YBX1 is required for maintaining myeloid leukemia cell survival by regulating BCL2 stability in an m⁶A-dependent manner

Eukaryotic cells regulate messenger RNA (mRNA) expression through transcriptional and posttranscriptional mechanisms. In these pro-cesses, RNA-binding proteins (RBPs) play essential roles by fine-tuning gene expression, RNA splicing, polyadenylation, stability, localization, translation, and degradation.¹  RBPs typically bind RNA via 1 or several RNA-binding domains (RBDs) such as the RNA recognition motif, DEAD box domain, or K-homology domain.^1,2  Recent advances uncovered additional RBPs that lack conventional RBDs.³  Alterations of RBPs have been implicated in different human cancers; thus, it is necessary to decipher the complicated regulatory mechanisms of RBPs and their cancer-related RNA targets, which will provide a better understanding of cancer biology and unveil potential therapeutic targets.⁴ 

Acute myeloid leukemia (AML) is an aggressive hematologic malignancy characterized by uncontrolled expansion of poorly differentiated myeloid cells.⁵  Accumulated evidence reveals that RBPs are essential for normal hematopoiesis and hematopoietic malignancies.^6-8  For instance, frequent mutations in RBPs that serve as splicing factors have been identified in myeloid dysplasia.^9,10 N⁶-methyladenosine (m⁶A) is a modification commonly found in mammalian mRNA,¹¹  and it plays a critical role in determining the fate of RNA.^12,13  This modification is reversible and is catalyzed by a methyltransferase complex (METTL3-METTL14-WTAP) and demethylases ALKBH5 and FTO.^14-16  Several key m⁶A regulators have emerged as critical players in normal and malignant hematopoiesis.^17-26  Our recent studies^27,28  reveal that the RNA m⁶A demethylase ALKBH5 is selectively required for maintaining the function of AML leukemic stem cells but not normal hematopoietic stem cells (HSCs). Despite these recent advances, it is still necessary to explore RBPs that are essential to leukemia maintenance and survival, which will contribute to understanding the pathogenesis of AML development.

YBX1 belongs to the RBP family and is a multifunctional protein that contains the evolutionarily conserved cold-shock domain (CSD). YBX1 has been implicated in various biological pro-cesses because it affects a wide range of genes involved in cell proliferation, survival, drug resistance, and chromatin destabilization.^29,30  YBX1 can also act as a versatile oncoprotein and is required for tumor cell proliferation, progression, and metastasis.^31,32  However, the role of YBX1 in leukemia remains elusive. Here, we show that YBX1 regulates m⁶A-tagged mRNA stability by interacting with insulin-like growth factor 2 mRNA-binding proteins (IGF2BPs) and is selectively required for maintaining the survival of myeloid leukemia cells.

C57BL/6J (CD45.2) background Ybx1^fl/fl and NOD.Cg-Prkdc^scidIL2rg^null (B-NSG) mice were obtained from Biocytogen. Mx1-cre mice and B6.SJL (CD45.1) were purchased from The Jackson Laboratory. All of the animal experiments were carried out according to the protocols approved by Animal Care and Use Committee of the Medical Research Institute, Wuhan University.

Patients with AML provided informed consent for collecting samples from bone marrow (BM) aspirations. All experiments that refer to human samples were conducted in compliance with all relevant ethical regulations and were approved by the ethics committees of medical research units of the various universities. Experimental details are provided in the supplemental Data, available at the Blood Web site.

BM cells were extracted from 8- to 12-week-old Ybx1^cKO and wild-type (WT) mice, and lineage-negative cells (Lin^–) were enriched and used for generating MLL-AF9 AML mice. Experimental details are provided in the supplemental Data.

Colony formation assay, competitive repopulation assays, cell proliferation, cell cycle differentiation and apoptosis analyses, cell staining, and flow cytometry analyses were performed. We performed RNA pull-down, RNA immunoprecipitation (RIP), protein expression and purification, western blotting, co-immunoprecipitation, RNA decay, luciferase reporter assays, RIP-polymerase chain reaction (RIP-PCR), quantitative reverse transcriptase-PCR (qRT-PCR), methylated RIP-PCR (MeRIP-PCR), RNA sequencing (RNA-seq), m⁶A sequencing (m⁶A-seq), and thiol(SH)-linked alkylation for the metabolic sequencing of RNA (SLAM-seq). Experimental details are provided in the supplemental Data.

To comprehensively investigate the expression of RBPs in AML, we first analyzed the expression levels of 1068 RBP mRNAs in patients with AML by surveying the publicly available data sets from The Cancer Genome Atlas (TCGA). Interestingly, we found that YBX1 had higher expression in patients with AML and ranked among the top candidates in all RBPs (Figure 1A). Moreover, YBX1 mRNA expression in human AML samples was significantly higher than that in other cancer types (Figure 1B). We also found that YBX1 was expressed at a significantly higher level in leukemia cells from de novo patients with AML, includ-ing patients with normal karyotype, inv(16), MLL-rearranged t(11q23), and t(8,21), than in healthy controls (Figure 1C). And YBX1 was also aberrantly elevated at the protein level in samples from AML patients compared with those from healthy control cells (Figure 1D). Similar higher YBX1 protein levels were detected in various human leukemia cell lines (Figure 1E). Together, these data suggest that YBX1 might play a critical role in myeloid leukemia.

To test the functional role of YBX1 in leukemic cells, we first knocked down YBX1 using 2 different short hairpin RNAs (shRNAs [shYBX1#1 and shYBX1#2]). Both shRNAs markedly deleted YBX1 expression at the protein and mRNA levels (Figure 1F; supplemental Figure 1A). We transduced Lin^–CD34⁺ cells derived from patients with AML, which are commonly considered a leukemia stem and progenitor cell–enriched population, with lentiviral shYBX1. As expected, YBX1 knockdown (KD) significantly reduced clonogenic capacity and impaired the in vivo leukemia reconstitution ability of these cells (Figure 1G-H), indicating that YBX1 is required for maintaining the function of primary leukemia stem and progenitor cells derived from patients with AML. We next explored the role of YBX1 in human myeloid leukemia cell lines. YBX1 KD resulted in a substantial inhibition of cell proliferation and clonogenicity in both MOLM-13 and THP-1 AML cells (Figure 1I-J). In addition, YBX1 KD induced differentiation, moderately affected cell cycles, and significantly promoted apoptosis of leukemia cells that showed a higher percentage of annexin V⁺ cells and increased cleaved caspase 3 of leukemia cells upon deletion of YBX1 (supplemental Figure 1B-G).

To rule out an off-target possibility, we restored YBX1 expression by inserting its complementary DNA, which was resistant to shYBX1 3′ targeting of the 3′ untranslated region (UTR) (supplemental Figure 1H). As expected, restoration of YBX1 completely rescued the defects in cellular growth, clonogenic ability, cell cycle arrest, and apoptosis caused by YBX1 KD (Figure 1K-N), indicating that the inhibitory function of YBX1 deficiency was not caused by an shRNA off-target effect. In addition, YBX1 KD cells exhibited delayed leukemia development, which was also reversed by restoring the expression of WT YBX1 (Figure 1O). Together, these data indicate that YBX1 is required for maintaining survival of human myeloid leukemia cells.

We next examined whether Ybx1 is required for murine leukemogenesis. We first knocked down Ybx1 in primary leukemia cells from MLL-AF9–induced AML mice (Figure 2A). The efficiency of Ybx1 KD was confirmed at the protein and mRNA levels (Figure 2B; supplemental Figure 2A). As expected, Ybx1 KD inhibited the proliferation and clonogenic capacity of leukemia cells (Figure 2C-E), moderately affected the cell cycle, and substantially induced apoptosis (Figure 2F-G; supplemental Figure 2B-C). Ybx1 KD leukemia cells were then transplanted into sublethally irradiated recipient mice. We found that recipients of Ybx1 KD leukemia cells developed AML and died as a result of AML significantly more slowly than did mice in the control group. This delayed AML development also correlated with a lower percentage of leukemia cells in peripheral blood (PB) and less severe splenomegaly (Figure 2H-J; supplemental Figure 2D-F). Together, these data indicate that Ybx1 is essential for murine AML development.

Next, we deleted Ybx1 in murine leukemia cells using CRISPR/Cas9-mediated knockout (KO) by 2 different guide RNAs (supplemental Figure 2G). Deletion of Ybx1 significantly reduced the clonogenic ability of leukemia cells (supplemental Figure 2H-I). We further generated Ybx1^fl/fl mice and crossed them with the Mx1-Cre strain to get Mx1-cre;Ybx1^fl/fl mice (supplemental Figure 3A). Eight- to 12-week-old Mx1-cre;Ybx1^fl/fl and Ybx1^fl/fl control mice were treated with polyinosinic-polycytidylic acid (pIpC) 3 times by intraperitoneal injection every other day. Efficient deletion of Ybx1 was observed at 2 and 4 weeks after the last injection of pIpC (supplemental Figure 3B-D). Because Mx1-Cre is also active in several major organs, we examined the histopathology 4 weeks after treatment with pIpC and did not detect any obvious defects in heart, liver, lung, kidney, or spleen tissue (supplemental Figure 3E). For simplicity, pIpC-treated Mx1-cre;Ybx1^fl/fl mice will be referred to as Ybx1^cKO mice, and pIpC-treated Ybx1^fl/fl control mice will be referred to as WT mice. We found that loss of Ybx1 significantly delayed the development of AML, and recipients of MLL-AF9–transduced Ybx1^cKO cells displayed markedly longer latency than did recipients of MLL-AF9–transduced WT Lin^– cells (Figure 2K-L). This was also confirmed by a colony formation assay that showed fewer colonies in the absence of Ybx1 (Figure 2M-N). Taken together, our data indicate that Ybx1 is required for murine leukemogenesis.

Next, we decided to determine whether YBX1 is required for normal hematopoiesis. We analyzed our in-house RNA-seq data sets (GSE165863) for 16 distinct hematopoietic populations from murine BM, but we did not find any significant difference in Ybx1 expression in these cells, which was further validated by qRT-PCR (supplemental Figure 3F-G). Compared with WT mice, Ybx1^cKO mice showed comparable blood counts and frequencies of different types of blood cells in PB, including white blood cells, lymphoma cells, granulocytes, monocytes, platelets, and red blood cells, at 4 weeks after the last pIpC injection (Figure 3A-D; supplemental Figure 3H). The percentages and total numbers of long-term HSCs, short-term HSCs, multipotent progenitor population 2 (MPP2), MPP3, MPP4, Lin⁻Sca-1⁺c-Kit⁺(LSK) cells, lymphoid-primed multipotent progenitor (LMPP) cells, common myeloid progenitor (CMP), granulocyte/macrophage progenitor (GMP), megakaryocyte/erythroid progenitor (MEP), and common lymphoic progenitor (CLP) in the BM of Ybx1^cKO mice were also similar to those in WT mice (Figure 3E-G; supplemental Figure 3I). Thus, these data indicate that Ybx1 is dispensable for normal hematopoiesis.

We next investigated the impact of Ybx1 deletion on HSC function. A colony formation assay showed normal clonogenic ability and differentiation potential of Ybx1^cKO stem and progenitor cells (Figure 3H-I). To determine whether Ybx1 affects the self-renewal capacity of HSCs, we performed a competitive repopulation assay. We found that the percentages of WT and Ybx1^cKO donor-derived total cells (CD45.2⁺), myeloid cells (Mac1⁺, Gr-1⁺), and T cells (CD3e⁺) in PB were comparable over 4 to 16 weeks after BM transplantation (Figure 3J). Interestingly, Ybx1 deficiency impaired B-cell reconstitution (Figure 3J), which needs to be studied in the future. At 16 weeks, we did not observe significant differences in the percentages of donor-derived WT and Ybx1^cKO stem and progenitor cells (long-term HSCs, short-term HSCs, MPP2, MPP3, MPP4, LSK, LMPP, progenitor cells, CMP, GMP, MEP, and CLP) (Figure 3K-L; supplemental Figure 3J-K). Collectively, these results indicate that loss of Ybx1 does not affect normal hematopoiesis.

YBX1 has been shown to bind and stabilize RNA.³³  To profile YBX1 binding RNA at the transcriptome level, we overexpressed FLAG-YBX1 in HEK293T cells and performed RIP after high-throughput sequencing (RIP-seq). We identified 9728 high-confidence binding sites within 2898 transcripts, which were highly enriched in promoter, exon, and 3′ UTRs (Figure 4A; supplemental Figure 4A). The majority of YBX1-binding RNAs (79.34%) were protein-encoding RNAs (supplemental Figure 4B). Gene ontology (GO) analysis showed that YBX1-binding transcripts were enriched in mRNA processing, splicing, and metabolic processes (supplemental Figure 4C). Surprisingly, we found that YBX1-binding sites clearly coincided with the m⁶A sites (Figure 4B; supplemental Figure 4D-E), which drove us to examine whether YBX1 binds to m⁶A-tagged RNAs. We conducted an RNA pull-down assay that used methylated single-stranded RNA bait (ss-m⁶A oligo with the consensus sequence GG(m⁶A)CU). Unmethylated control RNA (ss-A) was used as the control (Figure 4C). Interestingly, similar to positive control IGF2BPs, YBX1 from whole-cell nuclear lysate was pulled down by ss-m⁶A oligo but not ss-A oligo (Figure 4D), suggesting that YBX1 could bind to m⁶A-tagged RNAs. To assess whether YBX1 directly or indirectly binds to ss-m⁶A oligo, we purified glutathione S-transferase (GST)-YBX1 and used it in the RNA pull-down assay. Unlike IGF2BPs, purified YBX1 was not immunoprecipitated by ss-m⁶A (Figure 4E), implying that YBX1 indirectly binds to m⁶A-tagged RNAs.

In line with a previous study,³⁴  we found that YBX1 interacted with insulin-like growth factor 2 mRNA-binding proteins 1, 2, and 3 (IGF2BP1/2/3) (supplemental Figure 4F). Interactions between endogenous YBX1 and IGF2BP1/3 were observed (Figure 4F), and we also found that the YBX1 CSD mediated the interaction with IGF2BPs (Figure 4G-I; supplemental Figure 4G). Given that IGF2BPs are m⁶A readers, we proposed that IGF2BPs mediate the binding of YBX1 with m⁶A-tagged RNAs. To confirm our hypothesis, we knocked down IGF2BPs. An RNA pull-down assay showed that IGF2BP1/3 KD significantly abolished the binding of YBX1 with ss-m⁶A, whereas the YBX1 KD did not affect the binding of IGF2BPs with ss-m⁶A (Figure 4J-K). Thus, these data indicate that IGF2BPs are essential in mediating the binding of YBX1 to m⁶A-tagged RNAs. We further integrated our analysis with data from MeRIP-seq and RIP-seq for IGF2BPs and YBX1. Interestingly, ∼72.18% to 84.95% of the YBX1-bound transcripts were also bound by IGF2BPs (Figure 4L; supplemental Figure 4H), and both YBX1 and IGF2BPs preferentially bind the classic m⁶A motif containing the UGGAC consensus sequence (Figure 4M). As shown, representative high-confidence targets of IGF2BPs,³⁵  including MYC, MARCKLS1, and TK1, were also enriched by YBX1 (Figure 4N; supplemental Figure 4I). Interestingly, we observed strong binding of the BCL2 transcript with YBX1 (Figure 4N). Functionally, we found that decay of MYC, BCL2, MARCKLS1, and TK1 mRNA were accelerated upon KD of YBX1 in HEK293T cells (Figure 4O). Thus, these results indicate that YBX1 regulates m⁶A-tagged mRNA stability, which is mediated by IGF2BPs.

To determine whether the effect of YBX1 on mRNA stability is m⁶A dependent, we chose to study MYC and BCL2. The firefly luciferase (Fluc) reporter that was inserted into the 249-bp WT or the m⁶A site’s mutant coding region instability determinant (CRD) of the 3′ terminus of the MYC coding region was used.³⁵  We also constructed Fluc reporters with an ∼258-bp BCL2 3′ UTR that is enriched with m⁶A modifications or m⁶A site mutants. As expected, ectopic expression of YBX1 induced a significant increase in Fluc activity of the WT MYC and BCL2 reporters. Conversely, this increase was largely impaired by mutations in the m⁶A consensus sites (Figure 4P). Consistent with a previous study,³⁵  as a positive control, the relative Fluc mRNA level of CRD-wt, but not CDR-mutant (CRD-mut), was increased by IGF2BP1 or IGF2BP3 overexpression (supplemental Figure 4J). Moreover, deletion of IGF2BP1 significantly abolished the increase of luciferase activity induced by YBX1 (Figure 4Q), suggesting that IGF2BPs are required for mediating the role of YBX1 on mRNA stability. Next, we knocked down YBX1 to assess whether YBX1 is necessary for the functioning of IGF2BPs. We found that YBX1 KD caused a substantial reduction of luciferase activity induced by IGF2BP1/3 (Figure 4R; supplemental Figure 4K), indicating that YBX1 is also required for IGF2BPs in maintaining m⁶A-tagged mRNA stability.

We next performed crosslinking and immunoprecipitation sequencing (CLIP-seq) for YBX1 in AML cells and compared it with m⁶A-seq data for these cells. We consistently found that ∼57.7% of YBX1-bound transcripts were m⁶A tagged, and ∼78.17% of YBX1-bound targets were bound by IGF2BPs (supplemental Figure 5A-B). To further assess whether YBX1 maintains AML in an m⁶A-dependent manner, we abolished the binding of YBX1 with IGF2BPs. Because it is currently not possible to identify the exact site of YBX1 that binds to IGF2BPs, we focused on the YBX1 CSD because it mediates the interaction with IGF2BPs (Figure 4G-I). Compared with intact YBX1, restoration of YBX1 with deleted CSD (ΔCSD) failed to rescue the phenotypes of YBX1-deficient leukemia cells (supplemental Figure 5C-I), suggesting that CSD is required for YBX1 function in AML cells. In addition, YBX1 can function as a reader of RNA 5-methylcytosine (m⁵C) and stabilize mRNA through the indole ring of W65 in its CSD.^36,37  However, the YBX1-W65A mutant successfully rescued the phenotypes of YBX1-deficient leukemia cells, suggesting that the function of YBX1 as an m⁵C reader is not required for its effect on myeloid leukemia. Meanwhile, YBX1-W65A, but not ΔCSD, restored the expression of BCL2 and MYC (supplemental Figure 5J-K). Collectively, our data demonstrate that YBX1 fine-tunes m⁶A-tagged RNA stability by cooperating with IGF2BPs.

To better understand the role of YBX1 in leukemogenesis, we sorted leukemia-initiating cells from WT and Ybx1^cKO AML mice and compared their gene expression profiles using RNA-seq (Figure 5A). A total of 3631 genes (1578 downregulated and 2053 upregulated) was found to be significantly changed due to Ybx1 loss. GO analysis showed that these genes were enriched in mRNA processing, RNA splicing regulation, myeloid cell differentiation, and regulation of the intrinsic apoptotic signaling pathway (Figure 5B; supplemental Figure 6A). Gene set enrichment analysis also showed the enrichment of genes involved in unfolded protein response and Wilms tumor 1 (WT1)-associated protein (WTAP) downregulated targets in Ybx1^cKO cells (supplemental Figure 6B-D). Consistent with increased apoptosis of leukemia cells due to YBX1 loss, deletion of Ybx1 downregulated prosurvival genes including Myc, Bcl2l1, Mcl1, Tmbim6, and Birc6 (Figure 5C). In contrast, expression of proapoptotic genes such as Bax, Dapk3, and Siva1 was substantially increased in Ybx1^cKO leukemia cells (Figure 5D). Similar changes in apoptotic-related genes were also observed in human myeloid leukemia cells upon YBX1 KD (supplemental Figure 6E). Thus, these results indicate that YBX1 plays an important role in regulating the expression of apoptotic genes in myeloid leukemia cells.

Next, we explored how YBX1 regulates the expression of these genes. Considering our findings,s shown in Figure 4, we focused on m⁶A-mediated mRNA stability. As expected, YBX1 KD did not affect the global m⁶A level (supplemental Figure 6F). To identify transcripts that are methylated in leukemia-initiating cells from AML mice, we mapped the m⁶A methylome using m⁶A-seq, and identified 7547 m⁶A peaks that were primarily enriched in both exons and the 3′ UTR around the stop codon and displayed the classic DRACH (D, A/G/U; R, A or G; H, A/C/U) motif (Figure 5E). In addition, >87% of the m⁶A-tagged transcripts were protein encoding (Figure 5F). The integrative genomics viewer displayed high enrichment of m⁶A peaks in Bcl2l1 and Mcl1 (Figure 5G). Transcripts of Rbm7 and Zfp36l1 were also modified with m⁶A (supplemental Figure 6G-H). As an important subunit of m⁶A methyltransferase complex, WTAP is a novel oncogenic protein for leukemogenesis, and it regulates proliferation of acute myeloid leukemia cells.³⁸  Interestingly, we observed higher m⁶A peaks in a Wtap transcript (Figure 5G), suggesting that Ybx1 might affect Wtap expression in leukemia cells. MeRIP-qPCR further validated these m⁶A modifications on these transcripts (Figure 5H).

Notably, m⁶A modification affects mRNA stability.³⁹  So we measured the transcriptome-wide mRNA decay in leukemia-initiating cells from WT and Ybx1^cKO AML mice using SLAM-seq.⁴⁰  Interestingly, we observed a significant increase of mRNA decay at the global level in Ybx1^cKO leukemia cells (KO, 94.7 minutes vs WT, 130.3 minutes) (Figure 5I). To test the contribution of m⁶A modification to RNA degradation, we integrated m⁶A-seq data with SLAM-seq data and compared the degradation rates of m⁶A-tagged RNAs with non–m⁶A-tagged RNAs. As expected, m⁶A-tagged transcripts in both WT and Ybx1^cKO leukemia cells showed accelerated degradation when compared with non–m⁶A-tagged transcripts. Importantly, the loss of Ybx1 increased the degradation of m⁶A-tagged transcripts (Figure 5J). For instance, mRNA stabilities of Myc and Bcl2l1 were significantly decreased upon Ybx1 deletion (Figure 5K), but deletion of Ybx1 did not regulate the transcription and splicing of its targets in leukemia cells (supplemental Figure 6I-L). Taken together, our data indicate that Ybx1 regulates the expression of apoptotic genes in leukemia cells by affecting their mRNA stabilities.

We observed significant enrichment of MYC targets and genes of oxidative phosphorylation in YBX1^high cases from 2 AML patient cohorts (Figure 6A-C). BCL2 is involved in oxidative phosphorylation of AML stem cells,⁴¹  so we focused on c-MYC and BCL2. Downregulation of Myc and Bcl2 in Ybx1^cKO leukemia cells from AML mice was confirmed (Figure 6D). YBX1 KD in human leukemia MOLM-13 and THP-1 cells also caused marked reduction of MYC and BCL2 expression (Figure 6E-F; supplemental Figure 7A). We also observed an interaction of YBX1 with IGF2BP1/3 in leukemia cells (supplemental Figure 7B). We investigated whether YBX1 regulates BCL2 and MYC expression by modulating their mRNA stability in leukemia cells. Indeed, loss of Ybx1 caused an obvious decrease in the half-life of BCL2 and MYC mRNA (Figure 6G). Similar results were also found in human leukemia cells upon YBX1 KD (supplemental Figure 7C). From our m⁶A-seq data, we observed higher enrichment of m⁶A peaks in Myc and Bcl2 transcripts in leukemia cells, which was validated by MeRIP-qPCR (Figure 6H-I). YBX1 RIP-PCR showed that YBX1 directly binds BCL2 and MYC transcripts, which was impaired with knockdown of IGF2BPs (Figure 6J; supplemental Figure 7D). Therefore, our data indicate that YBX1 stabilizes BCL2 and MYC mRNA in an m⁶A-dependent manner.

Next, we examined whether MYC and BCL2 mediate the function of YBX1 in leukemia cells. We re-expressed either MYC or BCL2 in primary mouse Ybx1^cKO leukemia cells and YBX1 KD human leukemia cells, and we found that ectopic expression of MYC or BCL2 substantially rescued the clonogenic and proliferative defects resulting from YBX1 deficiency (Figure 6K-N; supplemental Figure 7E). Importantly, ectopic expression of MYC or BCL2 also prevented the induction of apoptosis and restored the cell cycle in YBX1 KD leukemia cells (Figure 6O-P; supplemental Figure 7F-G). Finally, by using the significantly downregulated targets in Ybx1^cKO leukemia cells as a gene cluster, we surveyed their expression in AML patients from the TCGA database. We found that the expression levels of 420 mRNAs were positively correlated with YBX1 expression in patients with AML, and higher expression of these 420 signature genes correlated with their shorter overall survival (supplemental Figure 7H-J). Taken together, these data indicate that BCL2 and MYC are functional downstream targets of YBX1.

Our study demonstrates that YBX1, as an RNA-binding protein, is selectively required for maintaining myeloid leukemia cell survival but is dispensable for normal hematopoiesis. This function of YBX1 is accomplished by stabilizing m⁶A-tagged mRNA (including BCL2 and MYC) by interacting with m⁶A reader IGF2BPs (Figure 6Q). Thus, our findings shed light on the underlying mechanism that controls survival of myeloid leukemia cells.

Although dysregulated expression of YBX1 has been observed in several human solid cancers,^42-45  to our knowledge, this study provides the first clear evidence for the role of YBX1 in hematopoiesis and leukemogenesis. Surprisingly, we did not observe any significant changes in hematopoietic stem and progenitor cells in Ybx1^cKO mice. Although a previous study using an EML cell line implied that YBX1 may function in the early stage of erythropoiesis,⁴⁶  we did not detect any obvious defects in erythroid development, but we did show that deletion of YBX1 impairs leukemogenesis by inducing apoptosis and inhibiting proliferation of myeloid leukemia cells. A new study⁴⁷  showed that YBX1 mediates persistence of JAK2-mutated myeloproliferative neoplasm cells, and the authors also found that deletion of YBX1 does not perturb normal hematopoiesis and the function of hematopoietic stem and progenitor cells. Therefore, our data have uncovered a unique and critical function of YBX1 in myeloid leukemia.

YBX1 is a multifunctional protein and is involved in regulating both transcription and translation.⁴⁸  Surprisingly, our study revealed an important role of YBX1 in stabilizing m⁶A-tagged RNA by interacting with m⁶A reader IGF2BPs. We found that YBX1 is not an m⁶A reader, suggesting that YBX1 acts as either a stabilizer or a scaffold for recruiting other proteins. YBX1 also interacts with other m⁶A readers such as YTHDF2 and YTHDC1/2 (data not shown), suggesting that it might affect mRNA fates in multiple ways. Thus, the exact mechanism of how YBX1 works in regulating m⁶A-tagged RNA may be context dependent. In addition, YBX1 functions as a reader for another RNA modification, m⁵C, and maintains the stability of its target mRNA in the pathogenesis of bladder cancer and zebrafish embryogenesis.^36,37  Given the multifaceted roles of YBX1 in regulating cotranscription and posttranscription, YBX1 may act in distinct manners on transcripts with different modifications. Thus, it will be important to further investigate the underlying mechanisms of YBX1 in different contexts.

Our findings establish a link between YBX1 and BCL2 expression. BCL2 is found to be upregulated in AML cells and leukemia stem cells,^41,49  and inhibiting BCL2 via a selective BH3 mimetic such as venetoclax has proven to be an efficient and successful strategy in eradicating leukemic stem cells in clinics.⁵⁰  Our study provides evidence that m⁶A modification is essential in maintaining the stability of BCL2 mRNA in myeloid leukemia cells, which is accomplished by YBX1. Modification of m⁶A could play multiple roles in determining BCL2 mRNA fate, and a recent study indicates that m⁶A promotes the translation of MYC and BCL2 mRNA in myeloid leukemia cells.¹⁸  Collectively, our results reveal novel mechanistic insights into how YBX1 selectively functions in regulating survival of myeloid leukemia cells and suggest a potential therapeutic strategy for treating myeloid leukemia.

The authors acknowledge the members of our laboratory for helpful discussion.

This work is supported by grants from the National Key Research and Development Program of China (2017YFA0505600) (H.Z.), the National Natural Science Foundation of China (81870124 and 81722003), the Wuhan Science and Technology Program for Application and Basic Research Project (2018060401011325), the Natural Science Foundation of Hubei Province (2019CFA073), the Hubei Provincial Natural Science Foundation for Creative Research Group (2018CFA018), and the Medical Science Advancement Program (Basic Medical Sciences) of Wuhan University (TFJC2018005). This work was also supported by a grant from the National Postdoctoral Fellowship (81500133) (J. Chang).

Contribution: M.F., X.X., G.H., and H.Z. designed the experiments; M.F. and H.Z. contributed to the experimental plan and data interpretation; G.H. and Yicun Li performed bioinformatic analyses with help from J. Chang and Y.C. and with conceptual input from H.Z.; M.F. performed growth curve and colony formation assays, qPCR, and western blotting on primary samples or cell lines with help from Tong Zhang and Tiantian Zhang; G.H. performed mouse experiments with help from M.F., K.G., and M.C.; M.F. and Tiantian Zhang performed plasmid constructions with help from J.W. and P.W.; M.F. constructed the RNA-sequencing library; X.X. performed the SLAM-seq with help from M.F., R.Y., and Q.W.; Yashu Li constructed the m⁶A-sequencing library with help from J.H., Z.G., and C.G.; J. Chai and W.L. performed hematoxylin and eosin staining; S.L. and J. Chen helped with or advised on experiments; L.L. and F.Z. provided AML samples, clinical data, and materials; H.Z. and M.F. designed and supervised the overall study; and M.F., X.X., G.H., and H.Z. performed statistical analysis and wrote the manuscript.

Conflict-of-interest disclosure: J. Chen is a scientific founder of and holds equities with Genovel Biotech Corporation. The remaining authors declare no competing financial interests.

Correspondence: Haojian Zhang, Frontier Science Center for Immunology and Metabolism, Medical Research Institute, School of Medicine, Wuhan University, No.185 East Lake Rd, Wuchang District, Wuhan, Hubei 430071, People’s Republic of China; e-mail: haojian_zhang@whu.edu.cn.