miR-146b-5p within BCR-ABL1–Positive Microvesicles Promotes Leukemic Transformation of Hematopoietic Cells

Microvesicles (MV) are extracellular vesicles released by most cells and act as mediators of intercellular communication (1). The functions of MVs are complex due to their various bioactive cargo, including DNA, mRNA, miRNA, and proteins (2). Tumor-derived MVs contain specific information of the tumor status, and thus MVs were studied as potential diagnostic markers and targets for therapeutic intervention (1, 3). Recently, MVs have received increasing attention for their roles in regulating and transferring active molecules responsible for tumor progression and metastasis (4). MVs play dual roles in cancer through transferring tumor-promoting molecules and tumor suppressors in different conditions (5, 6). Our previous work demonstrated that MVs derived from K562 chronic myelogenous leukemia (CML) cells can transform mononuclear cells (MNC) from normal hematopoietic transplants to acute leukemia-like cancer cells through genomic instability (7). During the transformation, a new group of leukemia-like cells could be observed after 14 days of consecutive incubation with MVs, and most of them were transformed into leukemia-like cells after 21 days (7). This transformation model provides an opportunity to explore the mechanism of blast crisis (BC) of CML and the occurrence of donor cell leukemia. As a common type of leukemia, CML was considered as a paradigm for understanding the molecular evolution of cancer because it was the first cancer shown to be initiated at the hematopoietic stem cell level by BCR-ABL1 and may undergo blastic transformation from the chronic phase (CP) to BC (8). However, the mechanisms of CML BC transformation are still poorly understood. Thus, elucidating the key factors and exploring their regulatory mechanisms of our model will shed light on the leukemogenesis and transformation of CML.

Transcription factors (TF) and miRNAs are important regulators in the gene expression of hematopoietic system. During leukemogenesis, the aberrant regulation and fusion of TFs, such as RUNX1, SPI1, GATA, AML1-ETO, and CBFB-MYH11, are key to the disease (9). MiRNAs, such as miR-15/16, miR-17-92, and miR-155, were reported with essential functions in the commitment and differentiation of hematopoietic stem cells, as well as the occurrence of leukemia (10, 11). The aberrations of TFs (e.g., AML1 and HOXs) and miRNAs (e.g., miR-150/17/19a/155) have been found to contribute to the disease progression of CML BC (12, 13). Furthermore, TF and miRNA can regulate mutually and coregulate the same target to form feedback loop or feed-forward loop regulatory motifs (14), which are vital and common regulatory motifs in different biologic processes and diseases (14–16).

Our previous work demonstrated that MVs lost their transforming abilities following RNase treatment, indicating that RNAs in MVs were responsible for the transformation (7). To further explore the key regulators and their regulations in the process of K562-MVs transforming MNCs into leukemia-like cells, we sequenced the mRNAs and small RNAs for samples of the critical time points. Then, deep analyses of differentially expressed genes, TFs and miRNAs, as well as the regulatory networks among them were performed. We identified that miR-146b-5p as a key regulator accelerated the transformation by targeting NUMB and other genes, and also caused genome instability and cell proliferation of the recipient cells, which will provide important insights into the leukemogenesis.

Total RNA was isolated from five samples to perform RNA-seq and small RNA-seq. Details of sequencing and expression analysis were in Supplementary Methods. We used the in-house scripts with Poisson distribution to test the differentially expressed genes and miRNAs between different samples. P values adjusted by the Benjamini and Hochberg procedure lower than 0.001 and fold changes higher than two were considered as significant. We also required RPKM ≥ 20 or RPM ≥ 100 in at least one sample for the differentially expressed genes (DEG) or miRNAs (DEM), respectively.

We obtained 161 differentially expressed TFs in the transformation process, and 42 key TFs were selected on the basis of their target information, importance to the hematopoietic system and cancer progression, and expression change during the transformation (upregulated or downregulated constantly, or highly expressed in 3W sample). The key differentially expressed miRNAs were chosen according to the RPM higher than 1,000 in at least one sample or the fold change greater than four times in the comparison of one stage. Finally, 39 key miRNAs were selected from 143 DEMs.

To analyze the cancer-related genes and pathways, we selected 16 cancer-related pathways from KEGG, as well as the chromatin modification gene list from nanoString PanCancer pathways (http://www.nanostring.com/products/pancancer) and tumor-suppressor genes from the TSGene database (17).

We applied the same method described in our review article to construct the miRNA and TF coregulatory network (see Supplementary Methods; 14). We used the expression correlation between regulatory factors and target genes to filter the false-positive regulatory interactions. We required that the absolute value of a TF correlation to its target should be larger than 0.5, and the correlation of an miRNA to its target should be less than 0.5. The network graphics were shown in Cytoscape (18).

The differentially expressed miRNAs and genes were hierarchically clustered using an average linkage algorithm and a Euclidean distance for the distance measure. We used MeV (19) to visualize the clustered data.

The human CML blast crisis cell line K562 was purchased from the China Center for Type Culture Collection (CCTCC) and was authenticated by CCTCC (Wuhan) using the STR genotyping method in December 2014. K562 was cultured in RPMI-1640 containing 15% FBS at 37°C in 5% CO₂. MNCs were extracted from the peripheral blood mobilization of healthy volunteers and were cultured in StemSpan SFEM (#09600; STEMCELL).

MVs isolation was performed by previous protocol: cells were centrifuged at 1,000 × g for 10 minutes. The supernatant was centrifuged at 5,000 × g for 20 minutes to remove cellular debris, and the remaining supernatant was centrifuged at 13,000 × g for 60 minutes to obtain MVs.

Isolated MVs were resuspended with serum-free RPMI-1640 and filtered using a Millipore Steriflip polyvinylidene difluoride filter with a pore size of 1.0 μm (to filter cells). MVs were quantified according to their copies of BCR-ABL1 mRNA. The MNCs were adjusted to 4 × 106 cells per well in a 6-well plate, and MVs were added to the cells three times a day for 13 to 32 days. The morphology of the transformed cells was observed using Wright's stain.

To confirm the effect of miR-146b-5p, K562 cells were transfected with miR-146b-5p mimics and inhibitor: K562 cells were seeded onto 6-well plates (6 × 105 cells/well) the day before transfection. Cells were transfected with 10 μL 20 μmol/L for miR-146b-5p inhibitor, 5 μL 20 μmol/L miR-146b-5p mimics (RiboBio), and 50 ng miR-X vector using riboFECTTM CP Reagent (RiboBio), respectively. Real-time PCR was performed to measure the level of miR-146b-5p in the cells and their MVs. The supernatant of transfected cells was collected 48 hours after transfection to isolate MVs.

To investigate whether miR-146b-5p could complete the transformation without BCR-ABL1, we performed extra work to incubate the MNCs with imatinib at the concentration of 0.25 and 0.5 μm/mL. K562-MV with elevated miR-146b-5p was added to the MNCs-imatinib mixtures to induce the transformation as described above.

Experiments of DNA breaks in recipient cells and intracellular reactive oxygen species (ROS) were performed as our previous study and also in Supplementary Methods. Other detailed experimental methods regarding the luciferase assay, RT-PCR, western blot, etc., were provided in the online Supplementary Files.

The RNA-seq and small RNA-seq data are available at NCBI Sequence Read Archive (SRA) with the accession SRP057826.

To determine the gene-expression change and the mechanism of the process that K562-MVs transform MNCs, we performed RNA-seq and small RNA-seq for five samples of the key time points, which were samples of K562-MVs, MNCs, 1 week(1W)/2 weeks (2W)/3 weeks (3W) cells after MVs incubation (see project design in Supplementary Fig. S1). The 1W sample was considered as reversible of the transformation. Whereas, the 2W sample was an irreversible status and a group of leukemia-like cells was observed at this time. The 3W sample was considered as the finish of the transformation, as the majorities were leukemia-like cells (7). The summary of sequenced data and mapping information were listed in Supplementary Table S1. In the five samples, we detected 18,614 expressed genes (RPKM > 0) and 1,425 expressed miRNAs (RPM > 0). In each sample, about 20% to 25% of the expressed genes were with RPKM > 20, and 3.5% to 5.8% of the expressed genes were at a high level with RPKM > 100 (Supplementary Table S2). For miRNAs, only 169 miRNAs (11.78%) were highly expressed (RPM > 100) in one or more samples, and they occupied more than 99% of the mapping reads. It is surprising that K562-MVs contained mRNAs of more than 14,000 genes, especially that it had the highest ratio (5.82%) and number (821) of highly expressed genes (RPKM > 100) among the five samples. However, K562-MVs embodied only 535 expressed miRNAs and 20 highly expressed miRNAs, which were less than other samples. When comparing the highly expressed genes in all samples (Fig. 1A), samples of MVs and 3W contained the most overlapped genes, which is consistent with their tumor status. There were 235 genes highly expressed in all samples, among them 80 genes encoded ribosomal protein and ribonucleoprotein, and others are enriched in terms “energy metabolism” and “glucose catabolic process.” It is remarkable that there were many highly expressed miRNAs, especially in the 3W sample. Almost all (18/20) of the highly expressed miRNAs in MVs were also highly expressed in the other four samples (Fig. 1A). The top five highly expressed miRNAs (RPM > 1,000) in MVs were hsa-miR-146b-5p, hsa-miR-486-5p, hsa-miR-92a-3p, hsa-miR-182-5p, and hsa-miR-191-5p.

Using the cutoff value described in Materials and Methods, we identified 3,717 DEGs (including 161 TFs) and 143 DEMs in the transformation process. Stage 3 (2W-3W) contained the largest number of DEGs and DEMs (Fig. 1B), suggesting that the biggest difference of cell population occurred during 2 to 3 weeks. Enrichment analysis revealed most of the DEGs in three stages were enriched in "immune response" and "response to stimulus" related terms on Gene Ontology (GO) biologic process (Supplementary Table S3). DEGs in stage 3 were also enriched in "cell-cycle" and "cell death" pathways. By mapping the DEGs into 18 cancer-related pathways or categories (Supplementary Table S4), we obtained 647 cancer-related DEGs, which are likely to play key roles in the transformation. We summarized the expression changes of these cancer-related DEGs through comparing the normal MNCs and 3W leukemia-like samples (Fig. 1C). DEGs in "cell growth" and "energy production" related pathways were mainly upregulated, whereas most DEGs in "apoptosis" and "immune system" pathways were downregulated. Surprisingly, nine of the 11 downregulated genes in the "cell-cycle" pathway were suppressors of the cell cycle. Some signaling pathways, such as TGFβ, Wnt, and PI3K–Akt pathways had equivalent up- and downregulated genes.

On the basis of the expression change in the three stages of the transformation, we classified these DEGs and DEMs into eight classes, respectively, which were DDD, DDU, DUD, DUU, UDD, UDU, UUD, and UUU (D, down and U, up; Supplementary Fig. S2 and Fig. 2). There were 702 genes continuously increased (UUU cluster) in the transformation, and their top enriched functions were “translational initiation” and “cell-cycle process” (Supplementary Table S5). The 406 continuously decreased (DDD) genes were enriched in functional clusters “immune response, response to stimulus” and “cytokine production.” According to the expression change (up or down) in stage 3 of the transformation, we divided the eight miRNA clusters into two groups (Fig. 2A and B). Many miRNAs upregulated in stage 3 were reported as oncomiRs. Especially, the highest expressed miRNAs in the UUU cluster were all oncomiRs, including miR-17/18a/20a/92a/378a/130b, which may promote cell proliferation and migration in the transformation. On the contrary, many tumor suppressor miRNAs or apoptosis-related miRNAs, such as miR-15/16/181a/30d/30e/26a/142-3p (UDD cluster), were downregulated in the transformation (20). Almost all miRNAs in the DDD cluster, including miR-148a/let-7/199ab/30e, were reported as antitumors or promoted apoptosis, which was consistent with the transformation.

To explore the perturbed mechanisms of cancer pathways, we constructed the regulatory network among miRNAs, TFs, and genes in cancer pathways. Among the 143 and 161 differentially expressed miRNAs and TFs, respectively, we refined 39 miRNAs and 42 TFs for further regulatory analysis based on their target information, expression, and importance to cancer (see Materials and Methods and Supplementary Fig. S3).

Different cancer-related pathways were activated or repressed in each stage of the transformation regulated by miRNAs and TFs (Fig. 3). In the first stage (MNCs-1W), many miRNAs (e.g., miR-26/27/181/146b), as well as TFs in IRF and STAT families, were up-regulated; whereas, members in CEBP family and AP1 complex were downregulated. These regulators were responsible for the downregulation of cell-cycle and mTOR pathways, as well as the upregulation of immune system and NFκB pathway in response to the stimuli of MVs. There were fewer regulators and pathways changed in stage 2 than the other two stages. Stage 3 constitutes a key stage of the transformation, with significant dysregulation of most regulators and pathways. Notably, all miRNAs in the miR-17-92 oncomiR cluster were upregulated, and most of the other miRNAs were downregulated in this stage. TFs, such as TP53, RUNX3, BCL6, and members in the STAT, CEBP, and IRF families were downregulated. Especially, members in the CEBP family were continuously repressed in all three stages of transformation. OncoTFs, such as MYC, MAZ, MYB, and MYBL2 were upregulated. For pathways in stage 3, the immune-related pathways, such as immune system, NFκB signaling, and Notch signaling were repressed; whereas, cell proliferation and energy-related pathways, including the cell-cycle, oxidative phosphorylation, DNA replication, and repair pathways were activated.

Next, we investigated the detailed regulations among TFs, miRNAs, and genes in the cell-cycle, DNA replication/repair, and Notch pathways.

We obtained 441 and 258 pairs of regulatory interactions for the cell-cycle and DNA replication/repair pathways, respectively (Supplementary Table S6). These two pathways are related to the proliferation of cancer cells and were upregulated during the transformation (Fig. 3). Interestingly, we observed that most of the miRNAs in the cell-cycle and all miRNAs in the DNA replication/repair pathways were downregulated in the 3W sample (Fig. 4A and B). Tumor repressor miRNAs miR-15a/16-5p, miR-155-5p, miR-26a/b-5p, and miR-24-3p regulated many genes of these two pathways, and some of the regulations were verified by experiments. The upregulation of miR-17-92, miR-130b-3p, and miR-301b repressed the expression of cell-cycle repressors RBL2, CDKN1A, ATM, and GADD45B. The upregulation of TFs (e.g., MYC, MYB, KLF1, and GATA1/2) and downregulation of other TFs (e.g., KLF6, RUNX3, TP53, and CEBPB) may lead to the activation of many DEGs in these two pathways. In addition, the TFs and miRNAs also regulated each other, such as MYC activated the expression of miR-17-92, miR-150-5p, and miR-15a-5p repressed MYB, and then combined to regulate the same genes to achieve subtle regulations.

Because the Notch signaling pathway plays a critical role in cell-fate determination and has a complex function in leukemia, we were especially interested in its regulation (21). We obtained 67 pairs of regulatory interactions among 15 TFs, 11 miRNAs, and six genes in the Notch signaling pathway, and constructed the TF and miRNA coregulatory network (Fig. 4C). The upregulation of miRNAs, such as miR-146b-5p and miR-125a/b-5p, and the downregulation of TFs, such as KLF6 and CEBPB, may be responsible for the downregulation of the Notch signaling pathway. Interestingly, we noticed that almost all members in the miR-17-92 cluster did not target the genes in the Notch signaling pathway. In this network, miR-146b-5p targeted NOTCH2, NUMB, and LFNG, and was targeted by TFs STAT5A and LEF1. As miR-146b-5p had the highest expression level on average and increased dramatically in the transformation, we highlighted it and constructed an miR-146b-5p–specific regulatory network (Fig. 4D). As a result, we found that miR-146b-5p regulated many other cancer-related genes combined with several TFs. Especially, genes BRCA1, IRAK1, and NFKB1 were confirmed targets of miR-146b-5p by previous studies (22, 23).

Apart from miR-146b-5p, we were also interested in the special regulatory network of other TFs and miRNAs that were highly expressed in MVs (Supplementary Fig. S4A). All of them were verified to be crucial to the occurrence of leukemia and other cancers. YBX1, as a highly expressed TF in K562-MVs, was upregulated constantly in the transformation process. Many studies verified that YBX1 promotes cell proliferation through activating the expression of positive regulatory genes in the cell cycle and repressing the expression of antiapoptotic genes (24). MYC and STAT5A could also activate the expression of genes in the cell-cycle, PI3K–Akt, and apoptosis pathways. All miRNAs in the miR-17-92 cluster were increased constantly in the transformation. Their overexpression is critical to the progress of multiple kinds of leukemia (25–27) through reducing the expression of tumor-suppressor genes and genes in apoptosis and cell-cycle pathways. These highly expressed TFs and miRNAs in MVs could also promote the transformation of MNCs through activating the onco-miRNAs/TFs and repressing the antitumor regulators in a cascading manner (Supplementary Fig. S4B).

According to the bioinformatics analysis, experiments were performed to verify the role of miR-146b-5p in the transformation. A significantly increased level of miR-146b-5p was observed in K562-MVs after transfecting miR-146b-5p mimics into the K562 cells, whereas a decrease could be found when transfected with miR-146b-5p inhibitor (Supplementary Fig. S5A and S5B). However, no dose-dependent effect was observed, and it was difficult to elevate the level of miR-146b-5p in the MVs by transfecting more mimics into the K562 cells (Supplementary Fig. S5C). To determine whether MVs-associated miR-146b-5p was functional for the transformation capability, we added K562-MVs with different levels of miR-146b-5p (normal K562, miR-146b-5p mimics, and miR-146b-5p inhibitors) into the recipient cells, taking K562-MVs as a control. We found that a high level of miR-146b-5p in K562-MVs could accelerate the transformation process, with an average of 9 days (P < 0.05, Table 1). Whereas, a decreased level of miR-146b-5p did not stop the transformation, but resulted in a 2-day short delay, leading to a transformation spanning 15 days (P > 0.05). To investigate whether BCR-ABL1 or miR-146b-5p is more important to the transformation, we incubated the MNCs with imatinib (0.25 and 0.5 μm/mL) and K562-MV with elevated miR-146b-5p. As a result, we discovered that no sign of transformation was observed when the MNCs were treated at both concentration of imatinib for 29 days. This indicates that BCR-ABL1 is an essential molecule for the transformation and miR-146b-5p might act as an acceleration factor with BCR-ABL1.

Furthermore, we tried to explore the mechanism of miR-146b-5p through its potential target genes. Among all the predicted target genes of miR-146b-5p, we considered the tumor-suppressor gene NUMB as a particularly promising candidate because of its importance in cell-fate determination and CML progression (28). Lower levels of NUMB in the target cells could be observed when miR-146b-5p was elevated in the K562-MVs (Fig. 5A and B). The luciferase reporter assay verified that miR-146b-5p could directly bind to the 3′-UTR of NUMB mRNA (Fig. 5C and D). The fluorescence intensity of NUMB 3′-UTR with miR-146b-5p transfection was only 75.89% of the negative control, whereas the repression disappeared when the binding site of miR-146b-5p on NUMB 3′-UTR was mutated. To prove that miR-146b-5p accelerated the transformation process via silencing the tumor-suppressor gene NUMB, we transfected the recipient cells with NUMB lentivirus. Elevated levels could be detected in target cells, although the efficacy of the transfection was approximately 15% (Supplementary Fig. S6A and S6B). Incubating with K562-MVs, a significant decay was observed when the target cells were transfected with NUMB (with an average of 30 days, P < 0.05, Table 1).

Genomic instability is an important hallmark of cancer, and it was increased in the malignant transformation of MNCs induced by MVs described in our previous work (7). Here, we also assessed whether the genomic instability of recipient cells was enhanced by the high level of miR-146b-5p. We found DNA breakage in the vast majority of recipient cells incubated with K562-MVs during transformation. There was a significant increase in DNA breakage when K562-MVs transfected with miR-146b-5p mimics were added to the recipient cells (Fig. 5E), especially at d6 (day 6), d9, and d14. A decrease in the DNA break was also detected by a gray value in the recipient cells when NUMB was elevated. However, it was not significant (Fig. 5F). Our previous work demonstrated that activation-induced cytidine deaminase (AICDA) and ROS might be associated with genomic instability induced by MVs. During transformation, AICDA mRNA and protein expression was increased in cells incubated with K562-MVs (Fig. 5G and H). The induction might be associated with the transfer of miR-146b-5p, as the level of AICDA increased when K562 were treated with miR-146b-5p mimics. Similar to AICDA, there was a significant increase of the ROS level when K562-MVs were treated with miR-146b-5p mimics (Fig. 5I).

Recent studies suggest that MVs within the tumor microenvironment are emerging as potent mediators for the communication of tumor cells (29, 30). MVs could act as an early warning indicator and a novel tool to detect and prevent the occurrence of cancers (31). Here, we showed that K562-MVs contained a huge amount of mRNAs and miRNAs, which are indispensable to the transformation of MNCs to leukemia-like cells. This transformation showed great implications for the process of CML BC, and provided further clues to the occurrence of donor cell leukemia (7). In addition, the process of malignant transformation might also serve as a rapid, convenient, and operable model for the investigation of leukemogenesis (7). In this work, we explored the potential molecules and mechanisms in the transformation by sequencing and bioinformatics analysis, which are BCR–ABL1 combined with other TFs and miRNAs, especially miR-146b-5p, to cause cell proliferation and genome instability.

The expression of thousands of genes and miRNAs was changed significantly during the transformation of MNCs to leukemia-like cells by MVs (Fig. 1B). At the starting stage (stage 1) of the transformation, many antitumor miRNAs were upregulated, and the immune response pathway was initiated (Fig. 3). Thus, this stage could be investigated for the suppression mechanisms of antitumor elements involved in the tumorigenesis. At stage 2, the immune response pathway began to downregulate, and the cell-cycle pathway was activated. At stage 3, all antitumor miRNAs were decreased and oncomiRs were upregulated as well as genes in the cell-cycle, DNA replication, and energy metabolism pathways. It has been reported that many miRNAs (miR-15a, miR-16, and miR-146a etc.) were at low levels in CML-AP patients compared with CML-CP (32). Our data also confirmed that these miRNAs were downregulated in stage 3 (Fig. 2). The expression tendency of these genes and miRNAs in the transformation provided evidence that it is a process of tumorigenesis and is similar to the process of CML CP to BC. Moreover, these results might serve as a resource for the process of transformation from normal cells to leukemia. The differentially expressed TFs and miRNAs coregulated many genes of the cell cycle, and DNA replication/repair, causing the activation of these pathways (Fig. 4A and B). The Notch pathway is important in the development of hematopoietic cells and leukemia progress (21). The TF and miRNA coregulatory network for the Notch pathway highlighted miR-146b-5p because it was predicted to be regulated by TFs STAT5A and LEF1, and regulate genes NUMB, NOTCH2, etc., which are key genes in leukemia and CML BC (Fig. 4C; refs. 28, 33). On the other hand, miR-146b-5p is the most highly expressed miRNA in five samples on average and the fourth highest in K562-MVs. It is also highly expressed in The Cancer Genome Atlas acute myelogenous leukemia samples (34) and had a higher expression in acute lymphoblastic leukemia than its controls, according to the HMED database (35).

During the transformation, we detected BCR–ABL1 fusion gene in the K562-MVs and 3W samples with dozens of reads mapping to the breakpoint site from RNA-seq data, especially in the 3W sample. BCR-ABL1 has diverse effects on DNA damage-response/DNA repair, checkpoint activation, proliferation, and apoptosis for the progression of CML CP to BC (36, 37). Although BCR–ABL1 fusion mRNA was essential to this transformation, other RNAs were also required, such as STAT5 (38). We also proved that MVs lost their transformative abilities following RNase treatment (7). Numerous studies have demonstrated that miRNA was one of the most promising and key regulatory molecules in MVs (39). miRNAs in MVs can be transferred into the recipient cells and function to reprogram the target cell transcriptome (40). Our expression and regulatory network analysis indicated that miR-146b-5p might play important roles in the transformation, and be activated by STATs, which were induced by BCR-ABL1 (Fig. 6; ref. 41). miR-146 contains two copies, miR-146a and miR-146b, located at different chromosomes, which do not seem to be redundant because of their different expression patterns (Fig. 2; refs. 42, 43). MiR-146a plays functional roles in hematopoiesis and innate immune responses, as a tumor suppressor in many solid tumors (44, 45). MiR-146b was found to function in thyroid cancer and glioma by targeting the TGFβ pathway and NFκB, and also targeted by TF STATs (41, 46, 47). Our work demonstrated that increasing miR-146b-5p in K562-MVs could accelerate the transformation (Table 1). However, the MNCs were unable to be transformed to leukemia-like cells when incubated with both imatinib and miR-146b-5p-elevated MVs. We think that miR-146b-5p might act as an acceleration factor with the essential molecule BCR-ABL1 in the transformation.

Furthermore, how did miR-146b-5p accelerate the transformation? We predicted that NUMB was a target of miR-146b-5p and also confirmed this by luciferase assay (Fig. 5C and D). Ito and colleagues (28) found that keeping NUMB at low levels may be essential for maintaining cells at an immature state and trigger CML BC transformation. Similar to the CML BC, decreased level of NUMB gene was critical to our transformation system induced by MVs. Increased NUMB in recipient cells by lentivirus delayed the transformation significantly, although transfecting with lentivirus only can also delay the transformation for the toxicity of the virus (Table 1). This indicates that NUMB might serve as a brake mechanism of the transformation. Thus, we could further authenticate the idea that increased miR-146b-5p in K562-MVs promoted the transformation via silencing the NUMB gene in the recipient cells. NUMB's effects on leukemia cell growth partially depend on p53 by preventing ubiquitination and degradation of p53 (28). NUMB is also an inhibitor of the Notch signaling pathway to control cell-fate and inhibit tumor cell proliferation (48). Thus, the increased miR-146b-5p promotes the transformation significantly through repressing the function of NUMB. However, decreased miR-146b-5p showed limited impact on the transformation. The mismatch might be explained by the fact that MVs are packages of a large number of bioactive molecules that contain not only contributing factors but also detractors. This provides a novel insight that the occurrence of leukemia induced by MV could be regulated by content intervention. On the other hand, overexpression of miR-146b-5p could elevate the level of AICDA and ROS, consistent with DNA break in the target cells (Fig. 5E and G-I). BRCA1 is a tumor-repressor gene involved in the cell-cycle and DNA damage repair pathways, and its deficiency will cause genome instability (49). It was reported that miR-146b-5p could target and inhibit BRCA1 accompanied by increased proliferation (22). Consequently, BRCA1 may be another way that miR-146b-5p leads to genomic instability in the transformation. We also predicted that miR-146b-5p targeted NOTCH2, a tumor-suppressor in myeloid leukemia (33), to inhibit the Notch pathway. As a result, we inferred that miR-146b-5p served as an oncomiR to promote proliferation and increase genome instability through its targets and downstream pathways in the transformation (Fig. 6).

In summary, using expression and regulatory network analysis, we explored the potential transformation mechanism for our previous model that K562-MVs transforming MNCs to leukemia-like cells. We identified that miR-146b-5p, as a downstream regulator of BCR–ABL1, promoted the transformation by targeting several important genes (NUMB, NOTCH2, BRCA1 etc.) to affect cell proliferation and genome instability. Our study provided an insight into the CML BC transformation and donor cell leukemia, as well as an opportunity for studying the changes of normal cells to cancer-like cells.

No potential conflicts of interest were disclosed.

Conception and design: H.-M. Zhang, Z. Chen, A.-Y. Guo

Development of methodology: Q. Li, H. Hu, T. Liu, Q. Li

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Q. Li, X. Zhu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H.-M. Zhang, Q. Li, X. Zhu, W. Liu, A.-Y. Guo

Writing, review, and/or revision of the manuscript: H.-M. Zhang, X. Zhu, A.-Y. Guo

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H.-M. Zhang, Q. Li, F. Cheng, Y. You, Z. Zhong, Z. Chen

Study supervision: P. Zou, Z. Chen, A.-Y. Guo

The work was supported by grants from the National Natural Science Foundation of China (NSFC; 31270885 and 31471247 to A.Y. Guo, 81470330 to Z. Chen, 81470348 to P. Zou, and 81470333 to Y. You), National Basic Research Program of China (973 Program, 2012CB932501, and the Program for New Century Excellent Talents in University (NCET; A.Y. Guo).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.