Expression profile of KEL in AML patients and cell lines
To better distinguish AEL (M6) from other AML subtypes, we screened The Cancer Genome Atlas (TCGA) database to select specific oncogene in AEL. Using this database, we found that KEL was specifically highly expressed in M6 patients compared with other subtype of AML patients (Fig. 1A and 1B). In addition, M6 cell lines--HEL and K562 showed higher expression of KEL than myeloid neoplasm cell lines (Fig. 1C and D, Additional file 4: Fig. S1A). The potential clinical significance of KEL was then investigated. RT-qPCR and agarose gel electrophoresis analysis demonstrated that, consistent with TCGA database, patients with AML-M6 showed higher level of KEL mRNA compared with non-M6 patients and healthy donors (Fig. 1E and Additional file 4: Fig. S1B). And the level of KEL protein was significantly elevated in AEL patients (Additional file 4: Fig. S1B). Further clinical analysis of TCGA database showed that KEL expression level has none business of age and gender (Additional file 4: Fig. S1C and D). However, patients presented with intermediate- or poor-risk assessment of molecular have higher KEL level compared to good ones (Fig. 1F). Combined the clinical characteristics of 20 AEL patients, analysis indicated that there are no significant differences in the levels of white blood cells (WBCs), platelets and hemoglobin between patients with different KEL expression (Additional file 4: Fig. S1E-G). In sum, KEL as a specific indicator in AEL exhibited its potential role in tumorigenesis and serving as diagnosis and prognosis biomarker.
KEL regulates AEL cell proliferation and its downregulation reverses drug resistance of JQ1
To explore the function of KEL in AEL, we designed three KEL siRNAs and separately transfected K562 and HEL cells, and then assessed the knockdown efficiency. Both qRT-PCR and western blot analysis confirmed that siRNA#2 was the most effective for knockdown of KEL expression (Additional file 5: Fig. S2A). We then constructed KEL shRNA using the sequence of siRNA#2 to stably knockdown the expression of KEL. And to manipulate the expression of KEL, we succeeded to construct the lentivirus-mediated over-expression (ov-KEL) vectors (Additional file 5: Fig. S2B). CCK-8 assay revealed that the K562 and HEL cells in which KEL expression was forced were significantly more likely to exhibit a malignant phenotype than the mock cells. Conversely, reduced KEL expression inhibited the proliferation ability (Fig. 2A). Next, we performed the protein array to find out abnormally activated pathways induced by KEL. The results were shown in Additional file 5: Fig. S2C and Additional file 6: Table. S4. And the additional file 5: Fig. S2D showed the overall phosphorylation change level. One key branch signaling pathway (RafB-MEK1-RSK2-CREB) involved in cell proliferation was finally picked out (Additional file 5: Fig. S2E). We hypothesized that KEL would serve essential functions in AEL, which depend on the pathway. The expression level of BTK and CyclinB1 were used to confirm the array results (Fig. 2B). Most importantly, associated genes of the key brunch proposed to be involved in KEL-mediated cell proliferation that was changed correspondingly with KEL up/down-regulation, indicating that KEL plays critical roles in AEL cell proliferation (Fig. 2C).
Currently, the treatment of AEL still follows the common AML therapeutic strategy. Small-molecule inhibitor JQ1 identified to led to robust antileukemic effects has been studied intensively in multiple subtypes of AML[17]. It has been reported that JQ1 is a hopeful choice that targets AML. However, drug-resistance occurs frequently and the mechanism underlying the difference in leukemia stem cells (LSCs) sensitivity to JQ1 remains elusive[19]. And forecasting analysis observed that JQ1 treatment seemed to be noneffective in K562 though it exerts an inhibitory effect in multiple cell lines and LSCs[20, 21]. According to the database of all the 318 small-molecule inhibitors, K562 was sensitive to 14 inhibitors and resistant only to JQ1 (Fig. 2D). Then we wondered the correlation among KEL, JQ1 and cell proliferation. To figure out whether KEL was associated with the resistance of K562 cells to JQ1, we firstly treated K562 at different dose of JQ1. High-dose of JQ1 induced modest decrement of KEL and cell proliferation signal transduction pathway. However, knocking down KEL significantly strengthened the role of JQ1 (Fig. 2E). CCK8 assay showed that knockdown of KEL forced the role of JQ1, as demonstrated by inhibition of cell proliferation (Fig. 2F). In K562 cells, inhibition of KEL reversed the relative resistance of JQ1 treatment alone, indicating that KEL plays major roles not only in cell proliferation but also in drug resistance.
KEL contributes to gain of H3K27 acetylation and promotes GATA1-induced erythroid differentiation
Hematopoietic process is the development and mature process of various types of blood cells in human. Erythroid differentiation as an important part plays portal role in AEL. K562 cell line established from a patient with chronic myeloid leukemia (CML) in blast crisis is a classical model to study erythroid differentiation in vitro[22]. Erythroid differentiation of K562 was induced with hemin or sodium butyrate (NaBu) treatment. As it is known that GATA1 is a core TF in erythroid differentiation process[23], we then explored the regulatory effect between KEL and GATA1. Inspiringly, we found that the change of KEL expression could affect erythroid differentiation potential. Upregulation of KEL promoted the erythroid differentiation of K562 cells induced by hemin or NaBu (Fig. 3A and Additional file 7: Fig. S3A). The change of KEL expression led to the corresponding increase or decrease of erythroid differentiation markers (γ-globin and fut1) and TF (GATA1) (Fig. 3B and 3C). TCGA database results of the strong positive relevance between KEL and GATA1 indicated the internal regulatory mechanism (Fig. 3D). Using bioinformatics prediction (http://dbtoolkit.cistrome.org/), we found high regulatory potential of H3K27ac and H3K4me3 at the region of GATA1 near transcription start site in K562 (Additional file 7: Fig. S3B). Western blot analysis showed that the level of H3K27ac not H3K4me3 changed with KEL (Fig. 3E). ChIP-seq exhibited the peak of H3K27ac at GATA1 locus region and different primers were designed according to various peaks (Additional file 7: Fig. S3C). Through ChIP-qPCR analysis, we observed the enrichment of H3K27ac at the promoter region of GATA1 and gain of H3K27ac in KEL over-expressed cells (Fig. 3F). Taken together, these data confirmed that KEL promoted the gain of histone sites H3K27ac of GATA1 promoter and partially accounted for the significant activation of GATA1 induced erythroid differentiation.
GATA1 and TAL1 as co-TFs regulate the expression of KEL
TFs networks exert essential roles in erythroid differentiation[25]. Through online database (Cistrome Data Browser), we have selected multiple TFs with high regulatory potential of KEL in K562. GATA1 and TAL1 as famous TFs and POLR2A which encodes the largest subunit of RNA polymerase II showed the highest regulatory potential (Fig. 4A). We've already showcased the highly positive correlation between KEL and GATA1 (Fig. 3D).
TCGA database also suggested that patients with higher level of TAL1 seem to possess higher KEL expression (Fig. 4B). To verify the hypothesis that TAL1 and GATA1 are TFs of KEL, we separately knocked down TAL1 and GATA1 with siRNAs. Results showed that downregulation of GATA1 and TAL1 reduced the expression of KEL (Fig. 4C and D). In addition, consistent with previous research[6], EMSA result revealed that K562 nuclear extract could specifically bind to biotin-labeled probe, and the competition occurred after the addition of the cold probe. With the increase of the cold probe concentration, the competition increased (Fig. 4E). Subsequently, in order to narrow down the area on which GATA1 and TAL1 exerted effects within KEL promoter, dual luciferase reporter assays were performed with truncated segments of KEL. Four shorter fragments with three 600bp and one 500bp of the promoter were cloned. The results implied that GATA1 affected all the promoter regions while TAL1 mainly affected the proximal 200bp (Fig. 4F). Thus, we concluded that the two TFs, GATA1 and TAL1, directly interacts with KEL to coactivate KEL in K562.
KEL enhances tumor cell proliferation and tumor growth in vivo
To verify the in vitro findings, we examined the biological functions of KEL in mediating proliferation in vivo. K562 cells with stably forced (ov-KEL group) and decreased KEL (sh-KEL group) expression were transplanted into NCG mice by tail intravenous injection. Phosphate buffered saline was injected for control group (PBS group). Consistent with the above in vitro findings, the over-expression of KEL dramatically promoted AEL progression. 23 days after injection of K562 cells, mice began to lose weight. Weight loss of mice in ov-KEL group started from 19 days was more dramatic than WT group and sh-KEL group (Fig. 5A). By week 5, the differences of white blood cells (WBCs) have turned up in experimental groups. 7 weeks later, 10 mice survived in WT group and 13 in sh-KEL group, whereas only 2 survived in the ov-KEL group with extremely high WBC counts (Fig. 5B). During the growth phase, lumps were observed in the abdominal cavity and hind limbs in AEL mice. Tumor burden rates were calculated and the tumor formation capability of ov-KEL group was greater than the WT and sh-KEL group (Fig. 5C). The growth state of tumors was observed by whole-body fluorescent imaging system at week 6 post injection (Fig. 5D). Using bioluminescent imaging, we found tumors harvested from ov-KEL group had significantly higher fluorescence signals, and sh-KEL group had lower fluorescence signals compared with those from the WT group. Immunohistochemical (IHC) assay was used to analyze the tumors biopsy specimen and evaluate the pathological feature. And the results proved that KEL could promote tumor cell proliferation (Fig. 5E). The histogram visually displayed the ratios and the significant difference of ki67 positive cells between the three groups (Fig. 5F). Importantly, our results showed that mice transplanted using cells with higher expression of KEL had a significantly worse prognosis. In contrast, knocking down of KEL prolonged the survival of AEL mice, which further verified the role of KEL in AEL (Fig. 5G). In general, these results demonstrated that KEL enhanced proliferation of tumor cells and was strongly associated with the progression and prognosis of AEL.
PD-L1 positively correlates with KEL may induce immune evasion of tumor cells
PD-L1 is known to be typically expressed on the surface of tumor cells and allow them to evade the immune system surveillance[26]. Unexpectedly, our data showed that the majority of proteins that reported to be involved in the biological process of PD-L1 were found to be upregulated in K562 cells with KEL over-expression (Fig. 6A and Additional file 6: Table. S4). The results were further verified by western blot which was consistent with protein array. Most importantly, PD-L1, lowly expressed in K562 cells, was significantly up-regulated after the enhancement of exogenous expression of KEL (Fig. 6B). To probe the functional consequences of PD-L1 expression in AEL and the relationship between PD-L1 and KEL, we screened the database and discovered a positive correlation between the two factors (Fig. 6C). Western blot conducted with total cells isolated from 3 pairs of AEL mice tumors exhibited higher levels of KEL and PD-L1 protein in ov-KEL group compared with WT group (Fig. 6D). Collectively, these results manifested that low expression of PD-L1 in K562 could be enhanced by the upregulation of KEL and perhaps provided possible checkpoint inhibitor therapy strategy for AEL patients.