Genes affecting DNA methylation (DNAme) are frequently comutated with splicing factors in acute myeloid leukemia (AML) and associate with more aggressive phenotypes. To elucidate the underlying molecular mechanisms, we deeply profiled wild-type and IDH2R140Q/SRSF2P95 single- or double-mutant AMLs. We find a unique set of mis-spliced genes and differentially methylated CpGs in double mutants. Mis-spliced exons are enriched in CCNG splicing enhancers and in the corresponding DNAme changes. Using a machine learning model, we can accurately predict exon inclusion levels from proximal CpGs. These CpGs are more likely to overlap footprints of RNA binding and chromatin-modifying complexes but not transcription factors. We also report unique gene expression profiles associated with each genotype; however, the differentially expressed genes do not overlap with mis-spliced transcripts. Instead, the mis-spliced genes encode for proteins that interact with the complexes regulating these differentially expressed genes. Thus, aberrant DNAme and splicing lead to the mis-splicing of key regulatory complexes, resulting in the aberrant gene expression profiles characteristic of these AMLs.

Oncogenic mutations of SF3B1 are common in myeloid cancers, chronic lymphocytic leukemia (CLL) and select solid tumors. Their mechanistic basis for promoting oncogenesis has been investigated in detail, with the stereotyped missplicing of mRNA protein coding sequences most intensively studied. These changes, in genes such as MAP3K7, BRD9, and ABCB7, typically lead to loss of function, thus contributing to cancer pathogenesis. Here, we systematically analyzed the impact of mutant SF3B1 on non-coding regions of mRNA transcripts across disease types, in both cell lines and primary patient specimens. This identified numerous novel and highly reproducible splicing alterations in such regions. Studies of one target gene, DCAF16, revealed multiple complex mutation-induced alterations in its 5’ and 3’ untranslated regions (5’, 3’ UTRs). Remarkably, these were mechanistically associated with increased DCAF16 protein levels in SF3B1 mutant cells, representing the first time that oncogenic SF3B1 has been shown to increase levels of a target protein in a gain-of-function manner. DCAF16 is a substrate recognition adapter for the DDB1/CUL4 E3 ubiquitin ligase complex. Novel protein degrader small molecules which co-opt DCAF16 to degrade BRD4 as a neosubstrate demonstrated preferential selectivity for SF3B1 mutant cancers and CLL primary patient specimens due to increased DCAF16 protein levels. In turn, this reveals the therapeutic relevance of mutant SF3B1 dysregulation of transcript untranslated regions and uncovers a novel strategy for the treatment of these important neoplasms.

ABSTRACT Myelodysplastic syndromes (MDS) are clonal hematopoietic malignancies characterized by ineffective hematopoiesis, dysplastic morphology, and risk of progression to acute myeloid leukemia. While genomic alterations intrinsic to malignant MDS disease-initiating cells have been well-characterized, molecular assessment of the bone marrow in situ has been limited. Here we present single cell spatial assessment of gene expression, T cell receptors, as well as MDS-defining mutations and RNA isoforms within fixed, decalcified human bone marrow core biopsies (41 MDS, 15 controls) paired with single cell immunogenomic analysis of bone marrow aspirates (35 MDS, 6 controls). Bone marrow spatial analyses of >7.47×10 6 cells identified hematopoietic and non-hematopoietic populations not readily captured in dissociated tissue. We developed computational methods to compare ecological niche structures, revealing enriched hematopoietic niches and reorganization of T cell immunity in MDS patient bone marrow. In situ genotyping of mutant cells revealed increased TGFβ expression in malignant megakaryocytes suppressing local T cell cytotoxicity. By contrast, TGFβ signaling was disrupted in mutant cells due to aberrant splicing of multiple TGFβ signaling components. This study provides a spatially resolved landscape of human MDS bone marrow and uncovers mechanisms by which malignant cells simultaneously promote intrinsic clonal persistence while rewiring the microenvironment for immune escape.

Developing chimeric antigen receptor (CAR) T cells for acute myeloid leukemia (AML) has been challenging due to a lack of known AML-associated antigens that spare normal hematopoietic precursor cells. Here we reasoned that donor autoantibodies from AML recipients cured following allogeneic transplant and responsible for graft-versus-leukemia effect could be engineered to create effective CAR-T cells. We generated CAR-T cells against one such antigen - U5 snRNP200, an RNA helicase localized to the AML cell surface and absent from normal hematopoietic precursors. Anti-U5 snRNP200 CAR-T cells were effective in human and syngeneic models of AML as well as B-cell acute lymphoblastic leukemia (B-ALL), a setting where surface U5 snRNP200 is also present. Armoring CAR-T cells with IL-18 led to antigen gain on AML, durable remission, and protection from AML rechallenge. These data thereby identify a CAR-T cell platform which addresses prior limitations in tumor-selectivity and safety for patients with acute leukemias.

Mutations in RNA splicing factors are prevalent across cancers and generate recurrently mis-spliced mRNA isoforms. Here, we identified a series of bona fide neoantigens translated from highly stereotyped splicing alterations promoted by neomorphic, leukemia-associated somatic splicing machinery mutations. We utilized feature-barcoded peptide-major histocompatibility complex (MHC) dextramers to isolate neoantigen-reactive T cell receptors (TCRs) from healthy donors, patients with active myeloid malignancy, and following curative allogeneic stem cell transplant. Neoantigen-reactive CD8+ T cells were present in the blood of patients with active cancer and had a distinct phenotype from virus-reactive T cells with evidence of impaired cytotoxic function. T cells engineered with TCRs recognizing SRSF2 mutant-induced neoantigens arising from mis-splicing events in CLK3 and RHOT2 resulted in specific recognition and cytotoxicity of SRSF2-mutant leukemia. These data identify recurrent RNA mis-splicing events as sources of actionable public neoantigens in myeloid leukemias and provide proof of concept for genetically redirecting T cells to recognize these targets.

Mutations in RNA splicing factors are the most common class of genetic alterations in MDS and are also prevalent in AML. These mutations cause recurrent splicing changes in a highly sequence-specific manner across patients and cancer types. We hypothesized that a subset of these aberrant splicing changes would generate “public” neoantigens (i.e. shared across patients) that can serve as potential targets of T cell-based immunotherapies. Here we identify a series of shared, bona fide neoantigens induced by leukemia-associated mutations in RNA splicing factors SRSF2 and ZRSR2. We validate in vitro that these neoantigens are presented on HLA class I (HLA-I) and immunogenic. We also discover T cell receptors (TCRs) reactive to identified neoantigens and demonstrate that introduction of these TCRs into primary human T cells redirects the T cells to selectively eliminate leukemic cells. To predict mis-splicing derived neoantigens, we analyzed RNA-seq datasets from five myeloid leukemia patient cohorts with SRSF2 mutations (n=107), ZRSR2 mutations (n=33), or no mutations in splicing factors (n=837). We identified mis-spliced mRNA isoforms consistently produced across SRSF2 or ZRSR2 mutant patients but minimally expressed in healthy bone marrow, PBMCs, and a panel of 14 normal tissues. We translated each tumor-specific mRNA isoform in silico, split into 8-12-mer peptides, and predicted high-affinity binders to HLA-A*02:01. In total, we selected 56 candidate mis-splicing derived neoantigens created by mutant SRSF2 and 19 by mutant ZRSR2 for further in vitro studies. We synthesized candidate peptides and validated their HLA-I binding using the T2 HLA-A2 shift assay. We then tested if the peptides are immunogenic (i.e. elicit effector CD8+ T cell responses) in PBMCs from 14 unique healthy donors. One particular peptide, derived from SRSF2 mutation-induced exon 4 skipping in CLK3 and confirmed by HLA-I immunopeptidomics, was immunogenic across multiple donors. To isolate CD8+ T cells reactive to the CLK3 neoantigen, we constructed dextramers, which are composed of ten peptide/HLA-I complexes on a dextran scaffold. We sorted CLK3 peptide-primed CD8+ T cells with dual-color CLK3 neoepitope dextramers and performed TCR-seq which identified 11 distinct TCR clonotypes from two healthy donors. Primary human T cells transduced with these TCRs demonstrated remarkably specific cytolysis of HLA-A*02:01+ leukemic cells expressing the CLK3 neoantigen as well as cells harboring SRSF2 mutations. The discovery of immunogenic neoantigens in splicing factor mutant cells begs the question of how such malignancies develop in the setting of potential immune responses to these antigens. To address this question, we synthesized a panel of DNA-barcoded dextramers against 43 SRSF2 and 12 ZRSR2 mutant-induced candidate peptides as well as one CMV and three negative control peptides. Using this dextramer panel, we isolated antigen-reactive CD8+ T cells from splicing factor mutant MDS/AML patient PBMCs and performed single-cell RNA-, TCR-, and dextramer barcode sequencing. Downstream single-cell analysis revealed that neoantigen-reactive CD8+ T cells are present in MDS and AML patient blood and are clonally expanded. However, they have a distinct gene expression profile from virus-reactive T cells with evidence of impaired T cell cytotoxic function at the time of active MDS and AML. For patients with high-risk MDS and AML, allogeneic stem cell transplantation (allo-SCT) remains the most established curative therapy and results in donor T cell-mediated graft-versus-leukemia effect. To test if donor T cells could recognize mis-splicing derived neoantigens in patient leukemia, we performed dextramer-based single-cell profiling of matched pre- and post-allo-SCT patient PBMC samples. This discovered a donor-derived TCR that is clonally expanded and cognate to a peptide derived from SRSF2 mutant-induced intron retention in RHOT2. Primary human T cells transduced with this TCR specifically recognized and lysed leukemic cells expressing the RHOT2 neoantigen. Overall, these data identify recurrent RNA mis-splicing events as sources of actionable public neoantigens in myeloid malignancies and provide proof-of-concept for genetically redirecting T cells to recognize these targets. These data have immediate therapeutic implications as the TCRs presented can be applied for transgenic TCR-T cell therapy.

Myelodysplastic syndromes (MDS) represent a heterogenous group of myeloid neoplasms characterized by ineffective hematopoiesis, dysplastic morphology, and risk of transformation to acute leukemia. While neoplastic cell-intrinsic mechanisms driving MDS pathogenesis have been intensely investigated, the spatial architecture and in situ immune microenvironment in human MDS has not been studied at single-cell resolution. We hypothesized that the MDS bone marrow microenvironment is characterized by altered hematopoietic and non-hematopoietic cell distributions and transcriptional profiles including the composition and location of adaptive immune subsets. We developed an approach for high-definition spatial transcriptomic analysis of fixed and EDTA-decalcified human marrow core biopsies, enabling interrogation of hundreds of genes simultaneously in the context of their cellular location. We designed a marrow and immune focused 433 gene panel for the Xenium platform (10x Genomics) and applied this to 7 decalcified core marrow biopsies from patients with early-stage untreated MDS, and 3 control marrows, yielding in situ spatial data for 750,688 cells. In parallel, we performed 32-color spectral flow cytometry along with 5' single-cell CITE- and T cell receptor- (TCR) sequencing (N=31 MDS, N=9 controls) capturing 521,853 cells and 145,176 TCRs. Single-cell spatial transcriptomics identified >20 distinct cell types including populations that are poorly captured by analyses of single-cell suspensions from bone marrow aspirates: megakaryocytes, osteoblasts, endothelial cells, mesenchymal stromal cells, vascular smooth muscle cells, and adipocytes. Nearest neighbor analyses revealed reduced cell type diversity in MDS compared to controls, particularly surrounding osteoblasts (p=1.7x10-2). Within the immune compartment, we captured B, T, and natural killer (NK) cell types for further stratification into FOXP3+ T regulatory cells, naïve and memory CD4 or CD8 T cells, along with numerous phenotypic and functional markers. Spatial data were integrated with single-cell proteomic, RNA, and TCR sequencing in the same individuals to further elucidate the adaptive immune composition and T cell repertoire of the MDS marrow. In MDS samples, B cells were reduced in frequency whereas T cells were the most abundant lymphocyte cell type. Although MDS T cell subset composition was similar to control marrows, the spatial distribution of immune cells in MDS was distinct. Spatial analysis identified aggregates of T, B, and plasmacytoid dendritic cells (pDC) reminiscent of tertiary lymphoid structures in the marrow (which we defined as clusters of B, T, NK and pDCs with > 10 T cells). Immune aggregates were larger in MDS (>75 cells/aggregate, range 83-390) compared to controls (no aggregates >72 cells) with increased proportions of T cells (p=1.2x10-3) and frequencies of CD4 and CD8 T cell subsets. TCF7, a transcription factor linked to T cell stemness, and CXCR4, a key marker of bone marrow homing, were increased in CD4 (p=7.0x10-4 and p=1.5x10-5 respectively) and CD8 (p=1.6x10-2 and p=8.9x10-7 respectively) T cells in MDS immune aggregates. These findings suggest the emergence of immunologically active niches within the MDS marrow and would not have been captured via single-cell RNA-seq alone. Outside of immune aggregates, we found increased memory CD8 T cells (p=1.4x10-2) in MDS. In contrast, control marrows showed no difference in T cell subset frequency between cells inside or outside of immune aggregates. CITE/TCR-seq revealed clonal T cell expansions in MDS within GZMB+ CD8s, distinct from controls (where clonal expansions were found in GZMK+ CD8s), supporting altered CD8 T cell immunity in MDS. Our integrative analyses, including single-cell spatial transcriptomic profiling, revealed active remodeling of the adaptive immune system in early stage MDS. The finding of large immune aggregates in the MDS marrow with more T cells bearing signatures of active cell recruitment and growth, combined with increased clonally expanded memory CD8 T cells outside immune aggregates, point toward localized changes in the adaptive immune response in early stage MDS. Spatial analyses revealed location-specific T cell differences within the MDS marrow, underscoring the importance of combined phenotypic and spatial approaches to elucidate disease pathophysiology in MDS.

Splicing factor mutations are common in myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), but how they alter cellular functions is unclear. We show that the pathogenic SRSF2P95H/+ mutation disrupts the splicing of mitochondrial mRNAs, impairs mitochondrial complex I function, and robustly increases mitophagy. We also identified a mitochondrial surveillance mechanism by which mitochondrial dysfunction modifies splicing of the mitophagy activator PINK1 to remove a poison intron, increasing the stability and abundance of PINK1 mRNA and protein. SRSF2P95H-induced mitochondrial dysfunction increased PINK1 expression through this mechanism, which is essential for survival of SRSF2P95H/+ cells. Inhibition of splicing with a glycogen synthase kinase 3 inhibitor promoted retention of the poison intron, impairing mitophagy and activating apoptosis in SRSF2P95H/+ cells. These data reveal a homeostatic mechanism for sensing mitochondrial stress through PINK1 splicing and identify increased mitophagy as a disease marker and a therapeutic vulnerability in SRSF2P95H mutant MDS and AML.

Mutations in the RNA splicing factor gene SF3B1 are common across hematologic and solid cancers and result in widespread alterations in splicing, yet there is currently no therapeutic means to correct this mis-splicing. Here, we utilize synthetic introns uniquely responsive to mutant SF3B1 to identify trans factors required for aberrant mutant SF3B1 splicing activity. This revealed the G-patch domain-containing protein GPATCH8 as required for mutant SF3B1-induced splicing alterations and impaired hematopoiesis. GPATCH8 is involved in quality control of branchpoint selection, interacts with the RNA helicase DHX15, and functionally opposes SURP and G-patch domain containing 1 (SUGP1), a G-patch protein recently implicated in SF3B1-mutant diseases. Silencing of GPATCH8 corrected one-third of mutant SF3B1-dependent splicing defects and was sufficient to improve dysfunctional hematopoiesis in SF3B1-mutant mice and primary human progenitors. These data identify GPATCH8 as a novel splicing factor required for mis-splicing by mutant SF3B1 and highlight the therapeutic impact of correcting aberrant splicing in SF3B1-mutant cancers.

Mutations in the core RNA splicing factor SF3B1 are common across MDS, CLL, and clonal hematopoiesis. Prior studies have elucidated that mutations in SF3B1 result in neomorphic widespread changes in splicing due to usage of aberrant intronic branchpoint nucleotides. However, the molecular basis by which mutant SF3B1 induces mis-splicing is not established and therapeutic means to correct mis-splicing due to mutant SF3B1 do not exist. Recently, the aberrant gain-of-function splicing activity of mutant SF3B1 was harnessed to regulate expression of proteins in a mutant selective manner using synthetic version of endogenous mRNA sequences uniquely recognized by mutant SF3B1. Here, we utilized this technology to engineer a synthetic intron derived from MAP3K7 that interrupts the coding sequence of the fluorescent protein mEmerald such that mutant SF3B1 produces lower mEmerald expression than SF3B1 wild-type (WT) cells (Fig.A). This fluorescent splicing reporter was then used to perform positive enrichment whole genome CRISPR screens to identify genes whose deletion corrects SF3B1 mutant aberrant splicing. GPATCH8 was the single most robust hit. GPATCH8 knockout strongly corrected mis-splicing of both synthetic and endogenous MAP3K7 introns (Fig.A). A previously unexplored protein, GPATCH8 has domains characteristic of RNA splicing factors including a G-patch motif, thought to be important in activating RNA helicases. Interestingly, recent work suggests that mutations in SF3B1 result in mis-splicing by disrupting physical interaction of SF3B1 to the RNA helicase DHX15 via the G-patch domain containing protein SUGP1. We therefore compared SUGP1 to GPATCH8 activity, evaluating first their transcriptome-wide binding sites using TRIBE-seq. We found that both proteins predominantly target intronic sequences over other genomic regions suggesting roles in splicing. RNA-seq of SF3B1 mutant and WT cells with or without GPATCH8/SUGP1 deletion revealed that ~33% of SF3B1 mutant splicing alterations are corrected by GPATCH8 deletion, while ~60% of SF3B1 mutant splicing alterations are recapitulated upon SUGP1 deletion. GPATCH8 regulated splicing events have stronger branchpoints indicating a potential role for GPATCH8 in 3' splice site recognition. To further understand the connection between GPATCH8 and mutant SF3B1, we performed mass spectrometry studies of immunoprecipitated endogenous GPATCH8 in SF3B1 WT and mutant knockin cells. This revealed strong interaction between GPATCH8 and DHX15. GPATCH8 interacts with DHX15 at the same sites previously shown to be occupied by SUGP1 in crystal structures of the SUGP1/DHX15 interaction. These data and further biochemical studies elucidated that GPATCH8 competes with SUGP1 for interaction to DHX15. As such, deletion of GPATCH8 corrects SF3B1 mutant mis-splicing by enhancing the interaction of SUGP1 and DHX15 to the mutant SF3b complex. Given that GPATCH8 is required for a large proportion of the splicing alterations induced by mutant SF3B1, we next evaluated the phenotypic effect of correcting SF3B1 mutant mis-splicing. We first tested the impact of anti-Gpatch8 shRNAs in bone marrow from WT mice as well as animals with conditional knockin of Sf3b1 K700E, K666N, or R625H mutations. While each Sf3b1 mutation results in impaired colony formation of hematopoietic precursors in methylcellulose assays, silencing of Gpatch8 rescued colony formation from hematopoietic precursors of each of the different Sf3b1 mutant mice and was tolerated by normal hematopoietic precursors (Fig.B). Moreover, while CRISPR base-edited knockin of the SF3B1 K700E mutation in adult CD34 + cells impaired erythroid development, this erythroid differentiation defect was rescued by GPATCH8 deletion in the same cells (Fig.B). These data identify GPATCH8 as a novel RNA splicing factor involved in quality control of RNA branchpoint selection whose expression is required for mis-splicing by the different mutant forms of SF3B1. GPATCH8 antagonizes the activity of SUGP1 by competing for interaction with DHX15. These findings suggest that disrupting GPATCH8/DHX15 interaction could have important therapeutic benefit for the multitude of SF3B1 mutant hematopoietic diseases. This study also demonstrates the power of synthetic intronic splicing assays for discovery of trans factors and druggable proteins required by leukemia-associated mutant RNA splicing factors.

Myelodysplastic syndrome (MDS) is a clonal disorder characterized by hyperproliferation and dysplasia of hematopoietic progenitor cells, deficiencies in circulating blood cells, and progression to acute myeloid leukemia (AML). Splicing factor mutations are common in MDS, but how they alter cellular functions remains unclear. We show that increased mitophagy is a common feature in SRSF2P95* mutated MDS and AML. The SRSF2P95H/+ mutation alters the splicing of multiple mRNAs encoding mitochondrial proteins, impairs mitochondrial function, and increases mitophagy compared to isogenic cells with wild-type SRSF2. We also identify a mechanism of mitochondrial surveillance mediated by PINK1 (PTEN Induced Kinase 1). PINK1 mRNA is alternatively spliced to a form that retains an intron with a premature termination codon, yielding an unstable mRNA and reduced PINK1 mRNA abundance. Mitochondrial stress promotes excision of this poison intron, stabilizing the mRNA and increasing PINK1 mRNA and protein. Similarly, SRSF2P95H induces mitochondrial defects and increases PINK1 expression by promoting removal of the poison intron. In contrast, disrupting splicing by inhibiting glycogen synthase kinase 3 promotes retention of the poison intron, reducing PINK1 mRNA, impairing mitophagy, and activating apoptosis in SRSF2P95H/+cells. These data reveal a mechanism for sensing mitochondrial stress through PINK1 splicing and identify increased mitophagy as both a hallmark and a therapeutic vulnerability in SRSF2P95H mutant MDS and AML.

Supplementary Figure from Impaired Proteolysis of Noncanonical RAS Proteins Drives Clonal Hematopoietic Transformation

Despite recent advances in the treatment of acute myeloid leukemia (AML), there has been limited success in targeting surface antigens in AML, in part due to shared expression across malignant and normal cells. Here, high-density immunophenotyping of AML coupled with proteogenomics identified unique expression of a variety of antigens, including the RNA helicase U5 snRNP200, on the surface of AML cells but not on normal hematopoietic precursors and skewed Fc receptor distribution in the AML immune microenvironment. Cell membrane localization of U5 snRNP200 was linked to surface expression of the Fcγ receptor IIIA (FcγIIIA, also known as CD32A) and correlated with expression of interferon-regulated immune response genes. Anti-U5 snRNP200 antibodies engaging activating Fcγ receptors were efficacious across immunocompetent AML models and were augmented by combination with azacitidine. These data provide a roadmap of AML-associated antigens with Fc receptor distribution in AML and highlight the potential for targeting the AML cell surface using Fc-optimized therapeutics.

ABSTRACTRNA splicing factors are recurrently affected by alteration-of-function mutations in clonal blood disorders, highlighting the importance of splicing regulation in hematopoiesis. However, our understanding of the impact of dysregulated RNA splicing has been hampered by the inability to distinguish mutant and wildtype cells in primary patient samples, the cell-type complexity of the hematopoietic system, and the sparse and biased coverage of splice junctions by short-read sequencing typically used in single-cell RNA sequencing. To overcome these limitations, we developed GoT-Splice by integrating Genotyping of Transcriptomes (GoT) with enhanced efficiency long-read single-cell transcriptome profiling, as well as proteogenomics (with CITE-seq). This allowed for the simultaneous single-cell profiling of gene expression, cell surface protein markers, somatic mutation status, and RNA splicing. We applied GoT-Splice to bone marrow progenitors from patients with myelodysplastic syndrome (MDS) affected by mutations in the most prevalent mutated RNA splicing factor – the core RNA splicing factor SF3B1. High-resolution mapping of SF3B1mut vs. SF3B1wt hematopoietic progenitors revealed a fitness advantage of SF3B1mut cells in the megakaryocytic-erythroid lineage, resulting in an expansion of SF3B1mut erythroid progenitor (EP) cells. SF3B1mut EP cells exhibited upregulation of genes involved in regulation of cell cycle and mRNA translation. Long-read single-cell transcriptomes revealed the previously reported increase of aberrant 3’ splicing site usage in SF3B1mut cells. However, the ability to profile splicing within individual cell populations uncovered distinct cryptic 3’ splice site usage across different progenitor populations, as well as stage-specific aberrant splicing during erythroid maturation. Lastly, as splice factor mutations occur in clonal hematopoiesis (CH) with increased risk of neoplastic transformation, we applied GoT-Splice to CH samples. These data revealed that the erythroid lineage bias, as well as cell-type specific cryptic 3’ splice site usage in SF3B1mut cells, precede overt MDS. Collectively, we present an expanded multi-omics single-cell toolkit to define the cell-type specific impact of somatic mutations on RNA splicing, from the earliest phases of clonal outgrowths to overt neoplasia, directly in human samples.

Many cancers carry recurrent, change-of-function mutations affecting RNA splicing factors. Here, we describe a method to harness this abnormal splicing activity to drive splicing factor mutation-dependent gene expression to selectively eliminate tumor cells. We engineered synthetic introns that were efficiently spliced in cancer cells bearing SF3B1 mutations, but unspliced in otherwise isogenic wild-type cells, to yield mutation-dependent protein production. A massively parallel screen of 8,878 introns delineated ideal intronic size and mapped elements underlying mutation-dependent splicing. Synthetic introns enabled mutation-dependent expression of herpes simplex virus–thymidine kinase (HSV–TK) and subsequent ganciclovir (GCV)-mediated killing of SF3B1-mutant leukemia, breast cancer, uveal melanoma and pancreatic cancer cells in vitro, while leaving wild-type cells unaffected. Delivery of synthetic intron-containing HSV–TK constructs to leukemia, breast cancer and uveal melanoma cells and GCV treatment in vivo significantly suppressed the growth of these otherwise lethal xenografts and improved mouse host survival. Synthetic introns provide a means to exploit tumor-specific changes in RNA splicing for cancer gene therapy. Synthetic introns tailored for specific splice-factor mutations enable targeted cancer gene therapy.

High-throughput sequencing and functional characterization of the cancer transcriptome have uncovered cancer-specific dysregulation of RNA splicing across a variety of cancers. Alterations in the cancer genome and dysregulation of RNA splicing factors lead to missplicing, splicing alteration-dependent gene expression and, in some cases, generation of novel splicing-derived proteins. Here, we review recent advances in our understanding of aberrant splicing in cancer pathogenesis and present strategies to harness cancer-specific aberrant splicing for therapeutic intent.

Abstract Recently, screens for mediators of resistance to FLT3 and ABL kinase inhibitors in leukemia resulted in the discovery of LZTR1 as an adapter of a Cullin-3 RING E3 ubiquitin ligase complex responsible for the degradation of RAS GTPases. In parallel, dysregulated LZTR1 expression via aberrant splicing and mutations was identified in clonal hematopoietic conditions. Here we identify that loss of LZTR1, or leukemia-associated mutants in the LZTR1 substrate and RAS GTPase RIT1 that escape degradation, drives hematopoietic stem cell (HSC) expansion and leukemia in vivo. Although RIT1 stabilization was sufficient to drive hematopoietic transformation, transformation mediated by LZTR1 loss required MRAS. Proteolysis targeting chimeras (PROTAC) against RAS or reduction of GTP-loaded RAS overcomes LZTR1 loss-mediated resistance to FLT3 inhibitors. These data reveal proteolysis of noncanonical RAS proteins as novel regulators of HSC self-renewal, define the function of RIT1 and LZTR1 mutations in leukemia, and identify means to overcome drug resistance due to LZTR1 downregulation. Significance: Here we identify that impairing proteolysis of the noncanonical RAS GTPases RIT1 and MRAS via LZTR1 downregulation or leukemia-associated mutations stabilizing RIT1 enhances MAP kinase activation and drives leukemogenesis. Reducing the abundance of GTP-bound KRAS and NRAS overcomes the resistance to FLT3 kinase inhibitors associated with LZTR1 downregulation in leukemia. This article is highlighted in the In This Issue feature, p. 2221

Numerous antibody-based immunotherapies are in development for AML including antibodies targeting CD47, CD70, CD123, CD33, and TIM3. However, systematic evaluation of the distribution of AML-associated antigens in combination with Fc receptors and the immune microenvironment required for antibody-mediated therapeutic responses have been limited. Additionally, efforts to maximize antibody Fc receptor engagement on immune effector cells in AML patients via Fc region engineering have not been explored. Here we applied a custom 36-color flow cytometry panel tailored for AML patient bone marrow along with proteogenomic assessment via CITE-seq to simultaneously assess known and novel antibody target expression on AML cells, immune microenvironment, and Fc receptor expression across immune subsets. Application to samples from 7 age-matched healthy donors and 51 newly diagnosed AML patients allowed for unbiased identification of the malignant blast population via UMAP projection. Quantification of differential antigen expression on blasts and leukemic stem cells versus normal hematopoietic stem and progenitor cell (HSPC) subsets, identified that out of 9 AML putative antibody targets (CD123, TIM3, CD33, CD47, CD90, CD38, CD25, CD70, U5 snRNP200), only U5 snRNP200 is expressed on blasts but not normal HSPCs (Fig. A). Characterization of the distribution of Fc receptors revealed that AML patients exhibit reduced cell-surface activating Fc receptors (CD16, CD32A) on NK and non-classical monocytes with concomitant increases in inhibitory (CD32B) Fc receptors on monocytes and B-cells. Moreover, AML patients had fewer classical monocytes and type 2 conventional dendritic (cDC2) cells than normal age-match control marrow. Given the unique cell surface expression of U5 snRNP200 on AML blasts compared with normal HSPCs, we deeply interrogated the distribution of this marker throughout normal and malignant hematopoiesis. While U5 snRNP200 expression was not seen on normal HSPCs, U5 snRNP200 was present on mature B-cells and a subset of NK cells in normal adult human and mouse hematopoietic cells. U5 snRNP200 is an RNA helicase which is an essential conserved component of the spliceosome. Its known function is limited to the nucleus and it is therefore unexpected that it would be expressed on the cell surface. To rigorously confirm expression of U5 snRNP200 on the cell membrane, we knocked in a HaloTag at the N-terminus of the locus encoding U5 snRNP200 within K562 cells and traced expression of tagged endogenous U5 snRNP200 protein across cellular compartments. This clearly revealed cell surface localization of the full-length protein on the surface of AML cells. We next conducted genome-wide CRISPR screens in two human AML cell lines to identify regulators of surface U5 snRNP200 expression. In both cell lines, the top gene required for cell surface U5 snRNP200 expression was CD32A. CRISPR knockout of CD32A in U937 cells reduced U5 snRNP200 surface expression, while restoration of CD32A rescued U5 snRNP200 surface expression. AML-associated expression of U5 snRNP200 was present on a number of syngeneic immunocompetent mouse models of AML including MLL-AF9, RN2 (MLL-AF9 + NRAS G12D), transgenic EVI1 rearranged/SF3B1K700E mutant, and Tet2/Zrsr2 double knockout mice. To test the therapeutic potential of targeting cell surface U5 snRNP200, we engineered several anti-murine U5 snRNP200 antibody variants featuring modified Fc regions to assess the possible roles of engagement of Fc receptors on immune cell subsets to enhance AML cell killing. In both the RN2 and EVI1/SF3B1 mutant models, anti-U5 snRNP200 antibodies featuring the Fc region modified to engage activating Fc receptors and induce antibody-dependent cellular cytotoxicity (IgG2a) provided superior survival benefit compared to IgG1 or IgG2b U5 snRNP200 antibodies or a mutant IgG1 antibody (IgG1 D265A) which does not bind Fc receptors (Fig. B). These data provide a high-density roadmap of the distribution of known and novel AML-associated antigens together with the Fc receptor distribution and immune microenvironment in AML. The results motivated discovery of a novel antibody-therapeutic targeting aberrant cell-surface U5 snRNP200 and research into mechanisms for cell surface trafficking of U5 snRNP200. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Therapy resistance is a major challenge in the treatment of cancer. Here, we performed CRISPR-Cas9 screens across a broad range of therapies used in acute myeloid leukemia to identify genomic determinants of drug response. Our screens uncover a selective dependency on RNA splicing factors whose loss preferentially enhances response to the BCL2 inhibitor venetoclax. Loss of the splicing factor RBM10 augments response to venetoclax in leukemia yet is completely dispensable for normal hematopoiesis. Combined RBM10 and BCL2 inhibition leads to mis-splicing and inactivation of the inhibitor of apoptosis XIAP and downregulation of BCL2A1, an anti-apoptotic protein implicated in venetoclax resistance. Inhibition of splicing kinase families CLKs (CDC-like kinases) and DYRKs (dual-specificity tyrosine-regulated kinases) leads to aberrant splicing of key splicing and apoptotic factors that synergize with venetoclax, and overcomes resistance to BCL2 inhibition. Our findings underscore the importance of splicing in modulating response to therapies and provide a strategy to improve venetoclax-based treatments.

Hemophagocytic lymphohistiocytosis (HLH) is a clinical syndrome that can be inherited or acquired. Herein, we report a case of HLH and pulmonary alveolar proteinosis (PAP) in the setting of lysinuric protein intolerance (LPI) in a male toddler who presented with prolonged fever, respiratory distress, and failure to thrive. On histologic examination, hemophagocytosis was observed in lymph node, bone marrow sections and aspirates. Lung wedge resection was consistent with PAP. LPI was confirmed with genetic sequencing which revealed compound heterozygous mutations in the SLC7A7 gene. LPI is a rare inborn error of metabolism and is not widely known beyond the pediatric group. Though the association of LPI with HLH has been previously described, we believe this is the first reported case of HLH and PAP associated LPI with histopathological correlation. Early recognition of HLH is critical to successful treatment and LPI should be considered in any young infant who presents with HLH- and PAP-related symptoms.