苏州大学 尹斌 教授
1. BMI1 reprogrammes histone acetylation and enhances c-fos pathway via directly binding to Zmym3 in malignant myeloid progression（陈子兴课题组）
陈子兴教授课题组最近有一篇文章被Journal of cellular and molecular Medicine接受。在这篇文章中，陈子兴老师课题组新近报道了有趣的发现，通过对白血病病人标本的分析和分子、细胞水平的研究，深入地揭示了BMI1基因参与导致白血病发生的分子作用机制，可能通过抑制ZMYM3基因的转录，调控的组蛋白乙酰化水平，进一步影响了下游重要基因（包括FOS, RUNX1, PTEN, 等）的活性，从而增强了造血细胞的增殖和抗凋亡能力，促进了白血病的发生。”
Title: BMI1 reprogrammes histone acetylation and enhances c-fos pathway via directly binding to Zmym3 in malignant myeloid progression
Abstract: The polycomb group BMI1 is proved to be crucial in malignant myeloid progression. However, the underlying mechanism of the action of BMI1 in myeloid malignant progression was not well characterized. In this study, we found that the patients of both myelodysplastic syndromes and chronic myeloid leukaemia with BMI1 overexpression had a higher risk in malignant myeloid progression. In vitro gene transfection studies showed that BMI1 inhibited cell myeloid and erythroid differentiation induced by 12-O-tetradecanoyl phorbol-13-acetate (TPA) and histone deacetylase inhibitor sodium butyrate respectively. BMI1 also resisted apoptosis induced by arsenic trioxide. Moreover, the transcript levels of Runx1 and Pten were down-regulated in Bmi1-transfected cells in company with histone deacetylation modification. By using chromatin immunoprecipitation (ChIP) collaborated with secondary generation sequencing and verified by ChIP-PCR, we found that BMI1 directly bound to the promoter region of Zmym3, which encodes a component of histone deacetylase-containing complexes. In addition, as one of the downstream target genes of this complex, c-fos was activated with increasing histone acetylation when ZMYM3 was suppressed in the Bmi1-transfected cells. These results suggested that BMI1 may reprogramme the histone acetylation profile in multiple genes through either indirect or direct binding effects which probably contributes to the malignant progression of myeloid progenitor cells.
Results and Figures:
1. BMI1 correlated with a high risk of malignant myeloid progression in both MDS and CML patients
Q-PCR showed that Bmi1 transcription was up-regulated in MDSBMMCs compared to non-MDS cytopaenias. The Bmi1 transcription in MDS BMMCs was markedly decreased after MDS patients achievedmorphological and cytogenetic remission, P < 0.05 (Fig. 1A). Further statistical analysis showed that the Bmi1 transcription in MDS BMMCs displayed an increasing tendency from the group with blasts lower than 5% to the group with blasts higher than or equal to 5%, but a statistically significance was not reached (Fig. 1B). However, a significance difference of the Bmi1 transcription in MDS CD34+ cells had been found between these two groups, P < 0.05 (Fig. 1C). In addition, the Bmi1 transcription in MDS CD34+ cells also positively correlated with the IPSS, R2 = 0.597 P < 0.01. These results indicated that the abnormal transcript level of Bmi1 in MDS was mainly in undifferentiated stage. Interestingly, the MDS patients with higher transcript level of Bmi1 had a higher frequency of disease progression towards AML after diagnosis, as was demonstrated in one MDS-RA case progressed to MDS-RAEB and two MDS-RAEB cases transformed into AML in our patient cohort (Fig. 1D). Furthermore, the Bmi1 transcription of MDS-AML CD34+ cells was even much higher than that of dAML (Fig. 1C). Similar results were found in CML myeloid progression. The Bmi1 transcription in CML-BP BMMCs was much higher than that in CML-CP BMMCs, P < 0.05. Same as MDS, the Bmi1 transcription of CML-BP was also much higher than that of matched dAML, P < 0.05 (Fig. 1E).
Fig. 1 Bmi1 transcript expression in myeloid progression
2. The higher level of BMI1 in leukaemic cells changed their phenotypes
As there was no distinct MDS cell line available and the K562 cell line was established from the bone marrow of a CML patient, the open reading frame of Bmi1 was transfected into K562 and U937 leukaemic cell by retroviral expression system . Two subclones of K562 transfectant with more than 2-folds BMI1 levels and two subclones of U937 transfectant with about 2-folds BMI1 levels were obtained by limited dilution (Fig. 2A). After the cells were cultured in media without FBS for 72 hrs, the viabilities of subcloned transfectant cells were higher than that of controls, P < 0.05 (Fig. 2B). Although the overexpression of BMI1 in K562 and U937 cells seemed not to increase the cell proliferation, the colony-forming assay showed that the colony size of Bmi1-transfected cell colony was markedly bigger than that of control cells. After two cycles of colony-replating assay, the number of Bmi1-transfected cell colony, of which the diameter was larger than 0.3 mm, was 2.5 ±0.4*103/ml which was significantly greater than 1.3±0.3*103/ml of control cell colony, P < 0.05 (Fig. 2C). On the other hand, the annexin V assay performed by FCM demonstrated that the Bmi1-transfected K562 cells became less apoptotic compared with the control cells after treatment with 5 lM ATO for 48 hrs. BMI1 seemed not to reduce the ratio of early apoptotic cells, but mainly slow down the proceeding of early apoptotic cells into late apoptotic cells, indicating that the overexpression of BMI1 allowed cells becoming more resistant to apoptosis induced by ATO in K562 (Fig. 2D). Differentiation block is the major mechanism of MDS and CML progression into AML. To further explore the effect of BMI1 upon cell differentiation, we examined the TPA-induced myeloid differentiation and the NaB-induced erythroid differentiation in K562 which had the multiple linage differentiation ability. Nitroblue tetrazolium reduction assay showed that the percentage of differentiated cells in K562 parental cells was considerably higher than that in Bmi1-transfected cell after cells treated by 20 nM TPA for 72 hrs (Fig. 3A and B). The mean fluorescence intensity and percentage of CD15 positive cells determined by FCM were lower in Bmi1-transfected cell than that in control (Fig. 3C). Both the NBT reduction assay and the FCM results indicated that the up-regulation of BMI1 hindered TPA-induced myeloid differentiation in K562. Similar results were found in NaB-induced K562 differentiation. Benzidine staining showed that the Bmi1-transfected K562 had a lower erythroid differentiation compared to that ofthe parental control cells after treatment by 0.5 mM NaB for 72 hrs (Fig. 3D and E). Furthermore, FCM also showed that the Bmi1-transfected K562 underwent a slower NaB-induced erythroid differentiation by detecting CD71 and GPA (Fig. 3F). The mean fluorescence intensity of CD71-positive cells in K562 control was much higher than that in Bmi1-transfected K562. The ratio of CD71+GPA+ cells in control K562 was also much higher than that in Bmi1-transfected K562. These results indicated that the overexpression of BMI1 inhibited cell erythroid differentiation induced by HDAC inhibitor NaB, implying a possible influence of BMI1 on the histone modification in K562 cells.
Fig. 2 BMI1 enhances the malignancy of leukaemic cells
Fig. 3 BMI1 inhibits 12-O-tetradecanoyl phorbol-13-acetate (TPA) and histone deacetylase inhibitor sodium butyrate (NaB) induced differentiation in K562
3. BMI1 inhibits transcription of Runx1 and Pten via multiple histone modification but without direct binding
To further verify the correlation of BMI1 with the malignant myeloid progression and a poor prognosis in patients, especially, to explorethe molecular mechanism by which the Bmi1 transfection altered the phenotype of K562 cells, according to Bejar’s report, we have firstly analysed the transcription profile of Runx1, Ezh2, Idh2, Pten, Etv6, Cbl, Nras, Asxl1 and Tp53 genes in CD34+ cells of MDS patients by microarray (Fig. 4A) . Among the genes which were down-regulated in all subtypes of MDS patients, Runx1 and Pten were selected and focused on, because the former one is known to be responsible for the regulation of early haematopoietic differentiation, while the latter gene known as a tumour suppressor played pivotal role in distinguishing the leukaemic stem cell from the normal haematopoietic stem cell. Congruously, the transcript levels of both these genes were significantly suppressed in Bmi1-transfected K562 cells detected by Q-PCR, P < 0.05 (Fig. 4B). As BMI1, as a member of polycomb group, participates the epigenetic regulation of target gene via joining the PRC1, we have performed a series of ChIP assays to further elucidate the epigenetic regulatory effects on transcription of these two genes exerted by BMI1, including histone trimethylation of H3K9 and H3K27, histone acetylation of H3 and H4. In addition, the interaction of EZH2, a componant of PRC2, with BMI1 in the transfected K562 cells was also studied. The results indicated that the trimethylation of histone H3K27 (H3K27me3) in chromatin correlated with Runx1 and Pten promoter reduced in Bmi1-transfected K562 cells. As EZH2, which catalyses H3K27me3, was down-regulated in MDS CD34+ cells according to MDS CD34+ cell microarray and was suppressed in Bmi1-transfected K562 cells verified by western blotting. EZH2 suppressed by enforced BMI1 levels was probably responsible for the reduction in H3K27m3 for Runx1 and Pten promoter regions (Data S1). Meanwhile, the histone H3 acetylation in nucleosomes at Runx1 and Pten promoter regions was down-regulated in Bmi1-transfected K562 cells validated by ChIP and Q-PCR, P < 0.05 (Fig. 4C and D), which was consistent with the results that BMI1 counteracted the differentiation induced by HDAC inhibitor NaB. However, ChIP-PCR results failed to find evidence for the direct binding of BMI1 to the promoter regions of Runx1 and Pten. The ChIP-PCR using antibody against ubiquitin-H2A, a substrate of PRC1, also showed negative results. Therefore, the above epigenetic alterations relevant to suppressing the transcription of Runx1 and Pten must be an indirect effect of BMI1 in K562 cells.
Fig. 4 BMI1 indirectly inhibits Runx1 and Pten expression with histone deacetylation
4. BMI1 directly suppressed the transcription of Zmym3 and enhances expression of c-fos by modified histone acetylation
To seek the possible direct target gene of BMI1, we have performed ChIP sequencing on the K562 and SKM-1 cells. A list of several dozens of genes with various functions could be obtained and judged to be directly bound by BMI1. Among them, Zmym3, which translates a component of histone deacetylase-containing multiprotein complexes, was confirmed to be the direct target of BMI1. Furthermore, ChIP-PCR assay using antibody against ubiquitin-H2A, a substrate of PRC1, also indicated that Zmym3 gene promoter region could be amplified from the precipitations in K562, U937 and SKM-1. In comparison with the parental K562 cells and K562 cells transfected with MSCV vector, the amplified product of Zmym3 promoter fragment increased parallelly in both ChIP assays using antibody against BMI1 or against ubiquitin-H2A in the Bmi1-transfected K562 cells (Fig. 5A). Meanwhile, the transcription of Zmym3 was markedly reduced, P < 0.05 (Fig. 5B). To further confirm the effect of BMI1 on the expression of ZMYM3, we looked at the transcription of c-fos, one of the ZMYM3 downstream target genes. As being expected, BMI1 increased the transcription of c-fos, in company with a higher level of histone H3 and H3K27 acetylation detected by ChIP and Q-PCR, P < 0.05 (Fig. 5C and D). Taken together, these results indicated that BMI1 suppressed ZMYM3 by direct binding to Zmym3 promoter region and enhanced the downstream c-fos pathway in K562 cells by modifying histone acetylation.
Fig. 5 BMI1 directly suppresses Zmym3 and activates c-fos pathway