Splice Site Recognition Research Articles

Prp8 positioning of U5 snRNA is linked to 5' splice site recognition.

Prp8 is an essential protein that regulates spliceosome assembly and conformation during pre-mRNA splicing. Recent cryo-EM structures of the spliceosome model Prp8 as a scaffold for the spliceosome's catalytic U snRNA components. Using a new amino acid probing strategy, we identified a dynamic region in human Prp8 that is positioned to stabilize the pre-mRNA in the spliceosome active site through interactions with U5 snRNA. Mutagenesis of the identified Prp8 residues in yeast indicates a role in 5′ splice site recognition. Genetic interactions with spliceosome proteins Isy1, which buttresses the intron branch point, and Snu114, a regulatory GTPase that directly contacts Prp8, further corroborate a role for the same Prp8 residues in substrate positioning and activation. Together the data suggest that adjustments in interactions between Prp8 and U5 snRNA help establish proper positioning of the pre-mRNA into the active site to enhance 5′ splice site fidelity.

Splicing Site Recognition by Synergy of Three Domains in Splicing Factor RBM10.

Splicing factor RBM10 and its close homologues RBM5 and RBM6 govern the splicing of oncogenes such as Fas, NUMB, and Bcl-X. The molecular architecture of these proteins includes zinc fingers (ZnFs) and RNA recognition motifs (RRMs). Three of these domains in RBM10 that constitute the RNA binding part of this splicing factor were found to individually bind RNAs with micromolar affinities. It was thus of interest to further investigate the structural basis of the well-documented high-affinity RNA recognition by RMB10. Here, we investigated RNA binding by combinations of two or three of these domains and discovered that a polypeptide containing RRM1, ZnF1, and RRM2 connected by their natural linkers recognizes a specific sequence of the Fas exon 6 mRNA with an affinity of 20 nM. Nuclear magnetic resonance structures of the RBM10 domains RRM1 and ZnF1 and the natural V354del isoform of RRM2 further confirmed that the interactions with RNA are driven by canonical RNA recognition elements. The well-known high-fidelity RNA splice site recognition by RBM10, and probably by RBM5 and RBM6, can thus be largely rationalized by a cooperative binding action of RRM and ZnF domains.

Markovian encoding models in human splice site recognition using SVM

Splice site recognition is among the most significant and challenging tasks in bioinformatics due to its key role in gene annotation. Effective prediction of splice site requires nucleotide encoding methods that reveal the characteristics of DNA sequences to provide appropriate features to serve as input of machine learning classifiers. Markovian models are the most influential encoding methods that highly used for pattern recognition in biological data. However, a direct performance comparison of these methods in splice site domain has not been assessed yet. This study compares various Markovian encoding models for splice site prediction utilizing support vector machine, as the most outstanding learning method in the domain, and conducts a new precise evaluation of Markovian approaches that corrects this limitation. Moreover, a novel sequence encoding approach based on third order Markov model (MM3) is proposed. The experimental results show that the proposed method, namely MM3-SVM, performs significantly better than thirteen best known state-of-the-art algorithms, while tested on HS3D dataset considering several performance criteria. Further, it achieved higher prediction accuracy than several well-known tools like NNsplice, MEM, MM1, WMM, and GeneID, using an independent test set of 50 genes. We also developed MMSVM, a web tool to predict splice sites in any human sequence using the proposed approach. The MMSVM web server can be assessed at https://pashaei.shinyapps.io/mmsvm.

A Flexible Region of PRP8 Interacts with U5 snRNA Loop I and is Functionally Linked to 5' Splice Site Recognition

A Flexible Region of PRP8 Interacts with U5 snRNA Loop I and is Functionally Linked to 5' Splice Site Recognition

The kinetics of pre-mRNA splicing in the Drosophila genome and the influence of gene architecture.

Production of most eukaryotic mRNAs requires splicing of introns from pre-mRNA. The splicing reaction requires definition of splice sites, which are initially recognized in either intron-spanning ('intron definition') or exon-spanning ('exon definition') pairs. To understand how exon and intron length and splice site recognition mode impact splicing, we measured splicing rates genome-wide in Drosophila, using metabolic labeling/RNA sequencing and new mathematical models to estimate rates. We found that the modal intron length range of 60-70 nt represents a local maximum of splicing rates, but that much longer exon-defined introns are spliced even faster and more accurately. We observed unexpectedly low variation in splicing rates across introns in the same gene, suggesting the presence of gene-level influences, and we identified multiple gene level variables associated with splicing rate. Together our data suggest that developmental and stress response genes may have preferentially evolved exon definition in order to enhance the rate or accuracy of splicing.

VAGA: a novel viscosity-based accelerated gradient algorithm

Proximal Algorithms are known to be very popular in the area of signal processing, image reconstruction, variational inequality and convex optimization due to their small iteration costs and applicability to the non-smooth optimization problems. Various real-world machine learning problems have been solved utilizing the non-smooth convex loss minimization framework, and a recent trend is to design new accelerated algorithms to solve such frameworks efficiently. In this paper, we propose a novel viscosity-based accelerated gradient algorithm (VAGA), that utilizes the concept of viscosity approximation method of fixed point theory for solving the learning problems. We discuss the boundedness of the sequence generated by this iterative algorithm and prove the strong convergence of the algorithm under the few specific conditions. To test the practical performance of the algorithm on real-world problems, we applied it to solve the regularized multitask regression problem with sparsity-inducing regularizers. We present the detailed comparative analysis of our algorithm with few traditional proximal algorithms on three real benchmark multitask regression datasets. We also apply the proposed algorithm to the task of joint splice-site recognition problem of bio-informatics. The improved results demonstrate the efficacy of our algorithm over state-of-the-art proximal gradient descent algorithms. To the best of our knowledge, it is the first time that a viscosity-based iterative algorithm is applied to solve the real world problem of regression and recognition.

Noncanonical GA and GG 5' Intron Donor Splice Sites Are Common in the Copepod Eurytemora affinis.

The noncanonical 5′ intron donor splice sites GA and GG are exceedingly rare in described eukaryotic genomes; however, they are present in ∼12% of introns in the genome of the copepod Eurytemora affinis. Failure to recognize the high frequency of these donor sites compromised the modeling of genes in this newly sequenced genome, including 10 conserved ionotropic glutamate receptor (GluR) family genes curated herein. These introns appear to have been acquired recently, along with many additional idiosyncratic introns. Their high frequency implies the evolution of modified intron donor splice site recognition in this copepod.

Postcatalytic spliceosome structure reveals mechanism of 3'-splice site selection.

Introns are removed from eukaryotic messenger RNA precursors by the spliceosome in two transesterification reactions-branching and exon ligation. The mechanism of 3'-splice site recognition during exon ligation has remained unclear. Here we present the 3.7-angstrom cryo-electron microscopy structure of the yeast P-complex spliceosome immediately after exon ligation. The 3'-splice site AG dinucleotide is recognized through non-Watson-Crick pairing with the 5' splice site and the branch-point adenosine. After the branching reaction, protein factors work together to remodel the spliceosome and stabilize a conformation competent for 3'-splice site docking, thereby promoting exon ligation. The structure accounts for the strict conservation of the GU and AG dinucleotides at the 5' and 3' ends of introns and provides insight into the catalytic mechanism of exon ligation.

Abstract B14: Precision functional genomics for glioblastoma: Identifying molecular therapeutic targets using CRISPR-Cas9 and RNAi technologies in patient isolates

Abstract Glioblastoma (GBM) is the most aggressive and common form of adult brain cancer and is among the deadliest cancers, with a median survival of 15 months using standard-of-care therapies. Thus, improved treatments for GBM are desperately needed. To identify new GBM molecular therapeutic targets, our group has performed multiple functional genetic screens in patient-derived GBM stem-like cells (GSCs) and non-transformed human neural stem and progenitor cells (NPCs), which represent non-neoplastic controls. These screens, which have used both RNAi and CRISPR-Cas9 platforms, have led to the identification of several key molecular vulnerabilities in GSCs, including GBM-specific defects in: 3' splice site recognition, kinetochore function, and loss of redundancy between the kinase activities of PKMYT1 and WEE1. At this meeting we will present an overview of these studies, as well as our current efforts to: comprehensively retest all GBM-specific vulnerabilities scoring in these screens; address whether vulnerabilities arise from specific genetic alterations in patient samples (e.g. NF1 loss or PTEN loss); determine whether inhibition of specific molecular targets blocks tumor growth and/or maintenance; and demonstrate the mode of GBM-specific death for particular targets (e.g., cell cycle arrest, apoptosis, etc). In addition, we will highlight both strengths and limitations of applications of CRISPR-Cas9 technologies in patient samples. Collectively, our work illustrates the power of combining functional genetic technologies with the use of patient isolates to identify novel, patient-specific therapeutic strategies for GBM. Citation Format: Pia Hoellerbauer, Heather Feldman, Sonali Arora, Lucas Carter, Emily J. Girard, Philip Corrin, James M. Olson, Eric C. Holland, Patrick J. Paddison. Precision functional genomics for glioblastoma: Identifying molecular therapeutic targets using CRISPR-Cas9 and RNAi technologies in patient isolates [abstract]. In: Proceedings of the AACR Precision Medicine Series: Opportunities and Challenges of Exploiting Synthetic Lethality in Cancer; Jan 4-7, 2017; San Diego, CA. Philadelphia (PA): AACR; Mol Cancer Ther 2017;16(10 Suppl):Abstract nr B14.

Mutations of RNA splicing factors in hematological malignancies

Systematic large-scale cancer genomic studies have produced numerous significant findings. These studies have not only revealed new cancer-promoting genes, but they also have identified cancer-promoting functions of previously known “housekeeping” genes. These studies have identified numerous mutations in genes which play a fundamental role in nuclear precursor mRNA splicing. Somatic mutations and copy number variation in many of the splicing factors which participate in the formation of multiple spliceosomal complexes appear to play a role in many cancers and in particular in myelodysplastic syndromes (MDS). Mutated proteins seem to interfere with the recognition of the authentic splice sites (SS) leading to utilization of suboptimal alternative splicing sites generating aberrantly spliced mRNA isoforms. This short review is focusing on the function of the splice factors involved in the formation of splicing complexes and potential mechanisms which affect usage of the authentic splice site recognition.

Error-Prone Splicing Controlled by the Ubiquitin Relative Hub1

Accurate pre-mRNA splicing is needed for correct gene expression and relies on faithful splice site recognition. Here, we show that the ubiquitin-like protein Hub1 binds to the DEAD-box helicase Prp5, a key regulator of early spliceosome assembly, and stimulates its ATPase activity thereby enhancing splicing and relaxing fidelity. High Hub1 levels enhance splicing efficiency but also cause missplicing by tolerating suboptimal splice sites and branchpoint sequences. Notably, Prp5 itself is regulated by a Hub1-dependent negative feedback loop. Since Hub1-mediated splicing activation induces cryptic splicing of Prp5, it also represses Prp5 protein levels and thus curbs excessive missplicing. Our findings indicate that Hub1 mediates enhanced, but error-prone splicing, a mechanism that is tightly controlled by a feedback loop of PRP5 cryptic splicing activation.

A new target for gene therapy—splicing factor PUF60

Poly(U)-binding-splicing factor 60 KDa (PUF60), a U2-related splicing factor that facilitates 3′splice-site recognition at the early stage of spliceosome assembly. High expression of PUF60 is related to the occurrence of multiple tumors, such as colorectal cancer, ovarian cancer, gastric cancer, liver cancer, et al. PUF60 can regulate the expression of c-myc through the core-human transcription factor Ⅱ basal transcription factor. Anti-PUF60 antibodies are detected in the sera of patients with early-stage and recurrent tumor, and the levels are significantly decreased after the operation. PUF60 can be used as a combined or independent test index for the diagnosis of cancer, which can be a potential new target for gene therapy. Key words: Neoplasms; Genes, myc; DNA, recombinant

From intronization to intron loss: How the interplay between mRNA-associated processes can shape the architecture and the expression of eukaryotic genes.

Transcription-coupled processes such as capping, splicing, and cleavage/polyadenylation participate in the journey from genes to proteins. Although they are traditionally thought to serve only as steps in the generation of mature mRNAs, a synthesis of available data indicates that these processes could also act as a driving force for the evolution of eukaryotic genes. A theoretical framework for how mRNA-associated processes may shape gene structure and expression has recently been proposed. Factors that promote splicing and cleavage/polyadenylation in this framework compete for access to overlapping or neighboring signals throughout the transcription cycle. These antagonistic interactions allow mechanisms for intron gain and splice site recognition as well as common trends in eukaryotic gene structure and expression to be coherently integrated. Here, I extend this framework further. Observations that largely (but not exclusively) revolve around the formation of DNA-RNA hybrid structures, called R loops, and promoter directionality are integrated. Additionally, the interplay between splicing factors and cleavage/polyadenylation factors is theorized to also affect the formation of intragenic DNA double-stranded breaks thereby contributing to intron loss. The most notable prediction in this proposition is that RNA molecules can mediate intron loss by serving as a template to repair DNA double-stranded breaks. The framework presented here leverages a vast body of empirical observations, logically extending previous suggestions, and generating verifiable predictions to further substantiate the view that the intracellular environment plays an active role in shaping the structure and the expression of eukaryotic genes.

Abstract 4995: Premature polyadenylation causes oncogenic truncations of the tumor suppressor genes BRCA1, LATS1 and MAGI3 in breast cancer

Abstract Cleavage and polyadenylation is a fundamental process in the control of gene expression, yet how cancer cells deregulate this process to generate cancer-causing alterations is only beginning to be appreciated. We previously showed that in human breast cancer, premature polyadenylation of the tumor suppressor gene MAGI3 causes the expression of a truncated protein that dominantly promotes malignant growth via the Hippo pathway. This established a precedent for premature polyadenylation as an aberrant process causing transformation in vitro and in vivo. However, little is known about the molecular mechanism that causes premature polyadenylation, or what types of genes it affects in cancer. Despite the absence of cis-acting mutations, premature polyadenylation of MAGI3 occurs specifically following the gene’s large internal exon. We therefore asked whether an epigenetic mechanism might underlie the phenomenon of premature polyadenylation in cancer. Here, we report that genes with large internal exons are susceptible to truncation by premature polyadenylation. In addition to MAGI3, premature polyadenylation truncates the gene products of two tumor suppressor genes with large internal exons, BRCA1 and LATS1. These truncated proteins are upregulated in breast cancer cells compared to non-transformed mammary epithelial cells and enable downstream oncogenic events instead of providing tumor suppressive functions. Large internal exons are infrequently found in the genome as the mechanics of efficient splice site recognition favor small exons. Interestingly in transcripts, large internal exons are conspicuously enriched in N6-methyladenosine (m6A), though the functional significance of these epitranscriptomic marks remains unknown. Molecularly, we find that methylated levels of N6-adenosine in the large internal exons of BRCA1, LATS1 and MAGI3 transcripts are significantly reduced in breast cancer cells that upregulate the truncated gene products compared to non-transformed mammary epithelial cells. These results raise the intriguing possibility that m6A marks in large internal exons are mechanistically involved in repressing premature polyadenylation. We are currently investigating this hypothesis by perturbing N6-adenosine methylation in cells and examining the functional consequences. In summary, our study suggests that premature polyadenylation may be a more pervasive alteration mechanism than previously appreciated in cancer and begins to illuminate a gene region-specific mechanism responsible for repressing premature polyadenylation. Citation Format: Thomas K. Ni, Charlotte Kuperwasser. Premature polyadenylation causes oncogenic truncations of the tumor suppressor genes BRCA1, LATS1 and MAGI3 in breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4995. doi:10.1158/1538-7445.AM2017-4995

Abstract 4471: Novel SF3B1 deletion mutations result in aberrant RNA splicing in CLL patients

Abstract Recurrent heterozygous mutations in SF3B1, a component of the U2snRNP complex involved in branchsite recognition during the first step of splicing catalysis, are found in various solid tumors and hematologic cancers including myelodysplastic syndromes (MDS) and chronic lymphocytic leukemia (CLL). To provide broader insight into the effects of SF3B1 mutations on RNA splicing in CLL, RNAseq was performed on 215 CLL patient samples with unknown SF3B1 mutation status. 37 patients (17%) demonstrated cryptic 3’ splice site usage we and others have previously reported in patients with known SF3B1 hotspot mutations. Interestingly, 3 patient samples harbored previously unreported in-frame deletions in SF3B1 around K700, the most common amino acid mutated in hematologic cancers with SF3B1 mutations. In particular, 2 patient sequences contained 6 nucleotide (nt) deletions (c.2186 to c.2191) resulting in replacement of Q698, Q699 and K700 with Q (p.Q698_K700delinsQ: “p.Q698del”) and 1 patient sequence had a 3 nt deletion (c.2192 to c.2194) resulting in replacement of K700 and V701 with N (p.K700_V701delinsN: “p.K700del”) in SF3B1. The variant allele fraction (VAF) of these deletions was 20-40%. Targeted resequencing of DNA confirmed presence of these mutations at similar VAFs. Differential splicing analysis across patients with wild-type SF3B1, mutant SF3B1 and our novel deletions revealed similar alternative 3’ splice site patterns. To study the mechanism of alternative 3’ splice site selection induced by these novel p.Q698del and p.K700del in-frame deletions we used our ZDHHC16 minigene system with modifications in the canonical and cryptic 3’ splice sites and respective branchsites. After confirming expression of these deletion mutants and their incorporation in the SF3b complex by qPCR and co-immunoprecipitation, we demonstrated that recognition of both canonical 3’ splice site and alternate branchsite are important for aberrant splicing. This observation is identical to the intronic features required for aberrant splicing that we previously reported for the p.K700E SF3B1 hotspot mutation in CLL and MDS. By comparing the recent cryo-electron microscopy and X-ray crystal structures of SF3b complexes (PDB: 5GM6 and 5IFE, respectively), it is evident that the K700 present at the edge of heat repeat domain 6 is important for pre-mRNA interactions. Both the in-frame deletions and substitution mutations at K700 perturb SF3B1 charge, likely affecting a side-chain interaction with pre-mRNA, thus altering splice-site recognition through a mechanism that is still not well understood. Together, these data indicate that small in-frame deletions in SF3B1 around K700 offer a novel mechanism to induce aberrant splicing in patients with CLL. The common mechanism of action of these mutations and deletions suggest that they should also be sensitive to splicing modulators. Citation Format: Anant A. Agrawal, Michael Seiler, Lindsey Brinton, Rose Mantel, Rosa Lapalombella, Jennifer A. Woyach, Amy J. Johnson, Lihua Yu, John C. Byrd, Peter G. Smith, James S. Blachly*, Silvia Buonamici*. Novel SF3B1 deletion mutations result in aberrant RNA splicing in CLL patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4471. doi:10.1158/1538-7445.AM2017-4471

Restoration of correct splicing in IVSI-110 mutation of β-globin gene with antisense oligonucleotides: implications and applications in functional assay development.

Objective(s):The use of antisense oligonucleotides (AOs) to restore normal splicing by blocking the recognition of aberrant splice sites by the spliceosome represents an innovative means of potentially controlling certain inherited disorders affected by aberrant splicing. Selection of the appropriate target site is essential in the success of an AO therapy. In this study, in search for a splice model system to facilitate the evaluation of AOs to redirect defective splicing of IVSI-110 β-globin intron, an EGFP-based IVSI-110 specific cellular reporter assay system has been developed and a number of AOs were tested in this cellular splicing assay.Materials and Methods:A recombinant plasmid (pEGFP/I-110) carrying the EGFP gene interrupted by a mutated human β-globin intron 1 (IVSI-110) was developed and transfected into K562 cells. A number of AOs with a 2’-O-methyl oligoribonucleotide (2’-O-Me) backbone system were systematically tested in this cellular splicing assay.Results:The mutation in the intron causes aberrant splicing of EGFP pre-mRNA, preventing translation of EGFP; however, treatment of the cells with specific concentration of a sequence specific 2’-O-Me AO targeted to the aberrant splice site induced correct splicing and resulted in restoring of EGFP activity.Conclusion:This cellular splicing assay provides a novel functional assay system in assessing the cellular delivery efficiency of AOs and therapeutic effect of AOs in restoration of aberrant splicing.

RRM domain of Arabidopsis splicing factor SF1 is important for pre-mRNA splicing of a specific set of genes.

The RNA recognition motif of Arabidopsis splicing factor SF1 affects the alternative splicing of FLOWERING LOCUS M pre-mRNA and a heat shock transcription factor HsfA2 pre-mRNA. Splicing factor 1 (SF1) plays a crucial role in 3' splice site recognition by binding directly to the intron branch point. Although plant SF1 proteins possess an RNA recognition motif (RRM) domain that is absent in its fungal and metazoan counterparts, the role of the RRM domain in SF1 function has not been characterized. Here, we show that the RRM domain differentially affects the full function of the Arabidopsis thaliana AtSF1 protein under different experimental conditions. For example, the deletion of RRM domain influences AtSF1-mediated control of flowering time, but not the abscisic acid sensitivity response during seed germination. The alternative splicing of FLOWERING LOCUS M (FLM) pre-mRNA is involved in flowering time control. We found that the RRM domain of AtSF1 protein alters the production of alternatively spliced FLM-β transcripts. We also found that the RRM domain affects the alternative splicing of a heat shock transcription factor HsfA2 pre-mRNA, thereby mediating the heat stress response. Taken together, our results suggest the importance of RRM domain for AtSF1-mediated alternative splicing of a subset of genes involved in the regulation of flowering and adaptation to heat stress.

Activation of the intronic cryptic 5′ splice site depends on its distance to the upstream cassette exon

Splice site selection is a key step that determines the mRNA isoforms generated from a single transcript. The large diversity in splice site sequences emphasizes the plasticity of splice site recognition and selection. In this report, a cell-based reporter system using a SMN1/2 cassette exon was applied to study the roles governing the activation of a cryptic 5′SS from the intron 4 of the CT/CGRP gene. We found that the cryptic site was activated when placed within 124nt downstream the cassette exon, and the level of activation was negatively correlated with its distance from the exon. In addition, activation was not affected by PTB but was eliminated by an insertion extending the exon length. Activated cryptic 5′SSs in intron or exon could override the original alternative 5′SS, obeying the U1 base-pairing rule. These results suggest that the exon length itself could represent a factor in determining the splice site selection.

A Novel Alzheimer-Associated SNP in Tmp21 Increases Amyloidogenesis.

Recent studies suggest that TMP21 is a selective modulator of γ-secretase and its dysregulation affects APP processing, leading to increased Aβ generation. However, the genetic association between Tmp21 and Alzheimer's disease (AD) remains elusive. In this study, we identified that a novel single-nucleotide polymorphism (SNP) rs12435391 (IVS4-28T>C) in intron 4 of Tmp21 was genetically associated with AD. We found that allele C of the SNP rs12435391 did not affect splicing site recognition, but it significantly increased TMP21 gene expression. The stability of Tmp21 pre-mRNA and the transcription of Tmp21 were not affected by allele C of the SNP rs12435391. However, allele C of the SNP rs12435391 significantly increased the splicing efficiency of Tmp21 pre-mRNA, leading to the elevation of mature mRNA. Furthermore, allele C of the SNP rs12435391 significantly reduced C83 level and increased Aβ generation. Taken together, our study suggests that TMP21 is genetically associated with Alzheimer's disease, with the novel Tmp21 SNP as a risk factor for Alzheimer's pathogenesis.

Blastic Plasmacytoid Dendritic Cell Neoplasm (BPDCN) Harbors Frequent Splicesosome Mutations That Cause Aberrant RNA Splicing Affecting Genes Critical in pDC Differentiation and Function

Blastic plasmacytoid dendritic cell neoplasm (BPDCN) is an aggressive malignancy thought to result from transformation of plasmacytoid dendritic cells (pDCs). Clinical outcomes are poor and pathogenesis is unclear. To better understand BPDCN genomics and disease mechanisms, we performed whole exome- (12 BPDCNs), targeted DNA- (additional 12 BPDCNs), bulk whole transcriptome RNA- (12 BPDCNs and 6 BPDCN patient-derived xenografts [PDXs]), and single cell RNA-sequencing (scRNA-seq) compared to normal DCs. We observed RNA splicing factor mutations in 16/24 cases (7 ZRSR2, 6 SRSF2, 1 each SF3B1, U2AF1, SF3A2, SF3B4). Additional recurrent alterations were in genes known to be mutated in other blood cancers: TET2, ASXL1, TP53, GNB1, NRAS, IDH2, ETV6, DNMT3A, and RUNX1. From exome sequencing we also discovered recurrent mutations in CRIPAK (6/12 cases), NEFH (4/12), HNF1A (2/12), PAX3 (2/12), and SSC5D (2/12) that may be unique to BPDCN. ZRSR2 is notable among the recurrently mutated splicing factors in hematologic malignancies in that all mutations are loss-of-function (e.g., nonsense, frameshift). Of note, BPDCN is very male predominant, ZRSR2 is located on chrX and all mutations are in males. ZRSR2 plays a critical role in "minor" or U12-type intron splicing (only 0.3% of all introns). Thus, we hypothesized that mis-splicing, possibly of U12 genes, contributes to BPDCN pathogenesis. Using RNA-seq, we measured aberrant splicing in BPDCN. Intron retention was the most frequent abnormality in ZRSR2 mutant BPDCNs and PDXs compared to non-mutant cases. ZRSR2 mutant intron retention predominantly affected U12 introns (patients: 29.4% of retained introns, P<0.0001; PDX: 94%, P<0.0001). To test if ZRSR2 loss directly causes U12 intron retention in otherwise isogenic cells, we performed ZRSR2 knockdown using doxycycline-inducible shRNAs in the BPDCN cell line, CAL1, which has no known splicing factor mutation. RNA-seq was performed 0, 2, and 7 days after addition of doxycycline in 3 independent clones each of control or ZRSR2 knockdown. Consistent with what we observed in primary BPDCN, intron retention events were higher in ZRSR2 compared to control shRNA cells after 7 days of doxycycline (mean 885.7 vs 122.7 events, P=0.041). Aberrant intron retention after ZRSR2 knockdown largely involved U12 introns (30/732 U12 vs 37/207,344 U2 introns, P<0.0001). SRSF2 and SF3B1 mutations in BPDCN were at hotspots seen in other cancers: SRSF2 P95H/L/R and SF3B1 K666N, mutants that induce specific types of aberrant splicing (Kim, Ca Cell 2015; Darman, Cell Rep 2015). Mutant BPDCNs demonstrated the same aberrations: SRSF2, exon inclusion/exclusion based on CCNG/GGNG exonic splicing enhancer motifs; SF3B1, aberrant 3’ splice site recognition. We hypothesized that aberrant splicing may affect RNAs important for pDC development or function. To further define genes uniquely important in BPDCN, we performed scRNA-seq on 4 BPDCNs and on DCs from healthy donors. By principal component analysis, BPDCNs were more similar to pDCs than to conventional DCs (cDCs) or other HLA-DR+ cells. However, several critical genes for pDC function had markedly lower expression in BPDCN including the transcription factors IRF4 and IRF7. Next we determined which genes were commonly mis-spliced in splicing factor mutant BPDCNs. Strikingly, this list included genes already known to be important in driving DC biology or identified in our scRNA-seq as being differentially expressed between BPDCN and healthy pDCs, including IRF7, IRF8, IKZF1, FLT3, and DERL3. To determine if splicing factor mutations affect DC function, we stimulated ZRSR2 knockdown or control CAL1 cells with Toll-like receptor (TLR) 7, 8, and 9 agonists (R848 or CpG oligo). ZRSR2 knockdown inhibited upregulation of the CD80 costimulatory molecule and aggregation of CAL1 cells, suggesting impairment in activation. Using mouse conditional knock-in bone marrow in ex vivo multipotent progenitor assays, DC differentiation induced by FLT3 ligand was biased toward pDCs and away from cDCs in SRSF2 P95H mutant compared to wild-type cells. However, cDC and monocyte differentiation in the presence of GM-CSF was not affected. In conclusion, splicing factors are frequently mutated in BPDCN and lead to specific splicing defects. Splicing factor mutations may promote BPDCN by affecting pathways important in DC maturation or activation, which could contribute to transformation. DisclosuresSeiler:H3 Biomedicine: Employment. Buonamici:H3 Biomedicine: Employment. Lane:Stemline Therapeutics: Research Funding; N-of-1: Consultancy.