Development and validation of AI/ML derived splice-switching oligonucleotides.

Splice-switching oligonucleotides (SSOs) are antisense compounds that act directly on pre-mRNA to modulate alternative splicing (AS). This study demonstrates the value that artificial intelligence/machine learning (AI/ML) provides for the identification of functional, verifiable, and therapeutic SSOs. We trained XGboost tree models using splicing factor (SF) pre-mRNA binding profiles and spliceosome assembly information to identify modulatory SSO binding sites on pre-mRNA. Using Shapley and out-of-bag analyses we also predicted the identity of specific SFs whose binding to pre-mRNA is blocked by SSOs. This step adds considerable transparency to AI/ML-driven drug discovery and informs biological insights useful in further validation steps. We applied this approach to previously established functional SSOs to retrospectively identify the SFs likely to regulate those events. We then took a prospective validation approach using a novel target in triple negative breast cancer (TNBC), NEDD4L exon 13 (NEDD4Le13). Targeting NEDD4Le13 with an AI/ML-designed SSO decreased the proliferative and migratory behavior of TNBC cells via downregulation of the TGFß pathway. Overall, this study illustrates the ability of AI/ML to extract actionable insights from RNA-seq data.

Differentiation of mammary tumors and reduction in metastasis upon Malat1 lncRNA loss.

Genome-wide analyses have identified thousands of long noncoding RNAs (lncRNAs). Malat1 (metastasis-associated lung adenocarcinoma transcript 1) is among the most abundant lncRNAs whose expression is altered in numerous cancers. Here we report that genetic loss or systemic knockdown of Malat1 using antisense oligonucleotides (ASOs) in the MMTV (mouse mammary tumor virus)-PyMT mouse mammary carcinoma model results in slower tumor growth accompanied by significant differentiation into cystic tumors and a reduction in metastasis. Furthermore, Malat1 loss results in a reduction of branching morphogenesis in MMTV-PyMT- and Her2/neu-amplified tumor organoids, increased cell adhesion, and loss of migration. At the molecular level, Malat1 knockdown results in alterations in gene expression and changes in splicing patterns of genes involved in differentiation and protumorigenic signaling pathways. Together, these data demonstrate for the first time a functional role of Malat1 in regulating critical processes in mammary cancer pathogenesis. Thus, Malat1 represents an exciting therapeutic target, and Malat1 ASOs represent a potential therapy for inhibiting breast cancer progression.

SRSF1-Regulated Alternative Splicing in Breast Cancer.

Splicing factor SRSF1 is upregulated in human breast tumors, and its overexpression promotes transformation of mammary cells. Using RNA-seq, we identified SRSF1-regulated alternative splicing (AS) targets in organotypic three-dimensional MCF-10A cell cultures that mimic a context relevant to breast cancer. We identified and validated hundreds of endogenous SRSF1-regulated AS events. De novo discovery of the SRSF1 binding motif reconciled discrepancies in previous motif analyses. Using a Bayesian model, we determined positional effects of SRSF1 binding on cassette exons: binding close to the 5' splice site generally promoted exon inclusion, whereas binding near the 3' splice site promoted either exon skipping or inclusion. Finally, we identified SRSF1-regulated AS events deregulated in human tumors; overexpressing one such isoform, exon-9-included CASC4, increased acinar size and proliferation, and decreased apoptosis, partially recapitulating SRSF1's oncogenic effects. Thus, we uncovered SRSF1 positive and negative regulatory mechanisms, and oncogenic AS events that represent potential targets for therapeutics development.

Splicing-factor oncoprotein SRSF1 stabilizes p53 via RPL5 and induces cellular senescence.

Splicing and translation are highly regulated steps of gene expression. Altered expression of proteins involved in these processes can be deleterious. Therefore, the cell has many safeguards against such misregulation. We report that the oncogenic splicing factor SRSF1, which is overexpressed in many cancers, stabilizes the tumor suppressor protein p53 by abrogating its MDM2-dependent proteasomal degradation. We show that SRSF1 is a necessary component of an MDM2/ribosomal protein complex, separate from the ribosome, that functions in a p53-dependent ribosomal-stress checkpoint pathway. Consistent with the stabilization of p53, increased SRSF1 expression in primary human fibroblasts decreases cellular proliferation and ultimately triggers oncogene-induced senescence (OIS). These findings underscore the deleterious outcome of SRSF1 overexpression and identify a cellular defense mechanism against its aberrant function. Furthermore, they implicate the RPL5-MDM2 complex in OIS and demonstrate a link between spliceosomal and ribosomal components, functioning independently of their canonical roles, to monitor cellular physiology and cell-cycle progression.

Downregulation of Survivin contributes to cell-cycle arrest during postnatal cardiac development in a severe spinal muscular atrophy mouse model.

Spinal muscular atrophy (SMA) is the leading genetic cause of infant mortality, characterized by progressive degeneration of spinal-cord motor neurons, leading to atrophy of skeletal muscles. However, accumulating evidence indicates that it is a multi-system disorder, particularly in its severe forms. Several studies delineated structural and functional cardiac abnormalities in SMA patients and mouse models, yet the abnormalities have been primarily attributed to autonomic dysfunction. Here, we show in a severe mouse model that its cardiomyocytes undergo G0/G1 cell-cycle arrest and enhanced apoptosis during postnatal development. Microarray and real-time RT-PCR analyses revealed that a set of genes associated with cell cycle and apoptosis were dysregulated in newborn pups. Of particular interest, the Birc5 gene, which encodes Survivin, an essential protein for heart development, was down-regulated even on pre-symptomatic postnatal day 0. Interestingly, cultured cardiomyocytes depleted of SMN recapitulated the gene expression changes including downregulation of Survivin and abnormal cell-cycle progression; and overexpression of Survivin rescued the cell-cycle defect. Finally, increasing SMN in SMA mice with a therapeutic antisense oligonucleotide improved heart pathology and recovered expression of deregulated genes. Collectively, our data demonstrate that the cardiac malfunction of the severe SMA mouse model is mainly a cell-autonomous defect, caused by widespread gene deregulation in heart tissue, particularly of Birc5, resulting in developmental abnormalities through cell-cycle arrest and apoptosis.

Widespread recognition of 5' splice sites by noncanonical base-pairing to U1 snRNA involving bulged nucleotides.

An established paradigm in pre-mRNA splicing is the recognition of the 5' splice site (5'ss) by canonical base-pairing to the 5' end of U1 small nuclear RNA (snRNA). We recently reported that a small subset of 5'ss base-pair to U1 in an alternate register that is shifted by 1 nucleotide. Using genetic suppression experiments in human cells, we now demonstrate that many other 5'ss are recognized via noncanonical base-pairing registers involving bulged nucleotides on either the 5'ss or U1 RNA strand, which we term "bulge registers." By combining experimental evidence with transcriptome-wide free-energy calculations of 5'ss/U1 base-pairing, we estimate that 10,248 5'ss (∼5% of human 5'ss) in 6577 genes use bulge registers. Several of these 5'ss occur in genes with mutations causing genetic diseases and are often associated with alternative splicing. These results call for a redefinition of an essential element for gene expression that incorporates these registers, with important implications for the molecular classification of splicing mutations and for alternative splicing.

Alternative splicing of SLC39A14 in colorectal cancer is regulated by the Wnt pathway.

Alternative splicing is a crucial step in the generation of protein diversity and its misregulation is observed in many human cancer types. By analyzing 143 colorectal samples using exon arrays, SLC39A14, a divalent cation transporter, was identified as being aberrantly spliced in tumor samples. SLC39A14 contains two mutually exclusive exons 4A and 4B and the exon 4A/4B ratio was significantly altered in adenomas (p = 3.6 × 10(-10)) and cancers (p = 9.4 × 10(-11)), independent of microsatellite stability status. The findings were validated in independent exon array data sets and by quantitative real-time reverse-transcription PCR (qRT-PCR). Aberrant Wnt signaling is a hallmark of colorectal tumorigenesis and is characterized by nuclear ß-catenin. Experimental inactivation of Wnt signaling in DLD1 and Ls174T cells by knockdown of ß-catenin or overexpression of dominant negative TCFs (TCF1 and TCF4) altered the 4A/4B ratio, indicating that SLC39A14 splicing is regulated by the Wnt pathway. An altered 4A/4B ratio was also observed in gastric and lung cancer where Wnt signaling is also known to be aberrantly activated. The splicing factor SRSF1 and its regulator, the kinase SRPK1, were found to be deregulated upon Wnt inactivation in colorectal carcinoma cells. SRPK1 was also found up-regulated in both adenoma samples (p = 1.5 × 10(-5)) and cancer samples (p = 5 × 10(-4)). In silico splicing factor binding analysis predicted SRSF1 to bind predominantly to the cancer associated exon 4B, hence, it was hypothesized that SRPK1 activates SRSF1 through phosphorylation, followed by SRSF1 binding to exon 4B and regulation of SLC39A14 splicing. Indeed, siRNA-mediated knockdown of SRPK1 and SRSF1 in DLD1 and SW480 colorectal cancer cells led to a change in the 4A/4B isoform ratio, supporting a role of these factors in the regulation of SLC39A14 splicing. In conclusion, alternative splicing of SLC39A14 was identified in colorectal tumors and found to be regulated by the Wnt pathway, most likely through regulation of SRPK1 and SRSF1.

Heterogeneous nuclear ribonucleoprotein H1/H2-dependent unsplicing of thymidine phosphorylase results in anticancer drug resistance.

Thymidine phosphorylase (TP) catalyzes the conversion of thymidine to thymine and 2-deoxyribose-1-phosphate. The latter plays an important role in induction of angiogenesis. As such, many human malignancies exhibit TP overexpression that correlates with increased microvessel density, formation of aggressive tumors, and dismal prognosis. Because TP is frequently overexpressed in cancer, pro-drugs were developed that utilize TP activity for their bioactivation to cytotoxic drugs. In this respect, TP is indispensable for the pharmacologic activity of the chemotherapeutic drug capecitabine, as it converts its intermediary metabolite 5'-deoxyfluorouridine to 5-fluorouracil. Thus, loss of TP function confers resistance to the prodrug capecitabine, currently used for the treatment of metastatic colorectal cancer and breast cancer. However, drug resistance phenomena may frequently emerge that compromise the pharmacologic activity of capecitabine. Deciphering the molecular mechanisms underlying resistance to TP-activated prodrugs is an important goal toward the overcoming of such drug resistance phenomena. Here, we discovered that lack of TP protein in drug-resistant tumor cells is due to unsplicing of its pre-mRNA. Advanced bioinformatics identified the family of heterogeneous nuclear ribonucleoproteins (hnRNP) H/F as candidate splicing factors potentially responsible for impaired TP splicing. Indeed, whereas parental cells lacked nuclear localization of hnRNPs H1/H2 and F, drug-resistant cells harbored marked levels of these splicing factors. Nuclear RNA immunoprecipitation experiments established a strong binding of hnRNP H1/H2 to TP pre-mRNA, hence implicating them in TP splicing. Moreover, introduction of hnRNP H2 into drug-sensitive parental cells recapitulated aberrant TP splicing and 5'-deoxyfluorouridine resistance. Thus, this is the first study identifying altered function of hnRNP H1/H2 in tumor cells as a novel determinant of aberrant TP splicing thereby resulting in acquired chemoresistance to TP-activated fluoropyrimidine anticancer drugs.

SpliceTrap: a method to quantify alternative splicing under single cellular conditions.

MOTIVATION: Alternative splicing (AS) is a pre-mRNA maturation process leading to the expression of multiple mRNA variants from the same primary transcript. More than 90% of human genes are expressed via AS. Therefore, quantifying the inclusion level of every exon is crucial for generating accurate transcriptomic maps and studying the regulation of AS. RESULTS: Here we introduce SpliceTrap, a method to quantify exon inclusion levels using paired-end RNA-seq data. Unlike other tools, which focus on full-length transcript isoforms, SpliceTrap approaches the expression-level estimation of each exon as an independent Bayesian inference problem. In addition, SpliceTrap can identify major classes of alternative splicing events under a single cellular condition, without requiring a background set of reads to estimate relative splicing changes. We tested SpliceTrap both by simulation and real data analysis, and compared it to state-of-the-art tools for transcript quantification. SpliceTrap demonstrated improved accuracy, robustness and reliability in quantifying exon-inclusion ratios. CONCLUSIONS: SpliceTrap is a useful tool to study alternative splicing regulation, especially for accurate quantification of local exon-inclusion ratios from RNA-seq data. AVAILABILITY AND IMPLEMENTATION: SpliceTrap can be implemented online through the CSH Galaxy server http://cancan.cshl.edu/splicetrap and is also available for download and installation at http://rulai.cshl.edu/splicetrap/. CONTACT: michael.zhang@utdallas.edu. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

SFmap: a web server for motif analysis and prediction of splicing factor binding sites.

Alternative splicing (AS) is a post-transcriptional process considered to be responsible for the huge diversity of proteins in higher eukaryotes. AS events are regulated by different splicing factors (SFs) that bind to sequence elements on the RNA. SFmap is a web server for predicting putative SF binding sites in genomic data (http://sfmap.technion.ac.il). SFmap implements the COS(WR) algorithm, which computes similarity scores for a given regulatory motif based on information derived from its sequence environment and its evolutionary conservation. Input for SFmap is a human genomic sequence or a list of sequences in FASTA format that can either be uploaded from a file or pasted into a window. SFmap searches within a given sequence for significant hits of binding motifs that are either stored in our database or defined by the user. SFmap results are provided both as a text file and as a graphical web interface.

Predicting and controlling the reactivity of immune cell populations against cancer.

Heterogeneous cell populations form an interconnected network that determine their collective output. One example of such a heterogeneous immune population is tumor-infiltrating lymphocytes (TILs), whose output can be measured in terms of its reactivity against tumors. While the degree of reactivity varies considerably between different TILs, ranging from null to a potent response, the underlying network that governs the reactivity is poorly understood. Here, we asked whether one can predict and even control this reactivity. To address this we measured the subpopulation compositions of 91 TILs surgically removed from 27 metastatic melanoma patients. Despite the large number of subpopulations compositions, we were able to computationally extract a simple set of subpopulation-based rules that accurately predict the degree of reactivity. This raised the conjecture of whether one could control reactivity of TILs by manipulating their subpopulation composition. Remarkably, by rationally enriching and depleting selected subsets of subpopulations, we were able to restore anti-tumor reactivity to nonreactive TILs. Altogether, this work describes a general framework for predicting and controlling the output of a cell mixture.

Does distance matter? Variations in alternative 3' splicing regulation.

Alternative splicing constitutes a major mechanism creating protein diversity in humans. This diversity can result from the alternative skipping of entire exons or by alternative selection of the 5' or 3' splice sites that define the exon boundaries. In this study, we analyze the sequence and evolutionary characteristics of alternative 3' splice sites conserved between human and mouse genomes for distances ranging from 3 to 100 nucleotides. We show that alternative splicing events can be distinguished from constitutive splicing by a combination of properties which vary depending on the distance between the splice sites. Among the unique features of alternative 3' splice sites, we observed an unexpectedly high occurrence of events in which a polypyrimidine tract was found to overlap the upstream splice site. By applying a machine-learning approach, we show that we can successfully discriminate true alternative 3' splice sites from constitutive 3' splice sites. Finally, we propose that the unique features of the intron flanking alternative splice sites are indicative of a regulatory mechanism that is involved in splice site selection. We postulate that the process of splice site selection is influenced by the distance between the competitive splice sites.

Differential Functions of Splicing Factors in Mammary Transformation and Breast Cancer Metastasis.

Misregulation of alternative splicing is a hallmark of human tumors, yet to what extent and how it contributes to malignancy are only beginning to be unraveled. Here, we define which members of the splicing factor SR and SR-like families contribute to breast cancer and uncover differences and redundancies in their targets and biological functions. We identify splicing factors frequently altered in human breast tumors and assay their oncogenic functions using breast organoid models. We demonstrate that not all splicing factors affect mammary tumorigenesis in MCF-10A cells. Specifically, the upregulation of SRSF4, SRSF6, or TRA2ß disrupts acinar morphogenesis and promotes cell proliferation and invasion in MCF-10A cells. By characterizing the targets of these oncogenic splicing factors, we identify shared spliced isoforms associated with well-established cancer hallmarks. Finally, we demonstrate that TRA2ß is regulated by the MYC oncogene, plays a role in metastasis maintenance in vivo, and its levels correlate with breast cancer patient survival.

Alternative splicing regulation at tandem 3' splice sites.

Alternative splicing (AS) constitutes a major mechanism creating protein diversity in humans. Previous bioinformatics studies based on expressed sequence tag and mRNA data have identified many AS events that are conserved between humans and mice. Of these events, approximately 25% are related to alternative choices of 3' and 5' splice sites. Surprisingly, half of all these events involve 3' splice sites that are exactly 3 nt apart. These tandem 3' splice sites result from the presence of the NAGNAG motif at the acceptor splice site, recently reported to be widely spread in the human genome. Although the NAGNAG motif is common in human genes, only a small subset of sites with this motif is confirmed to be involved in AS. We examined the NAGNAG motifs and observed specific features such as high sequence conservation of the motif, high conservation of approximately 30 bp at the intronic regions flanking the 3' splice site and overabundance of cis-regulatory elements, which are characteristic of alternatively spliced tandem acceptor sites and can distinguish them from the constitutive sites in which the proximal NAG splice site is selected. Our findings imply that AS at tandem splice sites and constitutive splicing of the distal NAG are highly regulated.

Differential connectivity of splicing activators and repressors to the human spliceosome.

BACKGROUND: During spliceosome assembly, protein-protein interactions (PPI) are sequentially formed and disrupted to accommodate the spatial requirements of pre-mRNA substrate recognition and catalysis. Splicing activators and repressors, such as SR proteins and hnRNPs, modulate spliceosome assembly and regulate alternative splicing. However, it remains unclear how they differentially interact with the core spliceosome to perform their functions. RESULTS: Here, we investigate the protein connectivity of SR and hnRNP proteins to the core spliceosome using probabilistic network reconstruction based on the integration of interactome and gene expression data. We validate our model by immunoprecipitation and mass spectrometry of the prototypical splicing factors SRSF1 and hnRNPA1. Network analysis reveals that a factor's properties as an activator or repressor can be predicted from its overall connectivity to the rest of the spliceosome. In addition, we discover and experimentally validate PPIs between the oncoprotein SRSF1 and members of the anti-tumor drug target SF3 complex. Our findings suggest that activators promote the formation of PPIs between spliceosomal sub-complexes, whereas repressors mostly operate through protein-RNA interactions. CONCLUSIONS: This study demonstrates that combining in-silico modeling with biochemistry can significantly advance the understanding of structure and function relationships in the human spliceosome.

SCT Promoter Methylation is a Highly Discriminative Biomarker for Lung and Many Other Cancers.

Aberrant DNA methylation has long been implicated in cancers. In this work we present a highly discriminative DNA methylation biomarker for non-small cell lung cancers and fourteen other cancers. Based on 69 NSCLC cell lines and 257 cancer-free lung tissues we identified a CpG island in SCT gene promoter which was verified by qMSP experiment in 15 NSCLC cell lines and 3 immortalized human respiratory epithelium cells. In addition, we found that SCT promoter was methylated in 23 cancer cell lines involving >10 cancer types profiled by ENCODE. We found that SCT promoter is hyper-methylated in primary tumors from TCGA lung cancer cohort. Additionally, we found that SCT promoter is methylated at high frequencies in fifteen malignancies and is not methylated in~1000 non-cancerous tissues across >30 organ types. Our study indicates that SCT promoter methylation is a highly discriminative biomarker for lung and many other cancers.

Oncogenic splicing factor SRSF1 is a critical transcriptional target of MYC.

The SR protein splicing factor SRSF1 is a potent proto-oncogene that is frequently upregulated in cancer. Here, we show that SRSF1 is a direct target of the transcription factor oncoprotein MYC. These two oncogenes are significantly coexpressed in lung carcinomas, and MYC knockdown downregulates SRSF1 expression in lung-cancer cell lines. MYC directly activates transcription of SRSF1 through two noncanonical E-boxes in its promoter. The resulting increase in SRSF1 protein is sufficient to modulate alternative splicing of a subset of transcripts. In particular, MYC induction leads to SRSF1-mediated alternative splicing of the signaling kinase MKNK2 and the transcription factor TEAD1. SRSF1 knockdown reduces MYC's oncogenic activity, decreasing proliferation and anchorage-independent growth. These results suggest a mechanism for SRSF1 upregulation in tumors with elevated MYC and identify SRSF1 as a critical MYC target that contributes to its oncogenic potential by enabling MYC to regulate the expression of specific protein isoforms through alternative splicing.

Exon-centric regulation of pyruvate kinase M alternative splicing via mutually exclusive exons.

Alternative splicing of the pyruvate kinase M gene (PK-M) can generate the M2 isoform and promote aerobic glycolysis and tumor growth. However, the cancer-specific alternative splicing regulation of PK-M is not completely understood. Here, we demonstrate that PK-M is regulated by reciprocal effects on the mutually exclusive exons 9 and 10, such that exon 9 is repressed and exon 10 is activated in cancer cells. Strikingly, exonic, rather than intronic, cis-elements are key determinants of PK-M splicing isoform ratios. Using a systematic sub-exonic duplication approach, we identify a potent exonic splicing enhancer in exon 10, which differs from its homologous counterpart in exon 9 by only two nucleotides. We identify SRSF3 as one of the cognate factors, and show that this serine/arginine-rich protein activates exon 10 and mediates changes in glucose metabolism. These findings provide mechanistic insights into the complex regulation of alternative splicing of a key regulator of the Warburg effect, and also have implications for other genes with a similar pattern of alternative splicing.

The splicing factor SRSF1 regulates apoptosis and proliferation to promote mammary epithelial cell transformation.

The splicing-factor oncoprotein SRSF1 (also known as SF2/ASF or ASF/SF2) is upregulated in breast cancers. We investigated the ability of SRSF1 to transform human and mouse mammary epithelial cells in vivo and in vitro. SRSF1-overexpressing COMMA-1D cells formed tumors, following orthotopic transplantation to reconstitute the mammary gland. In three-dimensional (3D) culture, SRSF1-overexpressing MCF-10A cells formed larger acini than control cells, reflecting increased proliferation and delayed apoptosis during acinar morphogenesis. These effects required the first RNA-recognition motif and nuclear functions of SRSF1. SRSF1 overexpression promoted alternative splicing of BIM (also known as BCL2L11) and BIN1 to produce isoforms that lack pro-apoptotic functions and contribute to the phenotype. Finally, SRSF1 cooperated specifically with MYC to transform mammary epithelial cells, in part by potentiating eIF4E activation, and these cooperating oncogenes are significantly coexpressed in human breast tumors. Thus, SRSF1 can promote breast cancer, and SRSF1 itself or its downstream effectors may be valuable targets for the development of therapeutics.

Ancient diversity of splicing motifs and protein surfaces in the wild emmer wheat (Triticum dicoccoides) LR10 coiled coil (CC) and leucine-rich repeat (LRR) domains.

In this study, we explore the diversity and its distribution along the wheat leaf rust resistance protein LR10 three-dimensional structure. Lr10 is a leaf rust resistance gene encoding a coiled coil-nucleotide-binding site-leucine-rich repeat (CC-NBS-LRR) class of protein. Lr10 was cloned and sequenced from 58 accessions representing diverse habitats of wild emmer wheat in Israel. Nucleotide diversity was very high relative to other wild emmer wheat genes (π= 0.029). The CC domain was found to be the most diverse domain and subject to positive selection. Superimposition of the diversity on the CC three-dimensional structure showed that some of the variable and positively selected residues were solvent exposed and may interact with other proteins. The LRR domain was relatively conserved, but showed a hotspot of amino acid variation between two haplotypes in the ninth repeat. This repeat was longer than the other LRRs, and three-dimensional modelling suggested that an extensive α helix structure was formed in this region. The two haplotypes also differed in splicing regulation motifs. In genotypes with one haplotype, an intron was alternatively spliced in this region, whereas, in genotypes with the other haplotype, this intron did not splice at all. The two haplotypes are proposed to be ancient and maintained by balancing selection.