The Myc proto-oncogene contributes to the pathogenesis of more than 50 percent of human cancers including prostate cancers. Malignant transformation by Myc is known to transcriptionally upregulate the core pre-mRNA splicing machinery and cause mis-regulation of alternative splicing. However, our understanding of how changes of splice isoforms affect the cancer phenotype is limited and how these changes are directed by Myc remains largely unknown. Here, we performed a signaling pathway-guided differential splicing analysis to identify Myc dependent splicing events, including an HRAS cassette exon that is repressed by Myc transformation across multiple tumor types. To molecularly dissect the regulation of this HRAS cassette exon, we used antisense oligonucleotide tiling to identify intronic splicing enhancers and silencers in the HRAS flanking introns. RNA binding motif prediction indicated the presence of multiple binding sites for hnRNP H and hnRNP F within these cis-regulatory elements. Using siRNA knockdown and cDNA expression, we found that both hnRNP H and F activate the HRAS cassette exon. Mutagenesis and targeted RNA immunoprecipitation implicate two downstream intronic G-rich elements in this H/F mediated splicing activation. Bioinformatic analyses of ENCODE RNAseq datasets also confirmed hnRNP H regulation of HRAS splicing. Consistent with the observed effects of hnRNP H on HRAS splicing, analyses of RNAseq datasets across multiple cancers showed a negative correlation of hnRNP H gene expression with Myc hallmark enrichment. Loss of hnRNP H/F altered the cell cycle progression and induced apoptosis in prostate cancer cells. In the second part of this dissertation, we again applied antisense oligonucleotide tiling to dissect the cis-regulatory elements of another MYC-dependent splicing event, SRSF3 cassette exon, whose inclusion is repressed by Myc transformation in prostate cancer. We identified multiple exonic splicing enhancers as well as downstream intronic splicing silencers. Motif-based RNA binding protein prediction suggested the presence of Sam68 binding motifs in the intronic splicing silencers. siRNA knockdown of Sam68 confirmed it functions as a SRSF3 splicing repressor. Lastly, using RNA-seq data generated from an in vitro model of prostate cancer with inducible Myc expression, we performed RNA binding protein-focused differential gene expression analysis and found majority of differentially expressed RNA binding proteins are downregulated with the depletion of MYC oncoprotein. CRISPRi screening of top Myc-responsive RNA binding proteins suggested that HRAS and SRSF3 are differentially spliced in the loss of RBM28, SNRPD1, EIF3B, and PINX1. Collectively, our results reveal mechanisms for the Myc-dependent regulation of splicing, and point to new possible therapeutic targets in advanced prostate cancers.

Alternative splicing is controlled by diverse RNA binding proteins that recognize elements in the pre-mRNA to alter spliceosome assembly. Separate from their RNA binding domains, these proteins often contain intrinsically disordered domains with regions of low-complexity (LC) sequences, but how LC sequences contribute to splicing regulation is not known. In earlier work, we found that splicing regulators of the Rbfox family are bound with a large complex of proteins called the Large Assembly of Splicing Regulators, LASR. Rbfox proteins were shown to regulate splicing in association with LASR and to alter the activity of LASR components in splicing, but the nature of the Rbfox and LASR interaction was not clear. Here, we show that C-terminal domain of the Rbfox protein interacts with LASR and this interaction is essential for Rbfox activity in splicing. We find that an LC region within the C-terminal domain mediates assembly of Rbfox proteins with LASR into higher-order structures. Repetitive tyrosine residues in this domain are essential to the formation of the higher-order assemblies. The Rbfox LC domain both spontaneously aggregates in solution and forms fibrous structures and hydrogels over time, suggesting a mechanism for higher-order assembly similar to the fibril formation with FUS and other RNA-binding proteins. Exon repression and activation by Rbfox proteins are lost with mutations that disrupt the interaction of Rbfox and LASR. However, blocking higher-order assembly while retaining the Rbfox interaction with LASR, results in selective loss of Rbfox-dependent exon activation. These findings demonstrate that the LC domains of RNA-binding proteins and their self-assembly play a crucial role in splicing regulation. In addition to simple RNA recognition, higher-order assembly and its associated aggregation properties of phase separation and/or fiber formation offer additional mechanisms for tuning regulatory activities.

Alternative splicing is an important form of gene regulation during tissue development. In the mammalian nervous system, complex splicing patterns produce new protein isoforms with different structures and functions specific to neurons. These splicing patterns are regulated in a temporal and cell-specific manner by the expression of specialized pre-mRNA binding proteins (RBPs). The polypyrimidine tract binding (PTB) proteins, PTBP1 and PTBP2, are RBPs that control extensive programs of alternative splicing during neuronal differentiation and maturation. However, the neuronal splicing programs regulated by PTBP1 and PTBP2 are not fully characterized, and the cellular function of their regulated isoforms remains largely unknown. Here we define the PTBP1 splicing program during in vitro neuronal differentiation and show that PTBP1 guides developmental gene expression programs by regulating an alternative exon in the homeodomain transcription factor Pbx1.

We use RNA-sequencing to identify thousands of alternative exons that are differentially spliced as mouse embryonic stem cells (ESC) differentiate into neuronal progenitor cells (NPC) and then into neurons, during the transition from PTBP1 expression to PTBP2. We then define the subset of these neuronal exons controlled by PTBP1 in ESCs and NPCs by RNA-sequencing analysis after PTBP1 depletion and PTBP1 crosslinking-immunoprecipitation (iCLIP-seq). Among these targets, we find that PTBP1 represses splicing of exon 7 in the Pbx1 transcript leading to expression of the Pbx1b isoform in ESCs and NPCs. To study the consequences of this Pbx1 splicing switch, we use the CRISPR-Cas9 technology to induce or eliminate exon 7 splicing in different cell types. We delete regulatory elements in Pbx1 intron 6 to force exon 7 splicing and Pbx1a expression in ESCs. We find that the early expression of Pbx1a enhances expression of a defined group of neurogenic genes, many of which were previously identified as Pbx1 targets. Cas9-targeted deletion of Pbx1 exon 7 forces expression of Pbx1b, with the loss of Pbx1a, in differentiated motor neurons. We find that these cells differentiate normally but exhibit altered expression of Pbx1 target genes. These data demonstrate that one component of the PTBP1 regulatory network is to control the activity of the transcription factor Pbx1 and thus to alter the transcriptional program of neuronal differentiation.

Alternative pre-mRNA splicing (AS) plays a major role in gene expression in higher eukaryotes, especially in vertebrates. Regulation of AS is primarily accomplished by splicing factors that bind to regulatory elements on pre-mRNAs and influence spliceosomal complex assembly. One important splicing factor is the polypyrimidine tract-binding protein 1 (PTBP1), well known for its repressor activity. In repressing an alternative exon, PTBP1 uses different mechanisms depending on the location of its binding sites. PTBP1 can compete with U2AF65 to bind the polypyrimidine tract at the 3’ splice site if it contains the PTBP1 binding sites. Alternatively, when PTBP1 is bound to an exon, it can block the recruitment of U2AF65 across an exon or the cross-intron interactions to an adjacent exon. However, PTBP1 binding sites are often located further within both introns flanking an alternative exon. The mechanism behind this type of alternative exon repression is poorly characterized. Therefore, we used a modified version of an alternative N1 exon of c-src pre-mRNA to characterize this type of repression mechanism. We found that PTBP1 stalls an exon in an early exon complex, inhibiting its transition to an exon definition complex (EDC), which is required for subsequent spliceosomal complex assembly. Using quantitative mass spectrometry to analyze PTBP1-repressed early exon complex, de-repressed early exon complex and the EDC, we identified that PTBP1 accomplishes this, in part, by blocking the association of two ATP-dependent RNA helicases—DDX5 and DDX17, which we found playing an important role promoting efficient formation of the EDC in vitro and exon inclusion in vivo. Interestingly, during the normal transition of the exon from early exon complex to the EDC, we detected the removal of hnRNP proteins and substantial recruitment of SR proteins, concurrent with U2 snRNP recruitment. Not only did these data reveal for the first time the assembly of the EDC, but also suggest that there are remodeling events required prior to the exon definition process. We believe that these remodeling events could be attributed to actions of ATP-dependent RNA helicases DDX5 and DDX17.

Splicing is an essential step in gene regulation to produce mature mRNA from a messenger RNA precursor. Through splicing, exons are assembled to generate a mature mRNA that can be translated into a desired protein by the cell. In mammals, splicing is extensively regulated by splicing factors, resulting in diverse exon arrangements and alternative splicing patterns. To know how splicing factors control genes through alternative splicing, it is critical to understand how they recognize their RNA and exon targets. In this study, we investigated targets of a family of strong splicing regulators, Polypyrimidine tract-binding (PTB) proteins. PTB proteins bind to RNA using their RNA binding domains (RBDs) and regulate splicing of exons. Among PTB proteins, PTBP1 is the most abundant and well-studied gene. Each RBD of PTBP1 binds to short, degenerate pyrimidine sequences allowing PTBP1 regulate a wide variety of targets. This variation makes it difficult to evaluate PTBP1 binding events and thus to identify PTBP1 target exons. We developed computational models that predict the binding and splicing targets of PTBP1. Models identify many previously unrecognized PTBP1 binding sites and novel exon targets. Encouraged by predictive PTBP1 models, we expanded these models to PTBP2, a neuronal paralog of PTBP1. PTBP1 is expressed in non-neuronal and neuronal progenitor cells down-regulating PTBP2. Upon neuronal differentiation, expression of PTBP1 decreases, and PTBP2 is up regulated. To understand why cells require switching between such similar genes, we compared targets for PTBP1 and PTBP2 on a global scale. We were able to assess redundancy and divergence of their binding and splicing codes. With the advance of highthrouput technologies, there are on going efforts to profile PTB-regulated transcriptomes in different tissues and cellular contexts. Integration of multiple datasets will highlight and prioritize direct targets of PTB proteins for future study. To this end, we integrated in vivo targets of PTBP2 from mouse mutant studies and identified potential direct targets. Our approaches can be applied to other splicing factors, and would be especially useful for studying multi-RBD proteins. We expect this study provides a general framework for the analysis of splicing factor binding and identification of functional targets.

Alternative splicing aberrations can cause disease or promote its progression, including the development of cancer. Reversal of pathologic splicing events is a potential therapy for those cases. Efficient methods for identification and characterization of small molecule modulators of specific splicing events are required to develop therapeutics. The cardiotonic steroids, long prescribed for congestive heart failure, alter splicing of many exons. Two methods are described to define their mechanism of action. First, an unbiased analysis of transcriptome-wide changes in alternative splicing predicted loss of function of two specific splicing factors, SRp20 and Tra2-beta. Both proteins are depleted by the cardiotonic steroid digitoxin, and their restoration blocks digitoxin induced splicing changes. This general method can be applied to any identified drugs in future screens. Second, a candidate approach was taken to identify which of the known cardiotonic steroid effectors is required for induced splicing changes. An inhibitor to protein phosphatase 1, tautomycin, antagonizes digitoxin induced changes, through a yet to be determined mechanism.

BIN1 exon 12a inclusion is associated with progression of metastatic melanoma. Using a high throughput approach, cDNA expression and small molecule libraries were screened for regulators of exon 12a splicing. Specific RNA-binding proteins including hnRNP-K, hnRNP-LL, Tra2-beta and PCBP1/4 were identified as repressors while hnRNP-R is an enhancer of 12a splicing. Furthermore, expression of cell signaling effectors such as Map3k3 and -11, as well as CLK2 and FAST can modulate 12a inclusion levels. hnRNP-K, which has conflicting roles in the development of cancer, represses 12a splicing through an exonic RNA element. Small molecules that modulate exon 12a include modulators of prostaglandin activity, the antibiotic puromycin, and the psychoactive MAOA inhibitor harmaline. Interaction between these molecules and the identified genetic regulators of 12a splicing could elucidate the mechanism by which BIN1 splicing changes to promote tumorigenesis.