HOXA9 is a master transcription factor of hematopoiesis and essential for maintaining self-renewal of leukemia stem cells (LSCs). Although mechanisms associated with its transcriptional regulation are extensively studied, how Hoxa9 mRNA translation is controlled remain poorly understood. We have previously identified the RNA binding protein (RBP) MUSASHI-2 (MSI2) as a central regulator of myeloid LSCs by controlling translation of crucial transcription factors including Hoxa9 (Park et al. JCI, 2015; Nguyen et al. Nature Comm, 2020). However, the mechanisms by which HOXA9 translation is modulated and the role of MSI2 remain unknown. To characterize the HOXA9 3'UTR mRNA regulon, we used it as a bait to perform an RNA-pull down followed by mass spectrometry (MS) with control or MSI2 depleted cells. To determine the direct MSI2's protein interactors on HOXA9, we also performed MSI2 co-immunoprecipitation (co-IP) MS in control or MSI2 deficient AML cells. We identified 10 RBPs that directly bind to MSI2 and are dependent on MSI2 for their binding to the HOXA9-3'UTR mRNA. Among these, we focused on SYNCRIP, a known MSI2 interactor (Vu et al. Nat Gen, 2017), HNRNPC and MOV10, of which binding ability on HOXA9 3'UTR was most reduced and enhanced, respectively, upon MSI2 loss. Reciprocal co-IP confirmed the interactions between these RBPs with MSI2. We then reasoned that if these RBPs are in the same regulatory complex with MSI2, they might phenocopy MSI2 loss of function in LSCs. To understand how each selected factor impacts LSC function and gene control, we first assessed MSI2's global regulation by performing integrative omic studies including MSI2-HyperTRIBE, RNA-seq and proteomics in MLL-AF9 murine LSCs that have acutely deleted (68hrs) of Msi2. We found that MSI2 binding promotes translation of 264 genes while suppresses 8 genes; and only 10 genes affected at mRNA level. These data found that MSI2 mediates translation of the self-renewal program in LSCs. We then found that SYNCRIP is required for both leukemia initiation and maintenance in the MLL-AF9 driven AML model with the Syncrip conditional knockout mice. Serial transplantation assays of Syncrip deleted cells show a continued delay in leukemogenesis in vivo, suggesting SYNCRIP is required for functional LSC activity. To identify SYNCRIP's targets, we performed SYNCRIP-HyperTRIBE and iCLIP, and found that majority (~64%) of SYNCRIP targets overlapped with MSI2 targets and their binding sites are proximity in LSCs. SYNCRIP knockout transcriptome enriched for HOXA9-MEIS1 targets, and MSI2 deficient HSC signature. These data suggest that SYNCRIP is required for LSC function by co-regulating HOXA9-associated program with MSI2. Furthermore, depleting HNRNPC or MOV10 reduces AML cell proliferation, colony formation and increases apoptosis. Both HNRNPC and MOV10 depletion significantly decreased HOXA9 protein abundance. HNRNPC knockout cells exhibited delayed leukemia development in vivo and serial plating assays shows that this RBP is required for maintaining LSC' self-renewal. Importantly, higher HNRNPC or MOV10 expression associated with a worse overall survival in AML patients (p=0.0026 for HNRNPC; p=0.0101 for MOV10). ENRICHR analysis shows HNRNPC correlatedly expressed genes in human AML patients enriched for HOXA9, MOV10 and MSI2 signature. These data suggest that HNRNPC and MOV10 co-regulate leukemia HOXA9-associated program with MSI2. We also discovered that these RBPs collaborate in the translational regulation mediated by MSI2. SYNCRIP is necessary for MSI2 recruitment to the HOXA9 3'UTR based on reduced MSI2 binding after SYNCRIP depletion (MSI2-HYPERTRIBE and MSI2-RNA-IP). Luciferase reporter assays with HOXA9-3'UTR found that knockdown each one of the components in the MSI2-SYNCRIP-HNRNPC-MOV10 complex alone significantly decreased luciferase signal. Moreover, loss of each of the core components SYNCRIP and HNRNPC exacerbated translation inhibition caused by MSI2 depletion and attenuated the increase in translation resulted from MSI2 overexpression. Our data suggest that SYNCRIP and HNRNPC facilitate MSI2-mediated translation regulation of HOXA9. Overall, our study uncovered a MSI2-associated RBP complex co-regulating HOXA9 translation and LSC program in AML. We revealed that SYNCRIP is required for LSC function, and HNRNPC and MOV10 as novel vulnerabilities in AML.

Proteins with predicted or known nuclear export signals and significantly enhanced or reduced in nuclear export in XPO1 mutant versus wild-type from SILAC analysis.

Supplementary Figures S1-S12

Number and frequency of XPO1 hotspot mutations found across cancer types.

578 genes sequenced in mouse genomic DNA.

Tissue homeostasis is maintained after stress by engaging and activating the hematopoietic stem and progenitor compartments in the blood. Hematopoietic stem cells (HSCs) are essential for long-term repopulation after secondary transplantation. Here, using a conditional knockout mouse model, we revealed that the RNA-binding protein SYNCRIP is required for maintenance of blood homeostasis especially after regenerative stress due to defects in HSCs and progenitors. Mechanistically, we find that SYNCRIP loss results in a failure to maintain proteome homeostasis that is essential for HSC maintenance. SYNCRIP depletion results in increased protein synthesis, a dysregulated epichaperome, an accumulation of misfolded proteins and induces endoplasmic reticulum stress. Additionally, we find that SYNCRIP is required for translation of CDC42 RHO-GTPase, and loss of SYNCRIP results in defects in polarity, asymmetric segregation, and dilution of unfolded proteins. Forced expression of CDC42 recovers polarity and in vitro replating activities of HSCs. Taken together, we uncovered a post-transcriptional regulatory program that safeguards HSC self-renewal capacity and blood homeostasis.

Fold-change in abundance and statistical significance of proteins in XPO1E571K/WT versus XPO1WT NALM-6 cells detected by SILAC analysis of cytoplasmic fractions

Characteristics of XPO1 mutant and wild-type CLL patients that underwent RNA-sequencing analysis.

Fold-change in abundance and statistical significance of proteins in XPO1E571K/WT versus XPO1WT NALM-6 cells detected by SILAC analysis of nuclear fractions

Supplementary Table Captions for Tables S1-S6

RNA binding proteins (RBPs) have been increasingly recognized as an important class of regulators of normal and malignant hematopoiesis. However, the exact function and underpinning mechanisms of the RBPs that govern hematopoietic stem cells (HSCs) remains poorly characterized. We had previously identified SYNCRIP as a critical RBP that controls leukemia stem cell program in myeloid leukemia. Here, using the novel murine genetic conditional knockout (cKO) model, we delineated the role of SYNCRIP in regulating the low-output HSC. We developed a Syncrip cKO allele and crossed Syncripf/f mice to the interferon (IFN) -a-inducible Mx-1-Cre mice to create Syncripf/f Mx-1-Cre+. We consistently obtained near complete depletion of SYNCRIP 3 weeks after two consecutive Poly(I:C) injections. We observed that SYNCRIP is dispensable for static hematopoiesis and Syncrip KO animals showed equivalent number and frequencies of stem and progenitor cells (Lin-Sca+cKit+ (LSK)- LT-HSC (CD48-CD150+); MPP1 (CD48-CD150-); MPP2 (CD48+CD150+); MPP4 (CD48-CD150-)). However, KO SyncripD/D deficient cells were outcompeted by WT Syncripf/f cells in the transplantation setting (bone marrow (BM) chimerism WT (n=9) 38% ± 7.8% vs. KO (n=9) 2.7% ± 0.8%, p<0.001 at 16 weeks post-transplant) and completely lost their ability to repopulate in secondary recipient animals (WT (n=5) 58% ± 7.4% vs. KO (n=5) 7.2 %± 2.9%, p<0.001 at 16 weeks post-transplant). These data strongly indicate that SYNCRIP is critical for maintenance of long-term self-renewal of HSCs. To decipher the effect of Syncrip deletion on the transcriptomic changes in different cell types upon Syncrip loss, we performed single cell RNA sequencing analysis (scRNA-seq) of sorted LK cells (Lin-cKit+ cells) from KO SyncripD/D (n=3) vs. WT Syncrip f/f (n=3) mice. While there is no significant change in frequencies of stem and progenitor compartments, we found defective trajectory from the HSC that is closely identified as low-output HSC based on previously performed barcoding studies. We observed a strong activation of cellular response to stress and unfolded proteins, in particular the HSF1-dependent pathways upon Syncrip depletion specifically within the HSC population. To further investigate the impacts of SYNCRIP loss in the HSC unfolded protein stress response, we evaluated unfolded proteins in cells using tetraphenylethene maleimide (TMI)-based flow cytometry. The abundance of accessible thiols in unfolded proteins, which is bound by TMI serves as a surrogate measurement for the state of the unfolded proteome. We consistently observed almost 2.5-fold increase in TMI signals specifically in LT-HSC, but not ST-HSCs or MPPs upon SYNCRIP deletion indicating that SYNCRIP is required to maintain high protein quality in HSCs. Similar results were obtained with the epichaperome probe PU-FITC, which consists of HSP90 inhibitor PU-H71 conjugated to FITC. PU-H71 selectively binds to the altered epichaperome, which reflects an accumulation of chaperon networks in an aberrant cellular stress condition. Altogether, these data further confirmed that SYNCRIP depletion tips off the proteostatic balance. To understand the molecular mechanisms underpinning the functional requirement of SYNCRIP in HSPCs, we identified 534 direct mRNA targets of SYNCRIP using hyper-TRIBE method. We performed transcriptomic and proteomic analysis of sorted LT- HSCs and LSKs respectively upon SYNCRIP deletion. We integrated these datasets and found a strong enrichment of SYNCRIP targets in control of cytoskeleton and RHO GTPase related pathways. Using immunofluorescence imaging, we confirmed that SYNCRIP deletion in HSCs resulted in 2-fold reduction in RHO GTPase CDC42 expression coupled with reduced tubulin and a loss of cellular polarity (percentage of tubulin polarized cells 56% WT vs. 40% KO). We also observed that Syncrip deficient HSCs demonstrated reduced expression of lysosomal-associated membrane protein 1 (LAMP-1) and less asymmetric distribution of LAMP1 marked lysosomes during cell division (LAMP1 asymmetric division 25% WT vs. 19% KO). Overexpression of CDC42 restored cell polarity and partly rescued ability of KO SyncripD/D to serially replate. Overall, SYNCRIP is required for maintenance of protein homeostasis and cell polarity of the reserve HSCs. Our study uncovers a new regulatory axis that controls stem cell stress responses to preserve HSC self-renewal.

Altered expression of XPO1, the main nuclear export receptor in eukaryotic cells, has been observed in cancer, and XPO1 has been a focus of anticancer drug development. However, mechanistic evidence for cancer-specific alterations in XPO1 function is lacking. Here, genomic analysis of 42,793 cancers identified recurrent and previously unrecognized mutational hotspots in XPO1. XPO1 mutations exhibited striking lineage specificity, with enrichment in a variety of B-cell malignancies, and introduction of single amino acid substitutions in XPO1 initiated clonal, B-cell malignancy in vivo. Proteomic characterization identified that mutant XPO1 altered the nucleocytoplasmic distribution of hundreds of proteins in a sequence-specific manner that promoted oncogenesis. XPO1 mutations preferentially sensitized cells to inhibitors of nuclear export, providing a biomarker of response to this family of drugs. These data reveal a new class of oncogenic alteration based on change-of-function mutations in nuclear export signal recognition and identify therapeutic targets based on altered nucleocytoplasmic trafficking. SIGNIFICANCE: Here, we identify that heterozygous mutations in the main nuclear exporter in eukaryotic cells, XPO1, are positively selected in cancer and promote the initiation of clonal B-cell malignancies. XPO1 mutations alter nuclear export signal recognition in a sequence-specific manner and sensitize cells to compounds in clinical development inhibiting XPO1 function.This article is highlighted in the In This Issue feature, p. 1325.