1. cDNA microarray assay 1 × 10 7 shRNA-NC-K562 and shRNA-CRKL-K562 group cells were harvested for total RNA extraction using Trizol™ reagent (Life Sciences). The RNA concentration and quality were assessed using a NanoDrop 2000 spectrophotometer (Thermo) and 1.5% denaturing agarose gel electrophoresis. cDNA was synthesized using SuperScript II kit and puri ed by QIAGEN RNeasy Mini Kit. cRNA was created using a Genechip IVT labeling kit. The biotin-labeled fragmented cRNA (≤ 200 nt) was hybridized at 45 °C for 16 h on a Affymetrix Genechip (Human Transcriptome Array 2.0). All the arrays were imaged by a 3000 7G Scanner and processed by Affymetrix Genechip Operating Software. The random variance model (RVM) t-test was performed to screen the differentially expressed genes between the shRNA-NC-K562 and shRNA-CRKL-K562 group cells.

Statistical analysis
Statistical analysis was performed using GraphPad Prism 5.0 software. The data were presented as mean ± SD of at least three independent experiments. Student's t-test was performed to measure the differences between two groups. Differences with P < 0.05 were considered statistically signi cant.

The expression patterns of CRKL and CRK in CML patients
To investigate the potential role of CRKL in CML, we examined the expression pattern of CRKL in 33 CML patient BM samples, 5 CML CR patient BM samples and 13 normal PB samples by qRT-PCR. Our results showed that CRKL was almost universally overexpressed in CML patient BM samples (29/33), the mRNA expression level of CRKL was upregulated 6.2-fold in CML patient BM samples compared with normal samples (P = 0.009, Fig. 1A). We further compared the mRNA expression level of CRKL in 5 pairs of CML primary and CR patient samples. Interestingly, the mRNA expression level of CRKL was lower in CR patient samples than in the corresponding CML primary patient samples, the mRNA expression level of CRKL was downregulated 47.1% in CR patient samples compared with CML patient samples (P = 0.0165, Fig. 1B). Our results indicate that CRKL is highly expressed in CML and plays a crucial role in the development and progression of CML, it may be a potential diagnostic and therapeutic biomarker for CML.
Meanwhile, we examined the expression pattern of CRK in 33 CML patient BM samples, 5 CR patient BM samples and 13 normal PB samples. Our results showed that CRK was slightly overexpressed in CML patient BM samples (20/33), although there was no statistically signi cant difference, the mRNA expression level of CRK was upregulated 1.8-fold in CML patient BM samples compared with normal samples (P = 0.0855, Fig. 1C). We further compared the mRNA expression level of CRK in 5 pairs of CML primary and CR patient samples. The mRNA expression level of CRK was lower in CR patient samples than in the corresponding CML primary patient samples, the mRNA expression level of CRK was downregulated 41.9% in CR patient samples compared with CML patient samples, but there was no signi cant difference (P = 0.1014, Fig. 1D). Our results indicate that CRK is only slightly overexpressed in CML and may not play a very important role in the development and progression of CML.
CRKL was down-regulated during hemin-induced erythroid differentiation of K562 cells K562 cells can be differentiated into erythroid cells by treatment with hemin, so we investigated the expression pattern of CRKL during hemin-induced erythroid differentiation of K562 cells. After treatment with hemin, K562 cells showed signi cant increases in the number of benzidine-positive cells in a timedependent manner. The benzidine-positive rates of K562 cells induced by hemin for 0, 1, 2 d were 0.4%, 32.9% (P = 0.0026) and 40.3% (P = 0.0009), respectively ( Fig. 2A). Meanwhile, the mRNA expression levels of erythroid differentiation markers γ-globin and glycophorin (GPA) were also increased in K562 cells after treatment with hemin ( Fig. 2B), indicating the erythroid differentiation of K562 cells was successfully induced by hemin. Then we measured the protein expression level of CRKL during erythroid differentiation of K562 cells. WB results showed that the CRKL protein expression level was signi cantly downregulated by 52.7% (P = 0.0007) and 54.5% (P = 0.0004) in K562 cells following treatment with hemin for 1 and 2 d (Fig. 2C), respectively. Taken together, CRKL expression is downregulated during erythroid differentiation of K562 cells, indicating a potential role for CRKL in erythroid differentiation.
Meanwhile, we investigated the expression pattern of CRK during erythroid differentiation of K562 cells. No obvious protein level change was observed for CRK during erythroid differentiation of K562 cells.
CRK protein expression level was only slightly upregulated by 7.0% (P = 0.2893) and 6.8% (P = 0.5675) in K562 cells following treatment with hemin for 1 and 2 d (Fig. 2D), respectively, indicating CRK might not be involved in erythroid differentiation of K562 cells.
Moreover, we screened the differentially expressed proteins between shRNA-CRKL-K562 and shRNA-NC-K562 cells by iTRAQ quantitative proteomic analysis. A total of 215 proteins were identi ed as up-or down-regulated over 1.2-fold (P < 0.05) in shRNA-CRKL-K562 cells compared with shRNA-NC-K562 cells. Among these differently expression proteins, 53 proteins were up-regulated and 162 proteins were downregulated in shRNA-CRKL-K562 cells compared to shRNA-NC-K562 cells, these proteins were clustered as shown in Fig. 4A. Gene ontology analysis was performed on the differentially expressed proteins, these differentially expressed proteins were related to positive regulation of erythrocyte differentiation, leukocyte differentiation, positive regulation of megakaryocyte differentiation, hemoglobin complex, regulation of erythrocyte differentiation and megakaryocyte differentiation, indicating CRKL deregulation is associated with differentiation of K562 cells.
Among these targeting proteins, we also focused on the molecules associated with erythroid differentiation. Consistently with microarray results, we found that hemoglobin molecules HBE1, HBD, HBZ, HBG1, erythroid speci c transcription factor GATA-1 and HMGB2 were upregulated 1.  (Table 2), respectively. We detected the expression level of GATA-1 and HMGB2 by WB and qRT-PCR to validate the proteomic analysis results. The protein expression levels of GATA-1 and HMGB2 were upregulated 63.0% (P = 0.004) and 54.0% (P = 0.0391) in shRNA-CRKL-K562 cells compared to shRNA-NC-K562 cells (Fig. 4B), respectively, while, the mRNA expression levels of GATA-1 and HMGB2 were upregulated 29.5% (P = 0.0041) and 44.1% (P = 0.0012) in shRNA-CRKL-K562 cells compared to shRNA-NC-K562 cells (Fig. 4C), respectively. The expression pro le was consistent with proteomic analysis results, indicating the proteomic analysis results were believable. Our results indicated that CRKL downregulation promoted hemoglobin molecules expression, which resulted in erythroid differentiation of K562 cells. Taken together, the gene microarray and iTRAQ quantitative proteomic analysis further con rmed that CRKL downregulation promotes erythroid differentiation of K562 cells.
CRKL knockdown promoted megakaryocyte differentiation of K562 cells In addition, we also evaluated the effect of CRKL knockdown on megakaryocyte differentiation of K562 cells. K562 cell approximates to the megakaryocyte-erythrocyte progenitor stage, which has the potential to differentiate into megakaryocytes. As shown in Fig. 5A, compared to shRNA-NC-K562 cells, shRNA-CRKL-K562 cells exhibited typical characters of megakaryocyte differentiation with an increase in cell size, polyploidization and the presence of vacuoles. The percentage of megakaryocyte cells in the shRNA-CRKL-K562 group was higher than in the shRNA-NC-K562 group. The megakaryocyte surface differentiation markers CD41 and CD61 were also determined by qRT-PCR. The mRNA expression levels of CD41 and CD61 were signi cantly increased 73.6% (P = 0.0302) and 47.15% (P = 0.0234) in shRNA-CRKL-K562 compared to shRNA-NC-K562 cells, respectively (Fig. 5B). Our results indicated that CRKL downregulation also promotes megakaryocyte differentiation of K562 cells.
CRKL knockdown promoted erythroid and megakaryocyte differentiation of K562 cells via activating the Raf/MEK/ERK/Elk-1 signaling pathway The underlying molecular mechanisms of CRKL on erythroid and megakaryocyte differentiation is unknown. Current work links CRKL downregulation to promoting K562 cell erythroid and megakaryocyte differentiation via activating the Raf/MEK/ERK/Elk-1 signaling pathway.
Elk-1 is the downstream molecule of ERK, which is associated with erythroid differentiation. Our results showed that CRKL downregulation increased the mRNA expression levels of Elk-1, the mRNA expression level of Elk-1 was upregulated 44.1% (P = 0.0131) in shRNA-CRKL-K562 cells compared with shRNA-NC-K562 cells (Fig. 6A). Clearly, CRKL mediates the K562 cell erythroid and megakaryocyte differentiation might be via the Raf/MEK/ERK/Elk-1 signaling pathway.
CRK has no effect on erythroid differentiation of K562 cells CRK may be unimportance for erythroid differentiation, to con rm the effect of CRK on erythroid differentiation, we transiently transfected K562 cells with siCRK to knockdown CRK . CRK protein and mRNA levels were decreased by 53.6% (P = 0.0015) and 43.5% (P = 0.0059) in siRNA-CRK -K562 cells compared with siRNA-NC-K562 cells (Fig. 7A), providing a control study for the downregulation effect of CRK on K562 cell erythroid differentiation. qRT-PCR detected the expression level changes of erythroid genes after CRK knockdown, there were no obvious changes in γ-globin (P = 0.088) and GPA (P = 0.133) mRNA expression levels between siRNA-CRK -K562 and siRNA-NC-K562 cells (Fig. 7B). Meanwhile, to show that CRK downregulation did not affect the Raf/MEK/ERK pathway, we measured the expression level changes of p-Raf, p-MEK, p-ERK1/2 after CRK knockdown by WB. No changes were observed for p-Raf, p-MEK, p-ERK1/2 (Fig. 7C). Clearly, our results further demonstrated CRK has no effect on erythroid differentiation of K562 cells.

Discussion
Hematopoiesis is a highly and precisely regulated multistage process by which all of the different cell lineages (erythroid cells, lymphocytes and myeloid cells) that form the immune and blood systems originate from pluripotent stem cells [33,34]. Erythropoiesis happens in human red bone marrow after kidneys responses to low levels of oxygen by releasing erythropoietin [35]. Erythropoiesis is a multi-step cellular course by which a primitive multipotent HSC experiences a series of differentiations resulting in production of erythroid lineage, undergoing erythroid progenitors (colony-forming unit erythroid [CFU-E] and burst-forming unit erythroid [BFU-E]), normoblasts, proerythroblasts, early basophilic erythroblasts, late basophilic erythroblasts, polychromatic erythroblasts, orthochromatic erythroblasts, reticulocytes, ultimately differentiating to mature erythrocytes [5,6]. Megakaryopoiesis occurs through a hierarchical series of progenitor cells, multipotent progenitor (MPP), common myeloid progenitor (CMP) and megakaryocyte-erythroid progenitor (MEP), megakaryocyte progenitor (MKP), ultimately differentiating to mature megakaryocytes [36]. The two dynamic processes are mediated by a balance of intrinsic and extrinsic factors, containing transcription factors, growth factors and miRNAs, and destruction of the two dynamic processes leads to CML. Tyrosine kinase inhibitors (TKIs) targeting BCR-ABL for CML therapy have effectively improved the survival of CML patients, however, about 20% of CML patients have not been bene ted from TKIs treatment, commonly due to TKIs resistance which lead to disease relapse and progression [37][38][39]. Therefore, it is urgent to seek more e cient therapeutic strategies to overcome the problem. Deeper study of the molecular mechanisms governing the development, progression and differentiation of CML can lead to nding novel therapeutic targets and improving the therapy effects for CML patients.
CRK proteins are the predominant phosphorylation substrates for BCR-ABL, which is found in over 95% of CML and 25% of acute lymphoblastic leukemias (ALL) [40]. Although CRKII and CRKL share a high degree of homology within their functional domains, CRKL is the major tyrosine-phosphorylated protein in BCR-ABL-driven CML patient neutrophils [40]. The preferential binding of BCR-ABL to CRKL, even in the presence of CRKII [41], suggests disparity in interaction properties and differential regulation of CRK proteins by BCR-ABL or ABL tyrosine kinases. These nds imply that CRKII and CRKL may play different role in CML, so in this work we investigated the exact effect of CRKII and CRKL on erythropoiesis and megakaryopoiesis of CML. The current study illustrated for the rst time that CRKL but not CRKII inhibits erythroid and megakaryocyte differentiation via the inactivating Raf/MEK/ERK/Elk pathway.
CRKL deregulation is linked to the development and progression of a variety of cancers [26][27][28]. As we summarized in our review [25], abnormal CRKL expression is associated with gastric cancer, glioblastoma multiforme, hepatocellular carcinoma, bladder cancer, lung cancer, colon cancer, ovarian cancer, leukemia, breast cancer, head and neck cancer, rhabdomyosarcoma and neuroblastoma. It is of promise as an indicator for cancer development, invasion and metastasis as well as an attractive target for the diagnosis and prognosis of cancer. CRKL is a major tyrosine-phosphorylated protein in CML cells, pCRKL plays a special role in CML pathogenesis, and the constitutive phosphorylation of CRKL is unique to CML, which makes it a useful target for therapeutic intervention [42][43][44]. We previous reported that CRKL is associated with proliferation, migration and invasion of hepatocarcinoma and clear cell renal cell carcinoma cells [45][46][47][48][49][50]. However, the exact role of CRKL in CML is unknown. Our current work showed that the upregulation of CRKL potentially promotes the clinical development and progression of CML patients and enhances CML cell aggressiveness. CRKL was universally overexpressed in CML patient samples compared with normal samples (Fig. 1A). Interestingly, CRKL expression level was lower in CR patient samples than in corresponding CML patient samples (Fig. 1B). Our results indicate that CRKL is a tumor promoter playing a vital role in the development and progression of CML. To the best of our knowledge, this work is the rst reporting the expression pattern of CRKL in CML patients, CR patients and normal samples. CRK deregulation is also linked to the development and progression of a variety of cancers [26][27][28], But our results show that CRK is only slightly overexpressed in CML (Fig. 1C, D), and that it may not play an important role in the development and progression of CML.
It is known that the CRK family plays important roles in the regulation of cell differentiation. v-CRK overexpression can increase rat pheochromocytoma PC12 cell differentiation [29], and both SH2 and SH3 domains of the CRK protein are required for neuronal differentiation of PC12 cells [30]. Moreover, CRK enhances osteoclast differentiation by activating Rac1, the overexpression of CRK and CRKL signi cantly enhances RANKL-induced osteoclast differentiation, and the downregulation of CRK and CRKL synergistically decreases RANKL-induced osteoclast differentiation [31]. The effect of CRKL and CRK on leukemia cell differentiation has not been reported, in our study we investigated the potential role of CRKL and CRK in erythroid and megakaryocyte differentiation of K562 cells. Hemin is an ironcontaining porphyrin which is involved in oxygen delivery and used to treat acute porphyria and thalasssemia intermedia, and is also a relatively strong inducer for heme biosynthesis of K562 cell erythroid differentiation [51]. Using K562 cells as a model, we found that CRKL expression level was downregulated in hemin-induced erythroid differentiation of K562 cells (Fig. 2C), indicating CRKL might play an important role in erythroid differentiation of K562 cells. In order to verify the hypothesis, we selected previously successfully constructed CRKL stably downregulated monoclonal cell lines to investigate the effect of endogenous CRKL on erythroid differentiation. We further found that CRKL downregulation promoted erythroid differentiation of K562 cells with more benzidine-positive cells and higher mRNA expression levels of γ-globin and GPA (Fig. 3). Moreover, CRKL downregulation enhanced megakaryocyte differentiation of K562 cells with increased number of megakaryocyte cells and higher mRNA expression levels of CD41 and CD61 (Fig. 5). Our results rst demonstrate CRKL is a new regulator of erythroid and megakaryocyte differentiation of K562 cells. Furthermore, we screened the differentially expressed molecules between shRNA-CRKL-K562 and shRNA-NC-K562 cells using gene microarray and iTRAQ quantitative proteomic analysis. Results showed hemoglobin molecules HBD, HBA1, HBA2, HBZ, HBE1 and HBG1 were more upregulated in shRNA-CRKL-K562 than in shRNA-NC-K562 cells (Tables 2 and  3). Moreover, GATA-1 and HMGB2 expression were increased in shRNA-CRKL-K562 than in shRNA-NC-K562 cells (Fig. 4), which are crucial for erythrocyte and megakaryocyte lineages. The zinc-nger transcription factor GATA-1 binds to GATA/AATC consensus elements in the globin gene clusters and other erythroid or megakaryocytic cell-speci c genes [52]. The zinc-nger proto-oncogene G -lb is an erythroid-speci c transcription factor that plays an important role in erythropoiesis [53], G -1B gene disruption results in embryonic death of mice due to failure to produce red blood [54]. The G -1B promoter contains 2 tandem sites which includes both a GATA-1-and an Oct-1-binding sequence [55]. HMGB2 bends DNA at the G -1B promoter by binding to the G -1B promoter to facilitate the binding of Oct-1 to the G -1B promoter [56], subsequently enhancing the binding of GATA-1 to the AATC sites of G -1B promoter and activating the transcription of G -1B [57]. Our results show that CRKL regulates erythroid and megakaryocyte differentiation of K562 cells by upregulating GATA-1 and HMGB2 expression.
However, the expression level of CRK was not changed in hemin-induced erythroid differentiation of K562 cells (Fig. 2D) and CRK downregulation did not affect erythroid differentiation of K562 cells (Fig. 7B). Moreover, we further investigated the effect of CRK knockdown on erythroid differentiation of K562 cells by transiently transfecting siCRK in shRNA-CRKL-K562 cells, interestingly, consistent with the above results, CRK knockdown in K562 cells with CRKL downregulation (Fig. 7F). Collectively, CRK is not associated with erythroid differentiation of K562 cells. Although CRKII and CRKL have a high degree of similarity in sequence, the two isoforms vary in ligand a nities and speci city, and the 3-dimensional structures of CRKII and CRKL differ to engage key signaling partners [21,28]. The respective knockout mice have distinct phenotypes but both proteins are required for embryonic development [58,59]. So CRKII and CRKL might function differently in leukemogenesis, erythropoiesis and megakaryopoiesis of CML, which deserves more attention to understand the differences between the two CRK adapter proteins.
Our results are also consistent with the previous report that CRKL expression level is highest in adult hematopoietic tissues and low in epithelial tissues, whereas CRKII exhibits the highest expression in the brain, lung, kidney and low expression in bone marrow [60].
The Ras/Raf/MEK/ERK signaling pathway is involved in erythropoiesis which mainly promotes growth, differentiation and prevents apoptosis of hematopoietic cells [61][62][63] indicating CRK has no effect on erythroid differentiation of K562 cells.
Taken together, we have illustrated for the rst time that CRKL can inhibit erythroid and megakaryocyte differentiation of K562 cells via inactivating the Raf/MEK/ERK/Elk pathway. The novel action mechanism is outlined in Fig. 8. CRKL downregulation promotes the expression of Raf, p-Raf, p-MEK, p-ERK1/2 and EIk-1, then HMGB2 binds to the G -1B promoter and enhances its transactivation by promoting the binding of Oct-1 and GATA-1 to the G -1B promoter, which induces erythroid and megakaryocyte differentiation of K562 cells by increasing globin, hemoglobin and differentiation-speci c genes expression. Taken together, we have established a new functional role and molecular pathway for CRKL during hematopoietic differentiation. These ndings could be fundamental to the development of a novel potential diagnostic biomarker and therapeutic target for CML patients.

Conclusions
In conclusion, the different features of CRK and CRKL indicates that they may serve differently in leukemogenesis. Our ndings point to CRKL rather than CRK as a biomarker associating with differentiation of CML. Their different functions in CML cells may result from different preferential interactions with binding partners, thereby activating different signaling pathways leading to different roles in CML.   Tables  Table 1  Synthesized sequences of primers for targeting genes   Targeting gene Primer sequence   CRKL  F:5'-GTGCTTATGACAAGACTGCCT-3      (C) γ-globin and GPA mRNA expression levels were detected in shRNA-CRKL-K562 and shRNA-NC-K562 cells by qRT-PCR. *, **, ***refer to P values <0.05, <0.01, <0.001.