DDX6 undergoes phase separation to modulate metabolic plasticity and chemoresistance

Bi, Hongjie; Li, Wei; Ren, Lili; Zhang, Honghai; Dong, Lei; Chan, Anthony; Wang, Xueer; Yang, Lu; Zhang, Xiaoxu; Xue, Meilin; Qin, Hanjun; Wu, Xiwei; Hoeve-Scott, Johanna Ten; Zhang, Bin; Li, Ling; Wunderlich, Mark; Mulloy, James C.; Rosen, Steven T.; Chen, Jianjun; Li, Xiaobo; Chen, Chun-Wei; Su, Rui

doi:10.1038/s41467-025-66966-4

Download PDF

Article
Open access
Published: 02 December 2025

DDX6 undergoes phase separation to modulate metabolic plasticity and chemoresistance

Nature Communications volume 17, Article number: 260 (2026) Cite this article

8361 Accesses
2 Citations
5 Altmetric
Metrics details

Subjects

Abstract

Stress granules (SGs) and processing bodies (PBs), assembled via liquid-liquid phase separation (LLPS), are critical for spatial regulation of gene expression in the cytoplasm. However, their roles in tumorigenesis remain poorly understood. Here, we show DEAD-box helicase 6 (DDX6) as the most promising vulnerability in acute myeloid leukemia (AML) through in vitro and in vivo CRISPR screenings using a specialized library targeting RNA-binding proteins enriched in SGs and PBs. Knockout (KO) of DDX6 significantly delays leukemogenesis with minimal impact on normal hematopoiesis. Importantly, the functions of DDX6 in AML depend largely on its ability to trigger LLPS and PB assembly. Mechanistically, PBs serve as “reservoirs” for the mRNAs interacting directly with DDX6 and having low GC content. DDX6 KO leads to rapid PB dissolution and release of PB-enriched mRNAs, such as BCAT1, into the cytosol, where these transcripts undergo degradation. By reducing BCAT1 levels, DDX6 KO reprograms amino acid metabolism and sensitizes AML cells to cytarabine chemotherapy.

m⁶A/IGF2BP3-driven serine biosynthesis fuels AML stemness and metabolic vulnerability

Article Open access 06 May 2025

A novel LncRNA risk model for disulfidptosis-related prognosis prediction and response to chemotherapy in acute myeloid leukemia

Article Open access 16 May 2025

Metabolomic profile of acute myeloid leukaemia parallels of prognosis and response to therapy

Article Open access 09 December 2023

Introduction

Eukaryotic cell organization relies on both membrane-bound and membrane-less organelles, enabling precise spatiotemporal regulation of gene expression, signal transduction, and biological processes. Evidence is now mounting that the membrane-less ribonucleoprotein (RNP) granules are dynamic liquid-liquid phase separation (LLPS) condensates formed through interactions between specific RNAs and RNA-binding proteins (RBPs), contributing to post-transcriptional regulation of gene expression¹. Stress granules (SGs) and processing bodies (PBs) are conserved and well-characterized RNP granules in the cytoplasm^2,3,4. SGs are formed when translation initiation is inhibited under stress conditions and contain mRNAs stalled in the process of translation^5,6. PBs are initially recognized as primary sites of mRNA decay due to the presence of decapping and degradation RBPs within them⁷. However, this concept is challenged by findings that the transcripts concentrated in PBs can exit and re-enter translation^4,8,9,10. Independent single-molecule imaging indicates that mRNAs are not degraded within PBs¹⁰. Those studies support a controversial conclusion that PBs are the sites of mRNA storage rather than decay³, underscoring the need to re-consider and investigate the precise role of PBs in regulating gene expression in specific biological contexts. Despite extensive studies reporting the crucial roles of SGs and PBs in determining mRNA translation and turnover⁵, their biological functions in various pathological processes, including leukemogenesis, are still not well understood.

Acute myeloid leukemia (AML), a fatal form of hematopoietic malignancy, is characterized by myeloid differentiation blockage and clonal expansion of immature progenitors^11,12,13. Although significant efforts have been dedicated to developing novel therapeutics, intensive chemotherapy, the “7 + 3 regimen” (cytarabine [Ara-C] for 7 days followed by anthracycline drug for 3 days), remains the mainstay of first-line AML treatment^14,15. With currently available therapeutics, only ~30% of AML patients survive beyond 5 years^13,16 and such survival rate has not notably increased in recent decades¹⁷. Over half of AML patients are either resistant to current chemotherapy or relapse after initial remission, with relapsed patients often succumbing to the disease within 1-2 years¹⁸. The high relapse rate results in unsatisfactory outcomes for AML patients and poses a significant challenge¹⁹. Thus, it is imperative to identify new vulnerabilities, especially those contributing to chemotherapy resistance. Emerging evidence suggests that drug-resistant cells rapidly adapt to stressful microenvironments, aiding their survival against chemotherapeutic drugs. For instance, SGs help cells to withstand various stressors, including chemotherapy, and promote chemoresistance by inducing cellular quiescence²⁰. Additionally, DDX3X, a member of DEAD-box helicases, confers drug resistance and radioresistance by facilitating the assembly of SGs^21,22. However, whether and how the cytosolic RNP granules are implicated in leukemogenesis or chemoresistance is still unclear.

AML cells, including leukemia stem cells (LSCs) that initiate the disease, exhibit remarkable plasticity in response to stressful microenvironments and adapt metabolically to sustain uncontrolled proliferation, a phenomenon termed metabolic reprogramming^{23,24,25,26,27}. Unlike normal hematopoietic stem cells (HSCs) which predominantly depend on glycolysis, AML cells, especially LSCs, heavily depend on mitochondrial oxidative phosphorylation (OxPhos) for energy conversion^28,29,30. This metabolic rewiring is increasingly recognized in hematological malignancies and is being leveraged to develop innovative therapeutic regimens. Numerous pre-clinical and clinical trials have demonstrated promising benefits of mitochondrion-based therapies in AML, such as BCL2 and IDH2 inhibitors^31,32,33,34. Reprogramming of amino acid metabolism in AML cells profoundly affects disease progression and chemotherapy response. Recent studies, including our own³⁵, have highlighted the significance of branched-chain amino acid (BCAA) metabolism in AML pathogenesis, LSC self-renewal, and drug resistance^35,36,37,38. BCAA transaminase 1 (BCAT1) is an enzyme that catalyzes BCAA catabolism, yielding branched-chain α-keto acids (BCKAs) and glutamate. The BCKAs can be further catabolized to acetyl-CoA and succinyl-CoA, fueling the tricarboxylic acid (TCA) cycle and OxPhos. Meanwhile, glutamine serves as the primary nitrogen source for synthesizing non-essential amino acids (NEAAs) and nucleotides to satisfy the heightened demands of leukemia cells. Despite these insights, the relationship between leukemogenesis, RNP granules, and metabolic reprogramming still puzzles us.

In this work, we design a specialized CRISPR library comprising over 2000 sgRNAs targeting the high-confidence SG and PB proteins to investigate the potential role of cytoplasmic LLPS granules in AML pathogenesis. High-throughput screenings are conducted both in vitro (using AML cell lines and patient-derived xenograft [PDX] cells) and in vivo (using AML PDX models). This analysis pinpoints DEAD-box helicase 6 (DDX6) as the most promising vulnerability in AML. Knockout (KO) of DDX6 significantly suppresses AML leukemogenesis and markedly reduces PB assembly, without profound impacts on normal hematopoiesis. Furthermore, the LLPS capability of DDX6 is essential for its biological functions in promoting AML. Additionally, multi-omics analysis reveals that PBs are the “reservoirs” for mRNAs that directly interact with DDX6 protein and have low GC content. DDX6 KO induces the disassembly of PBs, leading to the decay of these LLPS-dependent targets of DDX6, such as BCAT1. By decreasing BCAT1 expression, DDX6 KO reprograms amino acid metabolism, sensitizing AML cells to the standard chemotherapy drug cytarabine (Ara-C).

Results

In vitro and in vivo CRISPR screening reveals DDX6 as one of the most promising vulnerabilities in AML

To investigate whether cytoplasmic LLPS granules are involved in leukemogenesis, we designed a specialized CRISPR library with 2,084 sgRNAs targeting the 101 most confident RNA-binding proteins (RBPs) enriched in SGs and PBs (confidence score > 7; tier 1^39,40; almost all of them are RBPs) (Fig. 1a and Supplementary Data. S1). Gene Ontology analysis independently confirmed that these 101 genes are predominantly associated with RNPs, SGs, and PBs (Supplementary Fig. 1a). This library was lentivirally introduced into Cas9-expressing single clones of human Mono-mac-6 AML cell line (carrying MLL-AF9 fusion gene) and PDX-propagated AML blasts (carrying IDH2 and DNMT3A mutation) (Supplementary Fig. 1b, c). For in vitro screening, these cells were cultured until their amplification reached a range of 5,000× - 10,000×, allowing for the sufficient elimination of cells in which essential genes were disrupted. For in vivo screening, the AML PDX cells were transplanted into NSGS mice, and the donor AML cells were isolated from the bone marrow mononuclear cells (BM MNCs) of recipient mice upon their succumbing to AML. Integrative analysis of our in vitro and in vivo CRISPR screening results (Fig. 1b–d, Supplementary Fig. 1d–f, and Supplementary Data. S2-S4) along with DepMap CRISPR scores⁴¹ (Supplementary Fig. 1g and Supplementary Data. S5) revealed 5 top-ranked essential RBPs for AML cell survival, including EIF5A, DDX6, CNOT3, EIF4E, and PABPC1 (Fig. 1e–g and Supplementary Fig. 1h).

We next analyzed the phase separation capacity⁴² of the top 5 candidates and found that CNOT3, DDX6, EIF4E, and PABPC1 exhibited high phase separation scores (Fig. 1h, i). To investigate whether the crucial roles of these top candidates in AML cell survival are associated with their LLPS ability, we monitored the assembly of SGs and PBs in AML cells. Immunofluorescence (IF) imaging showed that PBs are consistently visible in AML cells (Dcp1b was used as PB marker⁴³; approximately 10-20 PBs per AML cell; Fig. 1j), whereas SGs are barely detectable under routine culture conditions (G3BP1 was used as SG marker; data not shown). Notably, among the top candidates, DDX6 KO had the most significant impact on suppressing PB assembly (Fig. 1j, k and Supplementary Fig. 1i, j). Our in-house CRISPR screens demonstrated robust decreases in DDX6 sgRNAs (Supplementary Fig. 1k–m), supporting its essential role in AML cell survival both in vitro and in vivo. Pan-cancer analysis indicated that DDX6 is most highly expressed in AML (Supplementary Fig. 1n). Moreover, DDX6 is significantly elevated in AML samples, including AML PDX cells (Fig. 1l–n), compared to healthy controls. To further validate our CRISPR findings, we utilized sgRNAs to deplete endogenous DDX6 and found that DDX6 KO significantly suppressed AML cell growth (Fig. 1o, p and Supplementary Fig. 1o, p). This effect could largely be reversed by overexpression of sgRNA-resistant DDX6 (Fig. 1p and Supplementary Fig. 1p), demonstrating the high specificity of our DDX6 sgRNA.

To elucidate the underlying mechanisms mediating the elevated DDX6 expression in AML, we integrated four independent datasets to predict potential transcriptional regulators of DDX6 and identified five common candidates: STAT3, KDM5B, EGR1, TRIM28, and E2F4 (Supplementary Fig. 2a). Correlation analysis using the TCGA AML cohort revealed a significant positive association between DDX6 expression and both STAT3 and KDM5B (Supplementary Fig. 2b–f). To further evaluate their regulatory roles, we used shRNAs to deplete STAT3 and KDM5B expression in AML cells and evaluated their impacts on DDX6 expression. Notably, depletion of KDM5B, but not STAT3, led to a significant reduction in DDX6 mRNA levels (Supplementary Fig. 2g, h). Consistently, KDM5B depletion substantially decreased DDX6 protein levels in AML (Supplementary Fig. 2i). Moreover, KDM5B depletion suppressed PB formation in AML (Supplementary Fig. 2j, k). Together, these results suggest that KDM5B regulates PB formation via DDX6 in AML.

In conclusion, our focused CRISPR screening highlights the essential role of DDX6 in the survival/proliferation of human AML cells in vitro and in vivo, and its biological function may be associated with its ability to facilitate PB assembly.

DDX6 undergoes phase separation in vitro and in cellulo

DDX6 has previously been reported to be required for PB formation^44,45,46,47, and PB assembly is typically driven by LLPS, which arises from multivalent interactions between RNAs and RBPs⁴⁸. Like other members of the DEAD-box RNA helicase family, DDX6 features a RecA-like DEAD domain and a helicase domain. Moreover, it contains two predicted intrinsically disordered regions (IDRs) at the N-terminus and C-terminus (Fig. 2a, b). IDRs are extensively implicated in LLPS, triggering the assembly of membrane-less organelles⁴⁹. To assess the subcellular localization of DDX6, we expressed a red fluorescent protein (RFP)-tagged DDX6 in mammalian cells and observed that DDX6 colocalized with the PB marker Dcp1b (Fig. 2c and Supplementary Fig. 3a). We then studied the dynamics of DDX6 granules using fluorescence recovery after photobleaching (FRAP) assay in live cells expressing DDX6-RFP and revealed that the average recovery half-time of DDX6 granules was approximately 4.2 s (Fig. 2d, e). To further validate whether DDX6 undergoes LLPS, we purified recombinant enhanced green fluorescent protein (EGFP)-tagged DDX6 protein from HEK293T cells and performed phase separation assays. The in vitro FRAP assay demonstrated a rapid recovery for DDX6 droplets, consolidating a strong propensity for LLPS (Fig. 2f, g). We also purified recombinant EGFP-tagged DDX6 protein from E. coli and repeated the phase separation assays. Our findings demonstrated that E. coli-derived DDX6 also underwent LLPS, although its phase separation capacity is weaker than mammalian cell-derived DDX6 protein (Supplementary Fig. 3b–d). Meanwhile, time-lapse imaging captured dynamic fusion events among DDX6 droplets, leading to the progressive formation of larger, more complex structures (Supplementary Fig. 3e). In addition, DDX6 started to form noticeable liquid droplets at a concentration of 1 µM; LLPS of DDX6 is enhanced by increased protein concentration and dampened in the presence of sodium chloride (Fig. 2h, i). Protein mapping further validated the dynamic behavior of DDX6 droplets (Fig. 2j). Moreover, in contrast to poly(C)-RNA, the addition of poly(U)-RNA greatly stimulated this process (Fig. 2kupper panel). Consistently, when incubating ATTO^TM 590 (NHS Ester)-labeled poly(U) RNA and ATTO^TM 425 (NHS Ester)-labeled poly(C) RNA together with DDX6 protein, we observed a significant enrichment of poly(U) RNA over the poly(C) RNA within DDX6 phase separation droplets (Supplementary Fig. 3f–h).

To determine which domain of DDX6 is crucial for its LLPS ability and biological functions, we engineered three truncated variants, DDX6-CT (C-terminus), DDX6-NT (N-terminus), and DDX6-ΔIDR (depletion of N-terminal IDR) for both in cellulo and in vitro (cell-free) assays (see Fig. 2b). In AML cells, DDX6 KO significantly decreased PB number, an effect that could be fully reversed by overexpression of DDX6-WT, partially reversed by DDX6-NT, but barely reversed by DDX6-CT or DDX6-ΔIDR (Supplementary Fig. 3i, j). This confirms that the N-terminus, particularly the N-terminal IDR domain, is essential for PB assembly. In vitro comparative analyses of DDX6-WT and its three truncated variants further highlighted the pivotal role of the N-terminal region, including the DEAD domain and IDR, in DDX6 LLPS (Fig. 2k, l and Supplementary Fig. 3k, l). Consistent with DDX6-WT, the presence of poly(U)-RNA, rather than poly(C)-RNA, dramatically promoted the droplet assembly of DDX6 truncated proteins (Fig. 2k, l). Compared to DDX6-NT, both DDX6-CT and DDX6-ΔIDR exhibited a more pronounced inhibition of LLPS capability. Based on in vitro droplet formation assays, the ranking of LLPS capacity among the DDX6 proteins is as follows: WT > NT > ΔIDR > CT. We also deleted the DEAD domain of DDX6 and found that its deletion moderately impaired DDX6’s ability to undergo LLPS (Supplementary Fig. 3m, n).

In summary, our in vitro and in cellulo results demonstrate that DDX6 undergoes LLPS, and N-terminal IDR and DEAD domain are essential for its LLPS capacity. Moreover, the addition of poly(U)-RNA, rather than poly(C)-RNA, substantially enhances the assembly of DDX6 droplets.

DDX6 promotes AML cell growth and LSC self-renewal while suppressing myeloid differentiation

To evaluate the role of DDX6 in leukemogenesis, we conducted both loss-of-function and gain-of-function assays in primary AML blasts and AML cell lines. We first isolated the Lin^- HSCs from Cas9 knock-in mice and retrovirally transduced MLL-AF9 fusion oncogene to induce malignancy transformation into pre-LSCs (Fig. 3a). Serial colony formation assays revealed that KO of Ddx6 (Fig. 3b) suppressed the colony-forming and repopulation capacities of MLL-AF9 LSCs (Fig. 3c, d). Next, we investigated the impact of DDX6 KO on the colony-forming ability of AML PDX cells. As shown in Fig. 3e, DDX6 KO markedly reduced colony number. Moreover, we included a variety of leukemia cell lines with distinct genetic backgrounds – Mono-mac-6, NOMO-1, THP-1, U-937, MV-4-11, OCI-AML3, and MA9.3RAS (in vitro-generated human cord blood CD34-derived AML as previously described⁵⁰)—to further validate the functions of DDX6. CRISPR-Cas9-mediated KO of DDX6 significantly suppressed AML cell proliferation in vitro, and these effects could be fully reversed by overexpression of DDX6-WT (Fig. 3f–i and Supplementary Fig. 4a–h). Wright-Giemsa staining and flow cytometry independently demonstrated that DDX6 KO induced myeloid differentiation (Fig. 3j–m and Supplementary Fig. 4i). Moreover, DDX6 KO promoted apoptosis in AML cells (Supplementary Fig. 4j–m) and induced cell cycle arrest at the G0 phase (Supplementary Fig. 4n–q). Importantly, these effects could be completely rescued by overexpression of DDX6-WT, confirming that the observed phenotypes in the DDX6 KO context were not due to any off-target effects.

To assess the extent to which DDX6’s role in AML depends on its LLPS ability, we employed additional rescue assays in DDX6 KO background. As shown in Fig. 3n–q, the DDX6-NT, which exhibits stronger LLPS ability than either DDX6-CT or DDX6-ΔIDR, more effectively rescued the anti-leukemic effects due to DDX6 KO, including reduced cell growth and enhanced myeloid differentiation. Given that our in vitro and in cellulo LLPS assays demonstrated that DDX6-CT and DDX6-ΔIDR largely lost their ability to assemble droplets, we hypothesized that DDX6’s LLPS capability is essential for its biological functions in promoting AML leukemogenesis. To test the hypothesis, we inserted the IDR domain of another LLPS protein (DDX3X-IDR1⁵¹) into the DDX6-ΔIDR construct, generating a chimeric protein (DDX3X-IDR1 + DDX6-ΔIDR), which showed robust LLPS ability (Supplementary Fig. 3m, n). Unlike DDX6-ΔIDR, which failed to reverse growth inhibition caused by DDX6 KO (Fig. 3q), the chimeric protein DDX3X-IDR1 + DDX6-ΔIDR could substantially reverse the growth inhibition (Supplementary Fig. 3o). Furthermore, we overexpressed LSM14A^47,52, a crucial component of PBs, in DDX6 KO cells to restore PB assembly. LSM14A OE not only rescued DDX6 KO-mediated PB disassembly but also partially reversed DDX6 KO-mediated growth inhibition and apoptosis (Fig. 3r–t). These findings demonstrate that DDX6 promotes AML survival through its LLPS-mediated assembly of PBs.

Deletion of Ddx6 minimally impacts normal hematopoiesis

To investigate the role of DDX6 in normal hematopoiesis, we generated Ddx6 conditional KO (cKO) mouse strains by crossing Ddx6^fl/fl mice (carrying loxP sites flanking exons 3 and 4) with Mx1-cre mice (Fig. 4a, b) to induce specific KO of Ddx6 in the hematopoietic system. We then administered poly (I:C) to Ddx6-WT, heterozygous KO, and homozygous KO mice to activate Cre recombinase. Lin^- HSCs were isolated from Ddx6^+/+;Mx1-cre (WT), Ddx6^fl/+;Mx1-cre (heterozygous), and Ddx6^fl/fl;Mx1-cre (homozygous) mice. Western blot confirmed Ddx6 KO efficacy in the Lin^- HSCs (Fig. 4c). We collected samples from peripheral blood, BM, and spleen from each group (Fig. 4d) to further evaluate the potential impacts of Ddx6 KO on normal hematopoiesis. Analysis of peripheral blood showed no significant impacts of Ddx6 KO on populations (or levels) of white blood cells (WBC; Fig. 4e), lymphocytes (LYM; Fig. 4f), platelets (PLT; Fig. 4g), and percentage of monocytes (MONO; Supplementary Fig. 5a). However, we observed a moderate but significant reduction in red blood cells (RBC; Fig. 4h), hemoglobin (HGB; Fig. 4i) and neutrophil (NEU; Fig. 4j) in the peripheral blood following Ddx6 KO. We also evaluated the effects of Ddx6 KO on various hematopoietic cell populations in BM and spleen, including both progenitor and mature cells, using flow cytometry. Neither homozygous nor heterozygous Ddx6 KO showed significant impacts on primitive Lin^-, LK (Lin^- c-Kit⁺), and LSK (Lin^- Sca-1⁺ c-Kit⁺) populations (Fig. 4k, l). The colony-forming assay did not identify any significant impact of Ddx6 KO on the colony-forming ability of Lin^- HSCs (Fig. 4m, n). The number of PBs was slightly but significantly reduced in Ddx6 KO compared to WT in Lin⁻ HSCs (Fig. 4o). Flow cytometry did not detect notable changes in populations of mature hematopoietic cells in the BM and spleen, including T cells, B cells, myeloid cells, and erythroid cells (Fig. 4p and Supplementary Fig. 5b). Additionally, we assessed the weight of the whole body and various organs, including spleen, liver, heart, lung, and kidney, in Ddx6 cKO and WT groups. This analysis showed that neither heterozygous nor homozygous KO of Ddx6 significantly affected body or organ weight (Fig. 4q, r and Supplementary Fig. 5c–k). Consistent with these findings, the histological analysis suggested that neither heterozygous nor homozygous KO of Ddx6 substantially influenced the structure or morphology of the organs (Supplementary Fig. 5l). Taken together, our characterization of Ddx6 cKO mice indicates that heterozygous or homozygous KO of Ddx6 has a very moderate effect on normal hematopoiesis.

DDX6 maintains the steady-state levels of its target transcripts that are enriched in PBs

To further delineate the molecular mechanisms by which DDX6 promotes AML survival and leukemogenesis, we conducted transcriptome-wide RNA-seq with AML cells following DDX6 KO (Fig. 5a and Supplementary Data. S6). We integrated the RNA-seq results with DDX6 cross-linking immunoprecipitation seq⁵³ (CLIP-seq; profiling DDX6-binding transcripts; Supplementary Data. S7), along with the dataset reporting PB-enriched and PB-depleted transcripts³ (Supplementary Data. S8). We categorized the RNA transcripts into four groups: (1) total mRNAs, (2) DDX6-binding mRNAs (determined by CLIP-seq), (3) DDX6-binding & PB-depleted mRNAs (transcripts that directly interact with DDX6 protein but are excluded from PBs), and (4) DDX6-binding & PB-enriched mRNAs (transcripts that directly interact with DDX6 protein and are enriched in PBs) (Fig. 5b). RNA-seq identified 1,368 significantly down-regulated mRNAs (p < 0.05; log₂[fold change] < -0.4) and 1,171 significantly up-regulated mRNAs (p < 0.05; log₂[fold change] > 0.4] (Fig. 5c). Among the DDX6-binding transcripts, 456 were significantly down-regulated, while 474 were significantly up-regulated following DDX6 KO (Fig. 5d). In the group of DDX6-binding & PB-depleted transcripts, we observed a greater tendency for up-regulation following DDX6 KO (309 up-regulated mRNAs vs 130 down-regulated mRNAs) (Fig. 5e). This could be attributed to the reported interaction between DDX6 and CNOT1, which mediates mRNA decay in the cytosol^54,55. Among the DDX6-binding & PB-enriched transcripts, we observed a greater number of downregulated mRNAs upon DDX6 KO (87 up-regulated mRNAs vs 226 down-regulated mRNAs) (Fig. 5f).

We next analyzed the GC content of DDX6-binding transcripts. In contrast to DDX6-binding & PB-depleted transcripts, the DDX6-binding & PB-enriched transcripts have significantly lower GC content (Fig. 5g, Supplementary Fig. 6a, and Supplementary Data. S9). Additionally, incorporating RNA-seq data revealed that DDX6-binding & PB-enriched transcripts with decreased levels upon DDX6 KO had a much lower GC content than the DDX6-binding & PB-depleted transcripts with increased levels upon DDX6 KO (Fig. 5h and Supplementary Fig. 6b). Furthermore, we compared the GC content between the DDX6 KO_Up & DDX6-binding & PB-depleted transcripts and the DDX6 KO_Down & DDX6-binding & PB-depleted transcripts, finding no significant difference (Supplementary Fig. 6c, d). The DDX6 KO_Down & DDX6-binding & PB-enriched transcripts displayed a much lower GC content than their upregulated counterparts (Supplementary Fig. 6e, f). We also randomly selected several GC-rich transcripts and GC-poor transcripts to evaluate their potential enrichment in cytoplasmic DDX6 LLPS granules (Supplementary Fig. 7a–c), identifying the enrichment of GC-poor mRNAs (Supplementary Fig. 7d–g), rather than GC-rich mRNAs (Supplementary Fig. 7h–n), within them. Based on these results, we propose that DDX6 may regulate the PB-enriched transcripts and the PB-depleted transcripts through distinct mechanisms: DDX6 preferentially decreases the expression levels of its binding transcripts that are excluded from PBs, while it preferentially maintains the steady-state levels of its binding transcripts that are enriched in PBs. Moreover, GC content plays a critical role in guiding the localization of DDX6-binding transcripts into PBs.

Considering that DDX6 LLPS ability is essential for AML cell survival and that DDX6-binding & PB-enriched mRNAs exhibit significantly lower GC content, we next focused on the DDX6 KO_Down & DDX6-binding & PB-enriched transcripts with low GC content to investigate how DDX6 modulates their expression (Fig. 5i). Gene Set Enrichment Analysis (GSEA) indicates that these potential DDX6 target transcripts are enriched in pathways related to cell cycle, cell division, cellular response to stress, and other processes (Fig. 5j, k). Further analysis of the top core-enriched genes in these pathways revealed that DDX6 KO substantially reduced their steady-state levels (Fig. 5l). We then assessed the correlation between DDX6 and the core-enriched genes in AML cell lines⁴¹ (Fig. 5m) and TCGA cohort (Fig. 5n), finding that most of these genes showed a significant positive correlation with DDX6. Of note, BCAT1 mRNA (GC content: 37%) stands out as the most significantly down-regulated transcript following DDX6 KO and demonstrates a significant positive correlation with DDX6 in AML (Fig. 5l–n).

In summary, our results suggest that transcripts with lower GC content tend to be enriched within PBs and that DDX6 positively maintains the steady-state levels of these PB-enriched transcripts.

DDX6 regulates AML metabolic plasticity by targeting BCAT1 mRNA

To investigate whether DDX6 directly interacts with BCAT1 mRNA, we performed gene-specific CLIP-qPCR using whole-cell lysis. A robust interaction was detected with WT DDX6, which was markedly reduced upon deletion of the IDR domain (Fig. 6a). Additionally, we isolated DDX6⁺ PBs for further CLIP-qPCR analysis, confirming that DDX6 directly interacts with BCAT1 mRNA within PBs (Supplementary Fig. 8a, b). Moreover, we conducted single-molecule fluorescent in situ hybridization (smFISH) to determine the localization of BCAT1 mRNA independently. The smFISH images revealed co-localization between BCAT1 mRNA and DDX6 protein within PBs (Fig. 6b, Supplementary Fig. 8c, d, and Supplementary Data. S10). Emetine-induced disassembly of PBs notably reduced the co-localization between BCAT1 mRNA and DDX6 protein (Supplementary Fig. 8d). An independent in vitro (cell-free) RNA pull-down assay further confirmed the direct interaction between BCAT1 mRNA and DDX6 (Fig. 6c–e). Our RNA stability assays indicated that DDX6 significantly increased the half-life of BCAT1 mRNA, and vice versa (Fig. 6f, g and Supplementary Fig. 8e–j). Additional qPCR and Western blotting showed that DDX6 KO markedly decreased BCAT1 expression at both mRNA and protein levels, which could be reversed by overexpressing DDX6-WT, but not the ΔIDR mutant (Fig. 6h, i and Supplementary Fig. 8k–s). These results suggest that BCAT1 mRNA is not degraded in PBs; rather, PBs likely serve as a storage site for DDX6-binding and GC-poor transcripts, such as BCAT1, protecting them from degradation in the liquid-like cytosol. This finding aligns with previous reports^3,56.

We then conducted rescue assays to determine whether BCAT1 is a bona fide target of DDX6, mediating its biological function in leukemogenesis. As shown in Fig. 6j, k and Supplementary Fig. 9a–l, the growth inhibition, apoptosis, and cell cycle arrest induced by DDX6 KO could be partially reversed by overexpression of BCAT1. BCAT1 converts BCAAs to BCKAs, which are subsequently decarboxylated to produce derivatives of acetyl-CoA and succinyl-CoA, thereby fueling the TCA cycle and OxPhos^35,57. In AML cells, forced expression of BCAT1 significantly increased the oxygen consumption rate (OCR) (Fig. 6l, m and Supplementary Fig. 9m, n). Moreover, DDX6 KO led to a significant decrease in OCR, an effect that could be largely reversed by overexpression of BCAT1 (Fig. 6n, o and Supplementary Fig. 9o, p). These findings demonstrate that BCAT1 acts as a functionally essential target of DDX6 in AML.

To systemically evaluate the impact of the DDX6/BCAT1 axis on metabolism, we performed stable isotope tracer (¹³C,¹⁵N-leucine) experiments coupled with liquid chromatography-mass spectrometry (LC-MS) in AML cells following DDX6 KO (Supplementary Fig. 10a and Supplementary Data. S11). Among the significantly dysregulated ¹³C- or ¹⁵N-labeled metabolites (p < 0.05), majority of them were downregulated after DDX6 KO (20 downregulated metabolites vs 4 increased metabolites) (Fig. 6p). Further analysis revealed that the 20 significantly reduced ¹³C- or ¹⁵N-labeled metabolites were enriched in the pathways related to (1) glyoxylate and dicarboxylate metabolism, (2) alanine, aspartate, and glutamate metabolism, (3) valine, leucine, and isoleucine biosynthesis, among others (Supplementary Fig. 10b, c). Consistent with the decrease in OCR following DDX6 KO and the role of BCAT1 in metabolism, we found that DDX6 KO markedly decreased the levels of metabolites associated with the TCA cycle, NEAA, and nucleotides (Fig. 6q). Specifically, DDX6 KO led to significant reductions in ¹⁵N-labeled NEAAs (Fig. 6r), ¹⁵N-labeled nucleotides (Fig. 6s), and ¹³C-labeled TCA metabolites such as malate and citrate (Fig. 6t, u). Additionally, we analyzed the abundance of all the metabolites detected in our LC-MS to comprehensively assess the effects of DDX6 KO on metabolism (Supplementary Fig. 10d, e). This analysis indicated that the significantly decreased metabolites upon DDX6 KO were enriched in pathways related to (1) arginine and proline metabolism, (2) arginine biosynthesis, (3) alanine, aspartate, and glutamate metabolism, among others (Supplementary Fig. 10f), highlighting the crucial role of DDX6 in amino acid metabolism. Once again, DDX6 KO significantly reduced the abundance of TCA cycle metabolites, nucleotides, and NEAAs (Supplementary Fig. 10g–i). These findings strongly suggest that DDX6 KO profoundly impacts BCAA metabolism in AML, primarily through the reduced expression of BCAT1.

DDX6 KO sensitizes AML cells to chemotherapy

Given the established role of BCAA metabolism in drug resistance^58,59 and our identification of BCAT1 as a functionally essential target of DDX6, we explored the potential impact of DDX6 on the response of leukemia cells to first-line chemotherapy drugs. Notably, DDX6 KO and BCAT1 knockdown (KD) significantly increased the sensitivity of leukemia cells to Ara-C treatment in vitro; and DDX6 KO-induced sensitivity could be partially reversed by BCAT1 overexpression (Fig. 7a–c and Supplementary Fig. 11a). Similar to DDX6 KO, BCAT1 KD significantly suppressed AML proliferation (Supplementary Fig. 11b, c). We also calculated the IC₅₀ values for Ara-C in control and DDX6 KO AML cells and identified that DDX6 KO substantially reduced the IC₅₀ values of Ara-C (Fig. 7d and Supplementary Fig. 11d).

To confirm the effect of DDX6 KO on chemotherapy sensitivity, we utilized AML “human-in-mouse” xenograft and PDX models. In vivo bioluminescence imaging revealed that Ara-C treatment or DDX6 KO significantly reduced leukemia burden. Notably, the combination of Ara-C and DDX6 KO resulted in a more pronounced inhibition of AML leukemogenesis (Fig. 7e and Supplementary Fig. 11e). We also monitored the survival across all groups, revealing that the combination of Ara-C and DDX6 KO significantly prolonged the overall survival of recipient mice compared to the control group, Ara-C alone, or DDX6 KO alone (Fig. 7f, g). In addition, Wright-Giemsa staining indicated that the combination treatment notably decreased the infiltration of AML cells into peripheral blood and induced AML cell differentiation (Supplementary Fig. 11f). H&E staining independently demonstrated that the combination treatment was more effective in suppressing AML infiltration into the liver and spleen than either treatment alone (Fig. 7h and Supplementary Fig. 11g). Consistently, flow cytometry analysis of AML engraftment into recipient mice showed that the combination of Ara-C and DDX6 KO had the most robust effect on inhibiting AML cell population in vivo (Fig. 7i and Supplementary Fig. 11h–k).

Altogether, our studies demonstrate that DDX6, a protein undergoing LLPS and triggering the assembly of PBs, preferentially deposits GC-poor RNAs, i.e., BCAT1 mRNA, into PBs, protecting them from degradation and maintaining their steady-state levels. By positively regulating BCAT1 expression and reprogramming amino acid metabolism, DDX6 mediates the resistance of AML cells to the standard chemotherapy drug Ara-C.

Discussion

DDX6, particularly its LLPS ability, promotes AML leukemogenesis

In this study, we conducted both in vitro and especially in vivo CRISPR screening to unbiasedly evaluate which RBPs concentrated within cytoplasmic RNP granules, SGs and PBs, are essential for AML pathogenesis. This high-throughput screening identified multiple RBPs that are required for AML growth in vitro and leukemogenesis in vivo, including EIF5A, DDX6, CNOT3, EIF4E, and PABPC1. Notably, several previous studies have reported the crucial oncogenic roles of these top hits, such as DDX6, CNOT3 and EIF4E^60,61,62,63, in leukemogenesis, independently consolidating our CRISPR screening results. Among these top hits, DDX6 stood out as the most important gene for PB assembly. Furthermore, additional compelling functional results across various AML cells with different genetic backgrounds demonstrated that DDX6 KO significantly suppressed AML proliferation, induced myeloid differentiation, and reduced OxPhos in vitro, and substantially delayed AML progression in vivo. Using Ddx6 cKO mice, we identified that Ddx6 is dispensable for normal hematopoiesis, suggesting a potential therapeutic window for targeting DDX6 in AML treatment.

It is worth mentioning that one recent study reported the role of DDX6 in myeloid leukemia⁶³, further strengthening our findings regarding DDX6’s function in AML pathogenesis. However, more comprehensive studies are warranted to delineate the mechanism underlying the importance of DDX6 in AML. In particular, two key questions remain: (1) does DDX6 indeed undergo LLPS in AML, and (2) to what extent does the important role of DDX6 in AML rely on its LLPS ability? Herein, we characterize the LLPS behavior of DDX6 protein in vitro (cell-free system) and in cellulo, including AML cells. By engineering truncated DDX6 proteins, we identified that the IDR at the N-terminus is necessary for DDX6 LLPS. Additionally, rescue assays in DDX6 KO AML cells demonstrated that DDX6’s LLPS ability is essential for its biological functions, mediating its oncogenic role in leukemogenesis. Thus, our study addresses a key knowledge gap regarding the involvement of DDX6 and LLPS in AML leukemogenesis.

DDX6-mediated LLPS and PB assembly in post-transcriptional regulation of gene expression

Integrative analysis of multi-omics sequencing data reveals that DDX6 tends to negatively regulate potential target transcripts that are depleted from PBs and have high GC content; while positively regulating those that are enriched in PBs and have low GC content. Notably, poly(U) RNA, rather than poly(C) RNA, significantly promotes DDX6 LLPS. Among the transcripts directly bound by DDX6, those with high GC content are preferentially excluded from PBs; whereas GC-poor transcripts tend to condense within PBs. These findings suggest that GC-poor RNA may enhance DDX6 LLPS both in vitro and in cellulo, a notion supported by other independent studies⁵⁶. It is worth mentioning that we found that DDX6 protein derived from mammalian cells exhibits significantly stronger LLPS capability than its bacteria-derived counterpart, suggesting that post-translational modifications may regulate DDX6 phase separation⁶⁴.

Given our established roles of DDX6, particularly its capacity to trigger LLPS and PB assembly to promote AML leukemogenesis, we are especially interested in understanding how DDX6 determines the fate of its target transcripts that are enriched in PBs. For most DDX6-binding and PB-enriched transcripts, their steady-state levels decrease following DDX6 KO. Interestingly, a similar reduction was observed in many DDX6-non-binding and PB-enriched transcripts (data not shown). These effects are likely due to the rapid disassembly of PBs triggered by DDX6 KO, which releases PB-enriched, GC-poor transcripts into the liquid cytosol, where they become susceptible to degradation. For instance, BCAT1 mRNA (GC vs AU: 37% vs 63%), one of the most significantly decreased transcripts following DDX6 KO, is normally enriched in PBs, protecting it from degradation. DDX6 KO induces PB disassembly and subsequent decay of BCAT1 mRNA. Thus, DDX6 plays a crucial role in the dynamic regulation of RNA metabolism by depositing RNA in PBs and releasing it into the cytoplasm in response to cellular needs, particularly through its LLPS ability. This concept is further supported by observations that translation-repressed mRNAs accumulated in PBs can be protected from decay and re-enter translation, thereby modulating tumorigenesis, normal development, and cellular plasticity^3,4,65. In AML, PBs driven by DDX6 may function as reservoirs for GC-poor RNAs, maintaining their steady-state levels. Our work provides a comprehensive understanding of the multifaceted functions of DDX6 and its impact on post-transcriptional regulation during oncogenesis, offering valuable insights into the mechanisms underlying PB-mediated RNA storage and dynamics.

LLPS granules: paradigm-shifting therapeutic targets

Numerous studies have emphasized the essential role of DDX6 in PB assembly^{48,63,66,67,68}. However, the extent to which DDX6’s functions in specific biological contexts depend on its LLPS ability to drive PB assembly remains elusive. Our research demonstrates that DDX6 is essential for AML leukemogenesis in vitro and in vivo, and that its LLPS ability is critical for these effects. Specifically, the anti-leukemia effects mediated by DDX6 KO could not be reversed by forced expression of truncated DDX6 that lacks LLPS ability. Moreover, DDX6 maintains BCAT1 expression level through its LLPS capacity, thereby reprogramming amino acid metabolism and contributing to chemotherapy resistance. Importantly, Ddx6 KO has minimal effect on normal hematopoiesis, strongly suggesting that DDX6 represents a promising vulnerability for treating AML and circumventing therapeutic resistance.

The crucial roles of DEAD-box helicase members are increasingly recognized across various cancers. For example, DDX3, another RNA helicase with robust LLPS ability, has been reported to promote oncogenesis in lung, breast, and lymphoma cancers^69,70,71. Proof-of-concept studies have shown that pharmacological inhibition of DDX3 suppresses lung cancer progression and sensitizes cancer cells to radiation⁷¹. Fused in sarcoma (FUS), another RBP with potent LLPS ability, is genetically and pathologically associated with aggressive forms of amyotrophic lateral sclerosis (ALS). The FUS granules are the pathological hallmark of this disease. Clinical trials have demonstrated that silencing FUS expression with an antisense oligonucleotide (ASO; ION363) is an attractive therapeutic strategy in FUS-dependent ALS⁷². Beyond these examples, a recent study reported that small-molecule inhibitors targeting G3BP1, a central hub for SGs, dramatically inhibit the formation of SGs⁷³. We are currently at the infant stage of understanding the biological functions and mechanisms of membrane-less LLPS granules. Discovery of inhibitors targeting these granules will not only enable more rigorous investigations to bridge the knowledge gap between LLPS granule assembly and disease pathogenesis but also reveal paradigm-shifting therapeutic targets. In particular, the future development of inhibitor(s) targeting DDX6’s LLPS ability or silencing DDX6 expression presents exciting opportunities, potentially opening new avenues for new treatments for leukemia and strategies to combat drug resistance.

Methods

Cell culture

The human leukemia cells THP-1 (ATCC), NOMO-1 (DSMZ), MV-4-11 (ATCC), ML-2 (DSMZ), MOLM-13 (DSMZ), Kasumi-1 (ATCC), U-937 (ATCC), and HL-60 (ATCC) were cultured in RPMI1640 (11875119, Thermo Fisher Scientific) supplemented with 10% FBS (FB-11, Omega Scientific). OCI-AML3 (DSMZ) was maintained in MEM α, nucleosides (12571063, Thermo Fisher Scientific) supplemented with 10% FBS. Mono-Mac-6 (DSMZ) was maintained in RPMI1640 supplemented with 10% FBS plus 2 mM L-glutamine (25030-081, Thermo Fisher Scientific),1x non-essential amino acids (11140050, Thermo Fisher Scientific), 1 mM sodium pyruvate (11360070, Thermo Fisher Scientific), and 10 μg/ml human insulin (12585014, Thermo Fisher Scientific). HEK293T (ATCC) was maintained in DMEM (11-995-073, Fisher Scientific) supplemented with 10% FBS. The AML PDX cells (Supplementary Data. S12) and MA9.3RAS cells were cultured in IMDM (12-440-061, Fisher Scientific) supplemented with 10% FBS, with human cytokines (10 ng/ml IL3 [200-03, PeproTech], 10 ng/ml IL6 [200-06, PeproTech], 10 ng/ml FLT3-ligand [300-19, PeproTech], 10 ng/ml TPO [300-18, PeproTech], 10 ng/ml SCF [300-07, PeproTech]). All AML patients and healthy donors signed an informed consent form. Sample acquisition was approved by the Institutional Review Board (IRB# 18067) at City of Hope. All cell lines were cultured in the medium containing 1% (100U/ml final concentration) penicillin-streptomycin (15140122, Thermo Fisher Scientific) 10 mM HEPES (15630080, Thermo Fisher Scientific), and 25 μg/ml plasmocin prophylactic (ant-mpp, InvivoGen), as specified by the manufacturers, and maintained in an incubator with 5% CO₂ at 37 °C. All cell lines were confirmed to be free of mycoplasma contamination prior to further study.

Animals

All the animal-related experiments were performed in accordance with institutional guidelines and IACUC protocol approved by the City of Hope. The human-in-mouse xenograft AML models and patient-derived xenograft (PDX) AML models were established by transplanting 6- to 8-week-old NSGS mice (NOD.Cg-Prkdc^scid Il2rg^tm1Wjl Tg (CMV-IL3, CSF2, KITLG)1Eav/MloySzJ; RRID: IMSR_JAX:013062) with human AML cells. The Ddx6 conditional knockout (cKO) mouse models (Ddx6^fl/+ and Ddx6^fl/fl; genetic background: C57BL/6 J [CD45.2]) were created with CRISPR/Cas-mediated genome engineering by Cyagen. Murine Ddx6 exons 3 and 4 were selected as the cKO region, with loxP sites inserted to flank these exons. Deletion of this region results in a frameshift in Ddx6. Mx1-cre (B6. Cg -Tg (Mx1-cre)1Cgn/J) mice were purchased from The Jackson Laboratory (RRID: IMSR_JAX:003556). The Ddx6^fl/fl, Ddx6^fl/+, Ddx6^+/+ mice were crossed with Mx1-cre mice to generate Ddx6^+/+; Mx1-cre, Ddx6^fl/+; Mx1-cre and Ddx6^fl/fl; Mx1-cre mice. To induce expression of Cre recombinase, those animals were administered with poly (I:C) (polyinosinic: polycytidylic acid [4287, R&D systems]) by intraperitoneal (i.p.) injection (10 mg/kg; every other day treatment for 7 treatments in total). All mice were bred at a certified City of Hope animal facility by standard procedures. All mice were housed on a 12 h:12 h light-dark cycle with ad libitum food and water. All animal experiments included both male and female mouse subjects, with equal distribution across experimental groups.

The Cas9 single clones of Mono-mac-6 or AML PDX cells with stable firefly luciferase expression (pLenti-CMV-Puro-LUC [17477, Addgene]) were lentivirally transduced with sgNS or sgDDX6-1. Transduction-positive cells were selected with 200 μg/ml hygromycin B (ant-hg-5, InvivoGen) and 10 μg/ml Blasticidin (ant-bl-1, InvivoGen) for nine days before transplantation into NSGS mice. For AML xenograft models, 0.3 × 10⁶ Mono-mac-6 cells with or without DDX6 KO were transplanted into each NSGS mouse, while 0.1 × 10⁶ AML PDX cells were transplanted into each mouse. One to two weeks post-transplantation, animals in the sgNS or sgDDX6-1 group were randomly divided into two subgroups for Ara-C treatment (10 mg/kg daily i.p. administration; C1768, Sigma) or PBS (vehicle control) treatment. The animals were euthanized when AML cell engraftment reached 90% in the recipient mice or when body weight decreased by 20%. At endpoints, BM and PB MNCs were collected and subjected to flow cytometry analysis, and Wright-Giemsa Stain; liver and spleen tissues were collected for H&E staining.

In vivo bioluminescence imaging

In vivo bioluminescence imaging assay was employed to monitor leukemia burden in NSGS recipient mice. D-luciferin (LUCK-2G, Gold bio) was dissolved in Dulbecco’s phosphate-buffered saline (DPBS, no calcium, no magnesium). For the imaging procedure, mice were weighed and intraperitoneally injected with 150 mg/kg D-luciferin, followed by anesthesia with isoflurane. Ten minutes after the D-luciferin injection, whole-body bioluminescence imaging was conducted using the Lago X system (Spectral Instruments Imaging). The bioluminescence signal was quantified in radiance, expressed in “photons/seconds/cm²/steradian”. Pseudocolor was used to indicate the signal strength corresponding to leukemia burden.

Lentiviral CRISPR library construction

CRISPR sgRNAs were selected using the Genetic Perturbation Platform (Broad Institute), and the CRISPR library was constructed following our previous publications^{74,75,76,77,78}. For the SG/PB-focused CRISPR library, the 2084 sgRNAs, including 2020 sgRNAs targeting the 101 genes, 24 sgRNAs targeting positive control genes, and 40 sgRNAs targeting negative genes were designed (Supplementary Data. S1). These sgRNA oligonucleotides were synthesized via microarray (CustomArray) and cloned into the ipUSEPR lentiviral sgRNA vector (hU6-driven sgRNA co-expressed with EF-1a-driven red fluorescent protein [RFP] and puromycin-resistance gene) using the BsmBI (R0580, New England Biolabs) restriction sites. The molecular cloning was performed using the NEB 5-alpha Competent E. coli (C2987H, New England Biolabs). The library lentivirus was produced in HEK293T cells by co-transfecting ipUSEPR vectors with the packaging plasmids pPAX2 (12260, Addgene) and pMD2.G (12259, Addgene).

In vitro and in vivo CRISPR library screening and data analysis

The CRISPR library screening was conducted as described previously^{74,75,76,77,78}. Briefly, the SG/PB-focused CRISPR library was introduced into Cas9 single clones of Mono-mac-6 or AML PDX cells. The cells were infected with the CRISPR library with a multiplicity of infection (MOI) = 0.2–0.3 (monitored by flow cytometry for RFP expression). Three days after infection, cells were selected with 0.5 μg/ml puromycin (ant-pr-1, InvivoGen) for Mono-mac-6 cells and 16 μg/ml for PDX cells. For in vitro CRISPR screening, the library-transduced cells were cultured for an additional 28 days. At the start (day 0) and end (day 28) time points, 2 million cells from each group were collected. For in vivo CRISPR screening, the library-transduced AML PDX cells were selected with puromycin for 5 days and then transplanted into NSGS mice (1 × 10⁷ cells/mouse and 6 mice in total). The cells from mouse BM were collected around week 4 post-transplantation when the overt disease was observed, and CD45 beads (130-045-801, Miltenyi Biotec) were used to isolate CD45⁺ AML cells from BM MNCs. The integrated sgRNA-containing regions were amplified by PCR using primers DCF01 5′-CTTGTGGAAAGGACGAAACACCG-3′ and DCR03 5′-CCTAGGAACAGCGGTTTAAAAAAG C-3′. Amplicon sequencing was performed on an Illumina NextSeq 500 sequencer. To quantify sgRNA reads, 20-nucleotide sequences that matched the sgRNA backbone structure (5’ prime CACCG and 3’ prime GTTT) were extracted and aligned to the sgRNA sequences of the CRISPR screening library using Bowtie2. The frequency for individual sgRNAs was calculated as the read counts of each sgRNA divided by the total read counts matched to the library. The normalized CRISPR score was defined as a log10-fold change in the frequency of individual sgRNAs between the start (day 0) and end (day 28) of the screened samples and normalized by the median score of the negative control sgRNA (defined as 0.0; sgRNA targeting non-essential sequences).

Plasmid construction

The human DDX6 gene was amplified by CloneAmp HiFi PCR Premix (639298, Takara) using human cDNA made from HEK293T cell and cloned into lentivector-based pSIN4 (61065, Addgene) or pCDH (46970, Addgene) or Ubic (119947, Addgene) vectors. Synonymous mutated DDX6 and truncated DDX6, including NT, CT, and ΔIDR, were constructed by In-Fusion Cloning Plus CE (638916, Takara) according to the manufacturer’s instructions. The sgRNAs against DDX6, CNOT3, EIF4E, and PABPC1 were designed using CRISPick, synthesized by Eton Bioscience, and then cloned into the lenti-sgRNA hydro plasmid (104991, Addgene) using the BsmBI (R0580, New England Biolabs) restriction sites. The shRNAs of BCAT1 (TRCN0000005906, TRCN0000005907), KDM5B (TRCN0000014762, TRCN0000358504), and STAT3 (TRCN0000020843, TRCN0000329886, TRCN0000329887) were purchased from Sigma-Aldrich. All the primers used for molecular cloning are listed in Supplementary Data. S13. The plasmids were extracted using a High-speed mini plasmid kit (IB47102, IBI Scientific) or E.Z.N.A.® Endo-free Plasmid DNA Mini Kit II (D6950-02, Omega) and validated by Sanger sequencing (Eton Bioscience and Plasmidsaurus lab).

Lentivirus production

The lentiviruses utilized in this study, including those to mediate knockout (KO) or overexpression (OE) of DDX6, BCAT1, Cas9, or other genes, were produced with HEK293T cells. Briefly, the lentiviral packaging vectors (2.25 μg psPAX2 and 0.75 μg pMD2.G) plus 3 μg lentiCas9-Blast plasmid (52962, Addgene), or 3 μg lentiGuide-hydro sgRNA, or 3 μg OE plasmids were co-transfected into HEK293T cells using X-tremeGENE HP DNA Transfection Reagent (6366546001, Roche; approximately 9 μl for each dish) in 6-cm dishes. The media with lentiviruses were harvested at 48 h and 72 h post-transfection and filtered using a 0.45 µm syringe filter. Then, the lentivirus particles were added to AML cells or HEK293T cells directly with 8 μg/ml polybrene (H9268, Sigma-Aldrich). After two rounds of infection and spinoculation (1200 rpm, 2 h,32 °C), the lentivirus transduction-positive cells were selected with 10 μg/ml blasticidin (for Cas9 expression) 200 μg/ml hygromycin B (for sgRNA expression) or 2 μg/ml puromycin (for DDX6 and BCAT1 OE) for 1–3 weeks. KO and OE efficiencies were validated by Western blotting.

Protein extraction and Western blotting

The cells were lysed using RIPA lysis and extraction buffer (89901, Thermo Fisher Scientific) supplemented with 1× protease inhibitor cocktail (78438, Thermo Fisher Scientific). For Western blotting, equal amounts of proteins were separated by 10% SDS-PAGE gel and then transferred onto polyvinylidene difluoride (PVDF) membranes (88518, Thermo Fisher Scientific). Chemiluminescence signaling was detected by Pierce ECL™ Prime Western Blotting Detection Reagent (GERPN2236, Sigma) or Thermo Scientific™ Pierce™ ECL Western Blotting Substrate (PI32106, Thermo Fisher Scientific). Antibodies used for Western blotting were listed as follows: DDX6 (ab70455, Abcam; 1:2000), DDX6 (14632-1-AP, Proteintech; 1:2000), BCAT1 (TA504360S, Origene Technologies; 1:2000), FLAG (F3165, Sigma-Aldrich; 1:2000), Goat Anti-Mouse IgG H&L (HRP) (ab6789, Abcam; 1:2000), and Goat Anti-Rabbit IgG H&L (HRP) (ab6721, Abcam; 1:2000). β-actin (3700S, Cell Signaling Technology; 1:2000), Vinculin (sc-73614, Santa Cruz; 1:2000), or GAPDH (sc-47724, Santa Cruz; 1:2000) were used as loading controls. All unprocessed gel data are available in Source Data 1.

Immunofluorescence (IF) imaging

For IF imaging, cells were seeded on 8-well chamber slides (C8-1.5H-N, Cellvis) at a density of 1 × 10⁴ cells per well. 24 h later, the following plasmids were transiently transfected into these cells using X-tremeGENE HP DNA Transfection Reagent (6366546001, Roche): Ubic-DDX6-WT-RFP and GFP-Dcp1b (153976, Addgene) (250 ng plasmid/well). IF images were captured 72 h post-transfection using a Zeiss LSM 880 confocal microscope with an Airy scan model (Zeiss, Germany).

RNA isolation, cDNA synthesis, and qPCR

Total RNA was isolated using QIAzol Lysis Reagent (79306, Qiagen) following the manufacturer’s protocol. RNA concentration was measured using a NanoDrop spectrometer. Reverse transcription was conducted with 0.5-1 μg total RNA using QuantiTect Reverse Transcription Kit (205314, Qiagen) according to the manufacturer’s instructions to generate cDNA. Quantitative real-time PCR (qPCR) was performed using the SYBR Green qPCR master mix (FERK0253, Thermo Fisher Scientific) and a QuantStudio^TM 7 Flex Real-time PCR system (Applied BioSystems). GAPDH,18S rRNA, or 28S rRNA served as the endogenous controls. Three biological replicates were included for each reaction. All the primers used for qPCR are listed in Supplementary Data. S13.

Fluorescence recovery after photobleaching (FRAP)

FRAP assays were conducted using the bleaching module of the Zeiss LSM 880 confocal microscope with a 63 × oil immersion lens. The 488 nm laser was employed to bleach the DDX6-EGFP signal, while the 561 nm laser was used to bleach the DDX6-RFP signal. Bleaching was applied to a rectangle region of interest (ROI) using 100% laser power, and time-lapse images were subsequently collected. An equal-sized rectangle area away from the bleaching point was selected as an unbleached control. Fluorescence intensity was measured using Zen software, and the values were reported relative to pre-bleaching time points. GraphPad Prism was used to plot the data. The halftime for each granule was calculated using the following formula: I = I0-I1*exp(-t/T1), where I0 is the recovery fraction, I1 is the mobile fraction, T1 is the fitted parameter, I is the delta immobile fraction. A two-tailed t-test was used to calculate the p values. For in cellulo FRAP assay, live HEK293T cells overexpressing Ubic-DDX6-RFP were utilized. For in vitro (cell-free) FRAP assay, recombinant DDX6-EGFP protein purified from HEK293T cells was employed.

Recombinant protein expression and purification from HEK293T cells

The recombinant proteins DDX6-WT-EGFP, DDX6-NT-EGFP, DDX6-CT-EGFP, DDX6-ΔIDR-EGFP, DDX6-ΔDEAD-EGFP and DDX3X-IDR1 + DDX6-ΔIDR-EGFP were purified from HEK293T cells. Briefly, the plasmids were transiently transfected into HEK293T cells to 50-60% confluency in 150 mm dishes. For each dish, 24 μg plasmids were mixed with 60 μg polyethylenimine (PEI) for transfection. To maximize protein yield, 10 dishes of HEK293T cells were prepared for each plasmid. 48-72 h post-transfection, the cells were harvested at approximately 80-90% confluency, washed with PBS, and suspended in the protein lysis buffer (50 mM HEPES, 150 mM NaCl, 100 nM DTT, 2‰ NP-40, and 1 × protease inhibitor cocktail). The cell lysates were sonicated and then centrifuged at 15,000 × g for 30 min at 4 °C to collect the supernatants. Protein purification was performed using a Ni-NTA Agarose column (11-0230-5x5ML, Marvelgent Biosciences) with an ÄKTA start protein purification system (GE Healthcare-Cytiva). The recombinant proteins were eluted from the agarose column using a gradient of imidazole with low imidazole buffer (100 mM imidazole, 300 mM NaCl, 50 mM Na₂HPO₄) and high imidazole buffer (500 mM imidazole, 300 mM NaCl, 50 mM Na₂HPO₄). The proteins were concentrated with Amicon® Ultra Centrifugal Filter, 50 kDa MWCO (UFC905008, Millipore Sigma), and stored at 4 °C for a short time before use. For long-term storage, they were aliquoted in the storage buffer (25 mM Tris-HCl, pH 7.5, 300 mM NaCl, 10% glycerol, and 2 mM DTT) and kept at -80 °C to prevent degradation from repeated freeze-thaw cycles.

Recombinant protein expression and purification from E. coli

The 6×His-DDX6-EGFP sequence was cloned into the pFN29A expression vector (Flexi backbone with T7 promoter, Promega) and transformed into KRX competent cells (Promega) via heat shock. Transformed colonies were selected on LB agar plates containing ampicillin and incubated overnight at 37 °C. A single colony was inoculated into 4 ml of LB medium supplemented with 0.4% glucose and ampicillin, and cultured overnight at 37 °C, 220 rpm. The overnight culture was diluted 1:100 into fresh LB medium containing ampicillin and induced the protein expression with 0.05% glucose and 0.05% L-rhamnose for 20 h at 20 °C, 220 rpm. Cells were harvested by centrifugation at 4000 x g for 20 min at 4 °C and then resuspended in purification buffer (50 mM Tris pH 7.5, 150 mM NaCl, 10% glycerol, and 1 μM DTT) containing 10 g/ml lysozyme, RQ1 RNase-Free DNase, and Protease Inhibitor Cocktail (Promega). Following a 10-minute incubation on ice, cells were then sonicated with a microtip sonicator (Model 705 Sonic Dismembrator, Fisher Scientific) with an amplitude of 24W for 15 min (10 s on, 10 s off). Lysate was clarified by centrifugation at 15,000 x g for 30 min at 4 °C and then filtered by a 0.45 μm column. The resulting supernatant was subjected to affinity purification using HisPur™ Ni-NTA Resin (88221, Thermo Fisher Scientific) according to the manufacturer’s instructions. The recombinant proteins were eluted from the HisPur resin columns using high imidazole buffer (250 mM imidazole, 300 mM NaCl, 50 mM Na₂HPO₄). For further purifications, a HiLoad Superdex 200 pg SEC column was used to increase the target protein purity one by one. The final protein fractions were concentrated with Amicon® Ultra Centrifugal Filter, 50 kDa MWCO (UFC905008, Millipore Sigma), and stored at 4 °C for a short time to use. For long-term storage, they were aliquoted in the storage buffer (25 mM Tris-HCl, pH 7.5, 300 mM NaCl, 10% glycerol, and 2 mM DTT) and kept at -80 °C to prevent degradation from repeated freeze-thaw cycles. The purified DDX6 protein was subsequently analyzed using appropriate instrumentation.

In vitro LLPS assay

The in vitro LLPS droplet was set up at room temperature in a total volume of 10 μl within a PCR tube. The concentrations of purified recombinant proteins were determined using the Bio-Rad Protein Assay Kit I (5000006, Bio-Rad). Subsequently, the proteins were diluted to specified concentrations with varying volumes of 5 M NaCl and water to achieve the indicated final NaCl concentration. 5 μl of the protein at various concentrations was transferred to microscope slides and incubated at room temperature for 5 minutes. Then, cover glasses were placed on the slides, and images were captured using a Zeiss LSM 880 confocal microscope equipped with a 63x oil immersion lens. To assess the impact of RNA presence on LLPS droplet assembly, 1 μl of ATTO^TM 590 (NHS Ester) -poly(U)-RNA (2 mg/ml in water; IDT), ATTO^TM 425 (NHS Ester) -poly(C)-RNA (2 mg/ml in water; IDT), poly(U)-RNA (2 mg/ml in water; P9528, Sigma) or poly(C)-RNA (2 mg/ml in water; P4903, Sigma) was added to 9 μl of the protein solution. 5 μl of the RNA and protein mixture was transferred to microscope slides and incubated at room temperature for 5 minutes. The images were captured as detailed above. ImageJ was used to analyze the integrated intensity of each type of LLPS droplet.

CLIP and qPCR

Briefly, one 150 mm dish cells at 80% confluency were washed with ice-cold PBS, crosslinked with 254 nm ultraviolet light (150 mJ/cm²; ×3), harvested, and resuspended in 2 ml M-PER buffer (PI78501, Fisher Scientific) supplemented with RNase inhibitor (final concentration: 1 U/μl; EO0381, Thermo Fisher Scientific) and protease inhibitor (final concentration: 1×). The cells were lysed on ice for 20 minutes with frequent mixing, followed by sonication. Then, the cell lysates were centrifuged at 12,000 g for 10 minutes at 4 °C. The supernatants were collected and pre-cleared using Protein A/G magnetic beads (88803, Thermo Fisher Scientific). One-tenth of the supernatants were saved as the input, while the remainder were subjected to IP. For IP, Flag (8 μg for each assay; F3165, Sigma-Aldrich) or IgG (12-371, Sigma) antibodies were conjugated to Protein A/G magnetic beads by gentle rotation at 4 °C for 4 h, followed by three washes with RIP buffer (150 mM KCl, 25 mM Tris-HCl pH 7.4, 5 mM EDTA, 0.5 mM dithiothreitol [DTT], 0.5% NP40, 100 U/ml RNase inhibitor, and 1× protease inhibitor). The antibody-conjugated beads were incubated overnight with the pre-cleared lysates with rotation at 4 °C. After three washes with RIP buffer, the beads were dispersed in 80 μl PBS, followed by DNase I (EN0521, Thermo Fisher Scientific) digestion and proteinase K (E00492, Thermo Fisher Scientific) digestion at 37 °C for 15 min. Both the input and CLIP RNA were recovered with Direct-zol RNA Miniprep Kits (R2052, ZYMO Research). Subsequently, these RNAs underwent reverse transcription and qPCR as detailed above (see “cDNA synthesis and qPCR”). All the primers used for CLIP-qPCR are listed in Supplementary Data. S13.

Determination of cell proliferation via MTT assay

In vitro cell proliferation was assessed using a TACS MTT Cell Proliferation Assay (4890-050-K, R&D Systems). Briefly, cells were seeded into 96-well non-tissue-culture plates at a density of 5000 to 10,000 cells per well. At each time point, 10 μl of MTT solution was added to each well. After incubation at 37 °C for 3 to 4 h, 100 μl of Detergent Reagent (4890-25-02, R&D Systems) was added to solubilize the formazan product. Finally, absorbance was measured at 570 nm using the BioTek Gen5 system (BioTek, USA).

Flow cytometry analysis

To evaluate myeloid differentiation status, approximately 0.1–0.5 × 10⁶ leukemia cells were harvested, washed with ice-cold PBS, and resuspended in 100 μl flow cytometry staining buffer (00-4222-26, eBioscience). The cells were stained on ice for 30 min with the following antibodies: PE-CD11b (50-112-2210, Fisher Scientific) and APC-CD14 (17-0149-42, Thermo Fisher Scientific). After staining, the cells were washed with ice-cold PBS and analyzed on a BD LSRFortessa X-20 cell analyzer (BD Biosciences, USA).

To assess cell cycle, ~0.1–0.5 × 10⁶ cells were harvested and resuspended in 1 ml of culture medium containing 10 μg/ml Hoechst33342 (BDB561908, Fisher Scientific). After incubation at 37 °C for 45 min, 5 μl of 100 μg/ml Pyronin Y (AAJ6106803, Fisher Scientific) was added directly to the cell suspension, followed by an additional 15 minutes of incubation at 37 °C. The cells were immediately analyzed on BD LSRFortessa X-20.

The BD Annexin V: PE Apoptosis Detection Kit I (BDB559763, Fisher Scientific) was utilized to evaluate cell apoptosis according to the manufacturer’s instructions. Briefly, leukemia cells following DDX6 KO or Ara-C treatment (50 nM for 96 h) were harvested, washed twice with PBS, suspended in Annexin V Binding Buffer, and stained with Annexin V-PE, 7-AAD for 15 min at room temperature in the dark. The cells were immediately analyzed on BD LSRFortessa X-20. All the flow cytometry results were analyzed using FlowJo software (Supplementary Fig. 12).

Wright-Giemsa staining

The Wright-Giemsa Stain (24985-1, Polysciences) and Wright-Giemsa Stain Phosphate Buffer pH 6.8 (24984-1, Polysciences) were utilized for Wright-Giemsa staining BM or PB smears according to the manufacturer’s instructions. After staining, all the slides were washed with flowing water, and images were captured using a Zeiss LSM Widefield microscope with a 40× or 100× oil lens.

In vitro colony-forming and replating assay

Mouse BM lineage-negative (Lin^-) hematopoietic stem cells (HSCs) were enriched from 6- to 8- week-old Cas9 knock-in B6 mice (RRID: IMSR_JAX:026179, The Jackson Laboratory) following treatment with 150 mg/kg 5-fluorouracil (5-FU) for 5 days, using the Lineage Cell Depletion Kit (130-090-858, Miltenyi Biotec). The Lin^- HSCs were then virally transduced with MLL-AF9 and sgRNAs targeting Ddx6 or a scrambled control. The cells were seeded at a density of 2 × 10⁴ per 35 mm culture dish (27150, STEMCELL Technologies) in ColonyGEL (1201, ReachBio Research Labs). The ColonyGEL was supplemented with 10 ng/ml of murine recombinant mouse IL-3, IL-6, GM-CSF, and 100 ng/ml of murine recombinant SCF, 1 mg/ml of G418 (10131027, Thermo Fisher Scientific), and 200 μg/ml hygromycin B. The colonies were cultured at 37 °C in a humidified atmosphere with 5% CO₂ for 6 to 7 days. Serial replating was conducted by collecting and replating colony cells every 7 days. Colony numbers were enumerated and compared for each passage.

Isolation of DDX6 PBs

The DDX6 PBs were isolated from whole-cell lysate through continuous centrifugation followed by pulldown assay with Flag antibody (recognizing Flag-DDX6 protein) as described previously^3,79,80 with minor modifications. HET293T cells overexpressing DDX6 were transfected with the pSIN4-Flag-DDX6 plasmid. After appropriate treatments (500 μM NaAsO₂ treatment 1 h), five 15 cm-dish HEK293T cells were washed twice with ice-cold PBS quickly (within 10 s) and then harvested in 5 ml ice-cold lysis buffer (50 mM Tris pH 7.6, 50 mM NaCl, 5 mM MgCl₂, 0.1% NP-40,1 mM β-mercaptoethanol, 1% EDTA-free protease inhibitor cocktail, and 1 U/ml RNase inhibitor) on ice for 20 minutes. After that, the resuspended cells were passed through a 25 G 5/8 needle seven times while on ice, followed by centrifugation at 2000 g for 5 min at 4 °C. The supernatant contains a cytosolic fraction, while the pellet contains nuclear fraction. Then, 5 ml of the supernatant was centrifuged at 18,000 g for 20 min at 4 °C to isolate the insoluble pellet fraction. After discarding the supernatant, the pellet was re-suspended in 1 ml of lysis buffer and centrifuged at 18,000 g for another 20 minutes at 4 °C. Then, the supernatant was discarded, whereas the pellet was re-suspended in 300 μl lysis buffer and centrifuged at 850 g for 2 min at 4 °C. The supernatants represented core-enriched fractions of cytosolic RNP granules.

For further enrichment of DDX6 PBs, the supernatants were subjected to an additional IP assay with Flag (or IgG)-conjugated Protein A/G Magnetic Beads (88803, Thermo Fisher Scientific). The samples were rotated on a shaker overnight at 4 °C. Magnet-separate Protein A/G Magnetic Beads and remove supernatant. The Protein A/G beads were washed three times with wash buffer 1 (20 mM Tris HCl pH 8.0, 200 mM NaCl, 1% EDTA-free protease inhibitor cocktail, and 1 U/ml RNase inhibitor) at 4 °C for 5 minutes, once with 1 ml of Wash buffer 2 (20 mM Tris HCl pH 8.0, 500 mM NaCl, 1% EDTA-free protease inhibitor cocktail, and 1 U/ ml RNase inhibitor) at 4 °C for 5 minutes, and once with 1 ml of Wash buffer 3 (lysis buffer with 2 M Urea) at 4 °C for 2 min. The beads were resuspended in 80 μl PBS with Proteinase K (50 μg) (FEREO0491, Fisher Scientific), DNase I (1.6U) (FEREN0521, Fisher Scientific), and digested at 37 °C for 15 minutes before RNA extraction with QIAzol (79306, Qiagen).

smFISH

The 48 BCAT1 FISH probes were designed using the Stellaris® RNA FISH Probe Designer (Biosearch Technologies, Inc.) available online at www.biosearchtech.com/stellarisdesigner. Samples were hybridized with the BCAT1 Stellaris RNA FISH Probe set labeled with Quasar® 570 dye (Biosearch Technologies, Inc.), following the manufacturer’s instructions. The smFISH assay was adapted from a previous publication⁸¹. Briefly, cells with overexpression of DDX6 (pCDH-DDX6-EGFP) in an 8-well chamber were treatment 2 h with ementine (100 µg/ml) or not, and then washed three times with PBSM (5 mM MgCl₂ in PBS), fixed with fixation buffer (4% formaldehyde in PBSM) for 10 min at room temperature, and quenched with 200 mM glycine (G7126, Sigma-Aldrich). The cells were washed with PBSM twice, permeabilized with 0.1% Triton™ X-100 Surfact-Amps™ Detergent Solution (85111, Thermo Fisher Scientific), and incubated with hybridization buffer plus BCAT1 probe set (final concentration: 250 nM) overnight at 37 °C. Subsequently, the cells were washed three times with pre-hybridization buffer and mounted with Fluoroshield® medium containing DAPI (F6057, Sigma-Aldrich). The images were captured LSM 880 confocal microscope (Zeiss, Germany).

RNA pull-down assay

A 3’ end biotin-labeled fragment of BCAT1 mRNA was synthesized by Horizon (Supplementary Data. S13). The in vitro (cell-free) RNA pull-down was conducted using Pierce Magnetic RNA Protein Pull Down Kit (20164, Thermo Fisher Scientific) following the manufacturer’s instructions. Briefly, 50 pmol 3’ end biotin-labeled BCAT1 mRNA were bound with 50 μl Dynabeads^TM-280 streptavidin (11205D, Thermo Fisher Scientific) in RNA capture buffer for 1 h at room temperature with agitation. Streptavidin beads-conjugated BCAT1 mRNA were then incubated with 2 μg/μl recombinant 6×His-DDX6-EGFP protein (purified from E. coli) in Protein-RNA binding buffer in a 100 μl final volume overnight at 4 °C. Finally, the beads were washed with washing buffer 3 times, and the RNA-protein complexes were eluted using Elution Buffer. Proteins were detected by Western blotting.

Cytoplasmic and nuclear protein extraction

Cytoplasmic and nuclear proteins were extracted from HEK293T cells overexpressing DDX6 and its truncated variants using the NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific, Cat# 78835). Briefly, cell pellets were resuspended in ice-cold CER I and incubated on ice for 10 minutes. CER II was then added, followed by vertexing and an additional 5-minute incubation on ice. The supernatant, containing the cytoplasmic fraction, was collected after centrifugation. The remaining pellet was resuspended in ice-cold NER and vortexed for 15 seconds every 10 minutes over a total of 40 minutes to extract the nuclear fraction.

RNA stability assay

The cells were treated with 5 μg/ml actinomycin D (A9415, Sigma-Aldrich) for the indicated time points to assess mRNA stability. Total RNA was extracted from cells using QIAzol reagent for RT-qPCR analysis. The turnover rate and half-time of mRNA were estimated. The mRNA degradation rate k_decay was estimated by:

$${{In}}\left({{C}}/{{C}}_{0}\right)=- {{K}}_{{decay}} {{t}}$$

(1)

C₀ is the concentration of mRNA at time 0, t is the transcription inhibition time, and C is the mRNA concentration at time t. Thus, the K_decay can be derived by the exponential decay fitting of C/C0 versus time t. The half-time (t_1/2), which means C/C0 = 1/2, can be calculated by the following equation:

$${{In}}\left({1}/{2}\right)=- {{K}}_{{decay}} {{t}}_{1/2}$$

(2)

The rearrangement of the above equation leads to the mRNA half-life time value, t_1/2 = ln2/K_decay.

Measurements of OCR using the XFe96 extracellular flux analyzer

Mitochondrial OCR was measured using a Seahorse XFe96 Analyzer with Seahorse XFe96 FluxPaks Kit (103793-100, Agilent Technologies) according to the manufacturer’s instructions. In brief, Seahorse plates were coated overnight with 50 µg/ml Poly-D-Lysine (A-003-M, Sigma). Then, cells were seeded in the coated 96-well Seahorse plates at a density of 0.5–1 × 10⁵ cells/well in XF RPMI medium (103576-100, Agilent) supplemented with 2 mM glutamine, 10 mM glucose, 1 mM Sodium Pyruvate. These cells were incubated in a CO₂-free incubator for 30 min prior to Seahorse assay. To assess mitochondrial respiratory activity, Oligomycin (O4876-5MG, Sigma), FCCP (C2920, Sigma), and Rotenone (R8875-1G, Sigma)/Antimycin A (A8674-25MG, Sigma) were added through ports A, B, and C at final concentrations of 2 μM, 2 μM, and 0.5 μM, respectively. All data were normalized to cell numbers.

Phenotypic analysis of hematopoiesis

Both male and female Ddx6^fl/fl;Mx1-cre, Ddx6^fl/+;Mx1-cre, and Ddx6^+/+;Mx1-cre mice were used for the experiments. Briefly, PB samples were collected and analyzed using an Element HT5 (HESKA) according to the manufacturer’s protocol. BM or spleen cells were collected, lysed with ACK LYSING Buffer (VWR) to remove red blood cells, and then subjected to flow cytometry staining using the following antibodies: mouse hematopoietic lineage antibody cocktail (88-7772-72, Thermo Fisher Scientific; 1:1000), Sca-1 (12-5981-82, Thermo Fisher Scientific; 1:1000), c-Kit⁺ (17-1171-82, Thermo Fisher Scientific; 1:1000), CD3 (17A2, Biolegend; 1:1000), Ter119 (560504, BD Biosciences; 1:1000), and B220 (561102, BD Biosciences; 1:1000), Mac-1 (562605, BD Biosciences; 1:1000), and Gr-1 (45-5931-80, Thermo Fisher Scientific; 1:1000). After staining, the cells were resuspended in FACS buffer before being analyzed using BD LSRFortessa X-20.

Leucine tracing

For stable-isotope tracing experiments, cells were plated in 6-well plates at a density of 1 × 10⁶ cells/well in 2 ml of pre-warmed RPMI (SKU:091629149, MP Biomedicals) supplemented with ¹³C, ¹⁵N-Leucine (CCN1600P01, CortecNet) instead of the regular leucine, and cultured for 24 h. After incubation, the cells were collected, rinsed with ice-cold 5% mannitol, and quenched with 1 ml 80% methanol (−80 °C). The samples were vigorously vortexed at least three times for 30 min on ice, then spun down at 16,000 g for 5 minutes at 4 °C. The cleared supernatant was transferred to a new Eppendorf tube (022600028, Eppendorf) and evaporated under vacuum at 30 °C using a Vacufuge plus Centrifuge Concentrator (Eppendorf). The extracted metabolites were stored at -80 °C. Mass spectrometry-based analysis of the extracted metabolites was conducted at the UCLA Metabolomics Center.

RNA-seq and data analysis

Total RNAs were extracted from leukemia cells with and without DDX6 KO using QIAzol Lysis Reagent (79306, QIAGEN) according to the manufacturer’s instructions. RNA concentration was measured by NanoDrop 2000 (Thermo Fisher Scientific), and the RNA integrity was determined by Bioanalyzer (Agilent). Then, polyA⁺ RNAs were enriched from total RNAs, and library preparation was performed using the KAPA Stranded mRNA-Seq Kit (Kapa Biosystems). Samples were sequenced on an Illumina NovaSeq platform with paired-end reads (PE100).

After sequencing, reads were trimmed for adaptor sequence, masked for low-quality sequence by Cutadapt, and then mapped to the GRCh38 reference genome by STAR. Gene expression (RPKM) was calculated by RSEM. Gene Set Enrichment Analysis (GSEA) with differentially expressed genes was performed to identify enriched pathways. Differentially expressed genes between the two groups were determined using a double threshold on gene expression changes and associated statistical significances.

Statistical analysis

Data are presented as mean ± standard error of the mean (SEM). Statistical comparisons were performed using two-tailed t-tests, as specified in the figure legends. Pearson tests were performed for correlation analysis. P values less than 0.05 were considered statistically significant. All statistical analyses were done using GraphPad Prism 10 (GraphPad Software). The intensity of fluorescence was analyzed using ZEN Black software (Zeiss). The number of replicates for each experiment is indicated in the figure legends. All Western blotting and immunofluorescence images are representative of three independent experiments.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All the sequencing data generated in this study have been deposited in NCBI’s Gene Expression Omnibus under the accession numbers GSE277986 and GSE310106. The sgRNAs were designed with CRISPick (https://portals.broadinstitute.org/gppx/crispick/public). The raw data for pan-cancer analysis was derived from cBioPortal (https://www.cbioportal.org/). The raw data for the DepMap CRISPR (CERES) score was derived from the Depmap portal (https://depmap.org/portal/). The 101 SG/PB-enriched RBPs were derived from the RNA granule database (http://rnagranuledb.lunenfeld.ca/). The relative expression of DDX6 in the AML subtype derived from the BloodSpot (https://www.fobinf.com/). IDR domain prediction of DDX6 was used with IUPred3 (https://iupred.elte.hu/plot). The Sankey diagram was plotted with SnakeyMatic (https://sankeymatic.com/). The metabolic analysis was conducted with the MetaboAnalyst (https://www.metaboanalyst.ca/MetaboAnalyst/). The prediction of protein–mRNA binding sequences was performed using RPISeq (http://pridb.gdcb.iastate.edu/RPISeq/results.php). Source data are provided with this paper.

References

Lyon, A. S., Peeples, W. B. & Rosen, M. K. A framework for understanding the functions of biomolecular condensates across scales. Nat. Rev. Mol. Cell Biol. 22, 215–235 (2021).
Article PubMed Google Scholar
Yang, P. et al. G3BP1 is a tunable switch that triggers phase separation to assemble stress granules. Cell 181, 325–345 e328 (2020).
Article PubMed PubMed Central Google Scholar
Hubstenberger, A. et al. P-Body Purification Reveals the Condensation of Repressed mRNA Regulons. Mol. Cell 68, 144–157 e145 (2017).
Article PubMed Google Scholar
Brengues, M., Teixeira, D. & Parker, R. Movement of eukaryotic mRNAs between polysomes and cytoplasmic processing bodies. Science 310, 486–489 (2005).
Article ADS PubMed PubMed Central Google Scholar
Ivanov, P., Kedersha, N. & Anderson, P. Stress granules and processing bodies in translational control. Cold Spring Harb Perspect Biol 11 (2019).
Buchan, J. R. & Parker, R. Eukaryotic stress granules: the ins and outs of translation. Mol. Cell 36, 932–941 (2009).
Article PubMed PubMed Central Google Scholar
Sheth, U. & Parker, R. Decapping and decay of messenger RNA occur in cytoplasmic processing bodies. Science 300, 805–808 (2003).
Article ADS PubMed PubMed Central Google Scholar
Bhattacharyya, S. N., Habermacher, R., Martine, U., Closs, E. I. & Filipowicz, W. Relief of microRNA-mediated translational repression in human cells subjected to stress. Cell 125, 1111–1124 (2006).
Article PubMed Google Scholar
Eulalio, A., Behm-Ansmant, I., Schweizer, D. & Izaurralde, E. P-body formation is a consequence, not the cause, of RNA-mediated gene silencing. Mol. Cell. Biol. 27, 3970–3981 (2007).
Article PubMed PubMed Central Google Scholar
Horvathova, I. et al. The dynamics of mRNA turnover revealed by single-molecule imaging in single cells. Mol. Cell 68, 615–625.e619 (2017).
Article PubMed Google Scholar
Hope, K. J., Jin, L. & Dick, J. E. Acute myeloid leukemia originates from a hierarchy of leukemic stem cell classes that differ in self-renewal capacity. Nat. Immunol. 5, 738–743 (2004).
Article PubMed Google Scholar
Goardon, N. et al. Coexistence of LMPP-like and GMP-like leukemia stem cells in acute myeloid leukemia. Cancer Cell 19, 138–152 (2011).
Article PubMed Google Scholar
Döhner, H., Weisdorf, D. J. & Bloomfield, C. D. Acute myeloid leukemia. N. Engl. J. Med. 373, 1136–1152 (2015).
Article PubMed Google Scholar
Molica, M., Breccia, M., Foa, R., Jabbour, E. & Kadia, T. M. Maintenance therapy in AML: The past, the present and the future. Am. J. Hematol. 94, 1254–1265 (2019).
Article PubMed Google Scholar
Short, N. J. et al. Advances in the treatment of acute myeloid leukemia: new drugs and new challenges. Cancer Discov. 10, 506–525 (2020).
Article PubMed Google Scholar
Dohner, H. et al. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet. Blood 115, 453–474 (2010).
Article PubMed Google Scholar
Shallis, R. M., Wang, R., Davidoff, A., Ma, X. & Zeidan, A. M. Epidemiology of acute myeloid leukemia: Recent progress and enduring challenges. Blood Rev. 36, 70–87 (2019).
Article PubMed Google Scholar
DeWolf, S. & Tallman, M. S. How I treat relapsed or refractory AML. Blood 136, 1023–1032 (2020).
Article PubMed PubMed Central Google Scholar
Thol, F. & Ganser, A. Treatment of relapsed acute myeloid leukemia. Curr. Treat. Options Oncol. 21, 66 (2020).
Article PubMed PubMed Central Google Scholar
Lavalée, M., Curdy, N., Laurent, C., Fournié, J. J. & Franchini, D. M. Cancer cell adaptability: turning ribonucleoprotein granules into targets. Trends Cancer 7, 902–915 (2021).
Article PubMed Google Scholar
Yao, L. et al. KLHL29-mediated DDX3X degradation promotes chemosensitivity by abrogating cell cycle checkpoint in triple-negative breast cancer. Oncogene 42, 3514–3528 (2023).
Article PubMed PubMed Central Google Scholar
Heerma van Voss, M. R. et al. Targeting mitochondrial translation by inhibiting DDX3: a novel radiosensitization strategy for cancer treatment. Oncogene 37, 63–74 (2018).
Article PubMed Google Scholar
Ward, P. S. & Thompson, C. B. Metabolic reprogramming: a cancer hallmark even Warburg did not anticipate. Cancer Cell 21, 297–308 (2012).
Article PubMed PubMed Central Google Scholar
Raffel, S. et al. BCAT1 restricts alphaKG levels in AML stem cells leading to IDHmut-like DNA hypermethylation. Nature 551, 384–388 (2017).
Article ADS PubMed Google Scholar
Jones, C. L., Inguva, A. & Jordan, C. T. Targeting energy metabolism in cancer stem cells: progress and challenges in leukemia and solid tumors. Cell Stem Cell 28, 378–393 (2021).
Article PubMed PubMed Central Google Scholar
Jones, C. L. et al. Inhibition of amino acid metabolism selectively targets human leukemia stem cells. Cancer Cell 34, 724–740 e724 (2018).
Article PubMed PubMed Central Google Scholar
Ye, H. et al. Leukemic stem cells evade chemotherapy by metabolic adaptation to an adipose tissue niche. Cell Stem Cell 19, 23–37 (2016).
Article PubMed PubMed Central Google Scholar
Baccelli, I. et al. Mubritinib targets the electron transport chain complex I and reveals the landscape of oxphos dependency in acute myeloid leukemia. Cancer Cell 36, 84–99.e88 (2019).
Article PubMed Google Scholar
Carter, J. L. et al. Targeting mitochondrial respiration for the treatment of acute myeloid leukemia. Biochem Pharm. 182, 114253 (2020).
Article PubMed Google Scholar
Amaya, M. L. et al. The STAT3-MYC axis promotes survival of leukemia stem cells by regulating SLC1A5 and oxidative phosphorylation. Blood 139, 584–596 (2022).
Article PubMed PubMed Central Google Scholar
Pan, R. et al. Selective BCL-2 inhibition by ABT-199 causes on-target cell death in acute myeloid leukemia. Cancer Discov. 4, 362–375 (2014).
Article PubMed Google Scholar
DiNardo, C. D. et al. Clinical experience with the BCL2-inhibitor venetoclax in combination therapy for relapsed and refractory acute myeloid leukemia and related myeloid malignancies. Am. J. Hematol. 93, 401–407 (2018).
Article PubMed Google Scholar
Bosc, C. et al. Mitochondrial inhibitors circumvent adaptive resistance to venetoclax and cytarabine combination therapy in acute myeloid leukemia. Nat. Cancer 2, 1204–1223 (2021).
Article PubMed Google Scholar
Yen, K. et al. AG-221, a first-in-class therapy targeting acute myeloid leukemia harboring oncogenic IDH2 mutations. Cancer Discov. 7, 478–493 (2017).
Article PubMed Google Scholar
Han, L. et al. METTL16 drives leukemogenesis and leukemia stem cell self-renewal by reprogramming BCAA metabolism. Cell Stem Cell 30, 52–68.e13 (2023).
Article PubMed PubMed Central Google Scholar
Gu, Z. et al. Loss of EZH2 Reprograms BCAA metabolism to drive leukemic transformation. Cancer Discov. 9, 1228–1247 (2019).
Article PubMed PubMed Central Google Scholar
Hattori, A. et al. Cancer progression by reprogrammed BCAA metabolism in myeloid leukaemia. Nature 545, 500–504 (2017).
Article ADS PubMed PubMed Central Google Scholar
Tönjes, M. et al. BCAT1 promotes cell proliferation through amino acid catabolism in gliomas carrying wild-type IDH1. Nat. Med. 19, 901–908 (2013).
Article PubMed PubMed Central Google Scholar
Youn, J. Y. et al. Properties of stress granule and P-body proteomes. Mol. Cell 76, 286–294 (2019).
Article PubMed Google Scholar
Millar, S. R. et al. A new phase of networking: the molecular composition and regulatory dynamics of mammalian stress granules. Chem. Rev. 123, 9036–9064 (2023).
Article ADS PubMed PubMed Central Google Scholar
Tsherniak, A. et al. Defining a cancer dependency map. Cell 170, 564–576.e516 (2017).
Article PubMed PubMed Central Google Scholar
Chen, Z. et al. Screening membraneless organelle participants with machine-learning models that integrate multimodal features. Proc. Natl. Acad. Sci. USA 119, e2115369119 (2022).
Article PubMed PubMed Central Google Scholar
Ferraiuolo, M. A. et al. A role for the eIF4E-binding protein 4E-T in P-body formation and mRNA decay. J. Cell Biol. 170, 913–924 (2005).
Article PubMed PubMed Central Google Scholar
Safieddine, A. et al. Cell-cycle-dependent mRNA localization in P-bodies. Mol. Cell 84, 4191–4208.e4197 (2024).
Article PubMed Google Scholar
Ripin, N., Macedo de Vasconcelos, L., Ugay, D. A. & Parker, R. DDX6 modulates P-body and stress granule assembly, composition, and docking. J. Cell Biol. 223 (2024).
Maneix, L. et al. Cyclophilin A supports translation of intrinsically disordered proteins and affects haematopoietic stem cell ageing. Nat. Cell Biol. 26, 593–603 (2024).
Article PubMed PubMed Central Google Scholar
Ayache, J. et al. P-body assembly requires DDX6 repression complexes rather than decay or Ataxin2/2L complexes. Mol. Biol. Cell 26, 2579–2595 (2015).
Article PubMed PubMed Central Google Scholar
Luo, Y., Na, Z. & Slavoff, S. A. P-Bodies: Composition, Properties, and Functions. Biochemistry 57, 2424–2431 (2018).
Article PubMed Google Scholar
Brocca, S., Grandori, R., Longhi, S. & Uversky, V. Liquid-Liquid Phase Separation by Intrinsically Disordered Protein Regions of Viruses: Roles in Viral Life Cycle and Control of Virus-Host Interactions. Int. J. Mol. Sci. 21 (2020).
Wunderlich, M. et al. AML xenograft efficiency is significantly improved in NOD/SCID-IL2RG mice constitutively expressing human SCF, GM-CSF and IL-3. Leukemia 24, 1785–1788 (2010).
Article PubMed PubMed Central Google Scholar
Shen, H. et al. Sexually dimorphic RNA helicases DDX3X and DDX3Y differentially regulate RNA metabolism through phase separation. Mol. Cell 82, 2588–2603.e2589 (2022).
Article PubMed PubMed Central Google Scholar
Chahar, H. S., Chen, S. & Manjunath, N. P-body components LSM1, GW182, DDX3, DDX6 and XRN1 are recruited to WNV replication sites and positively regulate viral replication. Virology 436, 1–7 (2013).
Article PubMed Google Scholar
Zhao, W. et al. POSTAR3: an updated platform for exploring post-transcriptional regulation coordinated by RNA-binding proteins. Nucleic Acids Res. 50, D287–d294 (2022).
Article MathSciNet PubMed Google Scholar
Rouya, C. et al. Human DDX6 effects miRNA-mediated gene silencing via direct binding to CNOT1. RNA 20, 1398–1409 (2014).
Article PubMed PubMed Central Google Scholar
Chen, Y. et al. A DDX6-CNOT1 complex and W-binding pockets in CNOT9 reveal direct links between miRNA target recognition and silencing. Mol. Cell 54, 737–750 (2014).
Article PubMed Google Scholar
Courel, M. et al. GC content shapes mRNA storage and decay in human cells. Elife 8 (2019).
Zhu, Z. et al. Tumour-reprogrammed stromal BCAT1 fuels branched-chain ketoacid dependency in stromal-rich PDAC tumours. Nat. Metab. 2, 775–792 (2020).
Article PubMed PubMed Central Google Scholar
Yoo, H. C. & Han, J. M. Amino acid metabolism in cancer drug resistance. Cells 11 (2022).
Wang, Y. et al. Branched-chain amino acid metabolic reprogramming orchestrates drug resistance to EGFR tyrosine kinase inhibitors. Cell Rep. 28, 512–525.e516 (2019).
Article PubMed Google Scholar
Assouline, S. et al. Molecular targeting of the oncogene eIF4E in acute myeloid leukemia (AML): a proof-of-principle clinical trial with ribavirin. Blood 114, 257–260 (2009).
Article PubMed Google Scholar
Ghashghaei, M. et al. Translation efficiency driven by CNOT3 subunit of the CCR4-NOT complex promotes leukemogenesis. Nat. Commun. 15, 2340 (2024).
Article ADS PubMed PubMed Central Google Scholar
Ghashghaei, M. et al. miR-148a-3p and DDX6 functional link promotes survival of myeloid leukemia cells. Blood Adv. 7, 3846–3861 (2023).
Article PubMed Google Scholar
Kodali, S. et al. RNA sequestration in P-bodies sustains myeloid leukaemia. Nat. Cell Biol. (2024).
Bah, A. & Forman-Kay, J. D. Modulation of Intrinsically Disordered Protein Function by Post-translational Modifications. J. Biol. Chem. 291, 6696–6705 (2016).
Article PubMed PubMed Central Google Scholar
Buchan, J. R. mRNP granules. Assembly, function, and connections with disease. RNA Biol. 11, 1019–1030 (2014).
Article PubMed PubMed Central Google Scholar
Majerciak, V., Zhou, T., Kruhlak, M. ichaelJ. & Zheng, Z.-M. RNA helicase DDX6 and scaffold protein GW182 in P-bodies promote biogenesis of stress granules. Nucleic Acids Res. 51, 9337–9355 (2023).
Article PubMed PubMed Central Google Scholar
Hondele, M. et al. DEAD-box ATPases are global regulators of phase-separated organelles. Nature 573, 144–148 (2019).
Article ADS PubMed PubMed Central Google Scholar
Di Stefano, B. et al. The RNA helicase DDX6 controls cellular plasticity by modulating P-body homeostasis. Cell Stem Cell 25, 622–638.e613 (2019).
Article PubMed PubMed Central Google Scholar
Gong, C. et al. Sequential inverse dysregulation of the RNA helicases DDX3X and DDX3Y facilitates MYC-driven lymphomagenesis. Mol. Cell 81, 4059–4075.e4011 (2021).
Article PubMed Google Scholar
Botlagunta, M. et al. Oncogenic role of DDX3 in breast cancer biogenesis. Oncogene 27, 3912–3922 (2008).
Article PubMed PubMed Central Google Scholar
Bol, G. M. et al. Targeting DDX3 with a small molecule inhibitor for lung cancer therapy. EMBO Mol. Med 7, 648–669 (2015).
Article PubMed PubMed Central Google Scholar
Korobeynikov, V. A., Lyashchenko, A. K., Blanco-Redondo, B., Jafar-Nejad, P. & Shneider, N. A. Antisense oligonucleotide silencing of FUS expression as a therapeutic approach in amyotrophic lateral sclerosis. Nat. Med 28, 104–116 (2022).
Article PubMed PubMed Central Google Scholar
Freibaum, B. D. et al. Identification of small molecule inhibitors of G3BP-driven stress granule formation. J. Cell Biol. 223 (2024).
Chan, A. K. N. et al. Therapeutic targeting Tudor domains in leukemia via CRISPR-Scan Assisted Drug Discovery. Sci. Adv. 10, eadk3127 (2024).
Article PubMed PubMed Central Google Scholar
Xue, M. et al. METTL16 promotes liver cancer stem cell self-renewal via controlling ribosome biogenesis and mRNA translation. J. Hematol. Oncol. 17, 7 (2024).
Article PubMed PubMed Central Google Scholar
Li, M. et al. Epigenetic control of translation checkpoint and tumor progression via RUVBL1-EEF1A1 Axis. Adv. Sci. 10, e2206584 (2023).
Article ADS Google Scholar
Xu, X. et al. ACTR5 controls CDKN2A and tumor progression in an INO80-independent manner. Sci. Adv. 8, eadc8911 (2022).
Article PubMed PubMed Central Google Scholar
Yang, L. et al. High-resolution characterization of gene function using single-cell CRISPR tiling screen. Nat. Commun. 12, 4063 (2021).
Article ADS PubMed PubMed Central Google Scholar
Zhao, Z. et al. QKI shuttles internal m(7)G-modified transcripts into stress granules and modulates mRNA metabolism. Cell 186, 3208–3226.e3227 (2023).
Article PubMed PubMed Central Google Scholar
Zou, Z., Sepich-Poore, C., Zhou, X., Wei, J. & He, C. The mechanism underlying redundant functions of the YTHDF proteins. Genome Biol. 24, 17 (2023).
Article PubMed PubMed Central Google Scholar
Haimovich, G. & Gerst, J. E. Single-molecule Fluorescence in situ Hybridization (smFISH) for RNA Detection in Adherent Animal Cells. Bio Protoc. 8, e3070 (2018).
Article PubMed PubMed Central Google Scholar

Download references

Acknowledgements

The work is supported by the US National Institutes of Health (NIH) grants (R01CA296394, R37CA292678, and U54CA285116 [to R.S.]) and R37CA233691, U54CA243124, and R01CA278050 [to C.-W.C.]), Margaret E. Early Medical Research Trust award (to R.S.), Leukemia Research Foundation award (to R.S.), and St. Baldrick’s Foundation Scholar award (to R.S.). Dr. Rui Su is a Double Deckers Destroy AML St. Baldrick’s Scholar.

Author information

These authors contributed equally: Hongjie Bi, Wei Li, Lili Ren.

Authors and Affiliations

Department of Systems Biology, Beckman Research Institute of City of Hope, Monrovia, CA, USA
Hongjie Bi, Wei Li, Lili Ren, Honghai Zhang, Lei Dong, Anthony Chan, Xueer Wang, Lu Yang, Xiaoxu Zhang, Meilin Xue, Jianjun Chen, Chun-Wei Chen & Rui Su
Center for RNA Biology and Therapeutics, City of Hope Comprehensive Cancer Center, Duarte, CA, USA
Hongjie Bi, Wei Li, Lili Ren, Honghai Zhang, Xueer Wang, Xiaoxu Zhang, Jianjun Chen & Rui Su
Department of Pathology, Harbin Medical University, Harbin, China
Hongjie Bi, Lili Ren & Xiaobo Li
Quantitative Biomedical Research Center, Peter O’Donnell Jr. School of Public Health, UT Southwestern Medical Center, Dallas, TX, USA
Lei Dong
Division of Epigenetic and Transcriptional Engineering, Beckman Research Institute, City of Hope, Duarte, CA, USA
Anthony Chan, Lu Yang & Chun-Wei Chen
Laboratory of Gene Editing and Cellular Engineering, Biochemistry and Molecular Biology, Mayo Clinic, Rochester, MN, USA
Anthony Chan & Chun-Wei Chen
The Integrative Genomics Core, Beckman Research Institute of City of Hope, Duarte, CA, USA
Hanjun Qin & Xiwei Wu
UCLA Metabolomics Center, Department of Molecular and Medical Pharmacology, University of California, Los Angeles, CA, USA
Johanna Ten Hoeve-Scott
Department of Hematological Malignancies Translational Science, Gehr Family Center for Leukemia Research, City of Hope Medical Center and Beckman Research Institute, Duarte, CA, USA
Bin Zhang & Ling Li
Division of Experimental Hematology and Cancer Biology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA
Mark Wunderlich & James C. Mulloy
Department of Hematology and Hematopoietic Cell Transplantation, City of Hope, Duarte, CA, USA
Steven T. Rosen

Authors

Hongjie Bi
View author publications
Search author on:PubMed Google Scholar
Wei Li
View author publications
Search author on:PubMed Google Scholar
Lili Ren
View author publications
Search author on:PubMed Google Scholar
Honghai Zhang
View author publications
Search author on:PubMed Google Scholar
Lei Dong
View author publications
Search author on:PubMed Google Scholar
Anthony Chan
View author publications
Search author on:PubMed Google Scholar
Xueer Wang
View author publications
Search author on:PubMed Google Scholar
Lu Yang
View author publications
Search author on:PubMed Google Scholar
Xiaoxu Zhang
View author publications
Search author on:PubMed Google Scholar
Meilin Xue
View author publications
Search author on:PubMed Google Scholar
Hanjun Qin
View author publications
Search author on:PubMed Google Scholar
Xiwei Wu
View author publications
Search author on:PubMed Google Scholar
Johanna Ten Hoeve-Scott
View author publications
Search author on:PubMed Google Scholar
Bin Zhang
View author publications
Search author on:PubMed Google Scholar
Ling Li
View author publications
Search author on:PubMed Google Scholar
Mark Wunderlich
View author publications
Search author on:PubMed Google Scholar
James C. Mulloy
View author publications
Search author on:PubMed Google Scholar
Steven T. Rosen
View author publications
Search author on:PubMed Google Scholar
Jianjun Chen
View author publications
Search author on:PubMed Google Scholar
Xiaobo Li
View author publications
Search author on:PubMed Google Scholar
Chun-Wei Chen
View author publications
Search author on:PubMed Google Scholar
Rui Su
View author publications
Search author on:PubMed Google Scholar

Contributions

R.S. conceived the project and supervised the research; H.B. performed the majority of the experiments; C.-W.C. designed the sgRNA library and CRISPR screenings and supervised the research; A.C. performed CRISPR screening; L.D. performed bioinformatics analysis; H.Q. and X.W. contributed to the RNA-seq; H.Z., W.L., X.W., L.Y., L.R., X.Z., M.X., and R.S. performed experiments and/or data analysis; J.H.-S., B.Z., L.L., M.W., J.C.M., S.T.R., J.C., X.L., C.-W.C., and R.S. provided materials and contributed to discussions; H.B., C.-W.C., and R.S. wrote the manuscript and all authors commented on the manuscript.

Corresponding authors

Correspondence to Xiaobo Li, Chun-Wei Chen or Rui Su.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information (download PDF )

Description of Additional Supplementary Files (download PDF )

Supplementary Data 1 (download XLSX )

Supplementary Data 2 (download XLSX )

Supplementary Data 3 (download XLSX )

Supplementary Data 4 (download XLSX )

Supplementary Data 5 (download XLSX )

Supplementary Data 6 (download XLSX )

Supplementary Data 7 (download XLSX )

Supplementary Data 8 (download XLSX )

Supplementary Data 9 (download XLSX )

Supplementary Data 10 (download XLSX )

Supplementary Data 11 (download XLSX )

Supplementary Data 12 (download XLSX )

Supplementary Data 13 (download XLSX )

Reporting Summary (download PDF )

Transparent Peer Review file (download PDF )

Source data

Source data (download ZIP )

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Bi, H., Li, W., Ren, L. et al. DDX6 undergoes phase separation to modulate metabolic plasticity and chemoresistance. Nat Commun 17, 260 (2026). https://doi.org/10.1038/s41467-025-66966-4

Download citation

Received: 21 November 2024
Accepted: 19 November 2025
Published: 02 December 2025
Version of record: 08 January 2026
DOI: https://doi.org/10.1038/s41467-025-66966-4