BPP: a sequence-based algorithm for branch point prediction

A deep intronic PHEX variant associated with X-linked hypophosphatemia in a Finnish family

Hypophosphatemic rickets is a rare bone disease characterized by short stature, bone deformities, impaired bone mineralization, and dental problems. Most commonly, hypophosphatemic rickets is caused by pathogenic variants in the X-chromosomal PHEX gene, but autosomal dominant and recessive forms also exist. We investigated a Finnish family in which the son (index, 29 yr) and mother (56 yr) had hypophosphatemia since childhood. Both patients had typical clinical, radiographic, and biochemical features of hypophosphatemic rickets, including a pathological fracture in the son. Gene panels and whole-exome sequencing did not reveal any pathogenic variants in the known hypophosphatemia genes. Therefore, we performed whole genome sequencing and identified a deep intronic variant (c.2147 + 1197A > G) in PHEX. Both the affected individuals, but none of the unaffected family members, had the same variant, as confirmed by Sanger sequencing. According to RT-PCR, whole transcriptomic data, and in silico analyses, the variant led to a new splice donor site in intron 21 and an 84 basepair pseudoexon between exons 21 and 22, likely leading to the synthesis of abnormal PHEX protein. Our study underscores the importance of intronic PHEX variants in X-linked hypophosphatemia (XLH). In patients with features of XLH but negative gene panel or whole-exome sequencing results, the combination of whole-genome sequencing and whole transcriptomics should be considered to detect possible deep intronic variants. The methodologies presented have the potential to be used more widely in other rare diseases.

Two novel deep intronic variants cause Duchenne muscular dystrophy by splice-altering mechanism

Duchenne muscular dystrophy (DMD) is a genetic disorder characterized by progressive muscle degeneration and weakness, due to mutations in the DMD gene, which encodes the dystrophin protein. While mutations within the coding regions of DMD have been extensively studied, recent focus has shifted to deep intronic variants for their potential impact on disease severity. Here, we characterize two deep intronic variants, c.8669-19_8669-24del and c.6439-1016_6439-3376del, in unrelated DMD patients. These variants were identified using targeted long-read sequencing on patients' DNA. RNA sequencing/reverse transcription polymerase chain reaction on RNA extracted from muscle biopsies revealed the presence of a pseudoexon or retention of part of the intron in the transcript, resulting in the introduction of premature termination codons. This study enhances our understanding of pseudoexon activation mechanisms in DMD and underscores the diverse genetic abnormalities contributing to the disease's complexity.

Refining clinically relevant parameters for mis-splicing risk in shortened introns with donor-to-branchpoint space constraint

Intronic deletions that critically shorten donor-to-branchpoint (D-BP) distance of a precursor mRNA impose biophysical space constraint on assembly of the U1/U2 spliceosomal complex, leading to canonical splicing failure. Here we use a series of β-globin (HBB) gene constructs with intron 1 deletions to define D-BP lengths that present low/no risk of mis-splicing and lengths which are critically short and likely elicit clinically relevant mis-splicing. We extend our previous observation in EMD intron 5 of 46 nt as the minimal productive D-BP length, demonstrating spliceosome assembly constraint persists at D-BP lengths of 47-56 nt. We exploit the common HBB exon 1 β-thalassemia variant that strengthens a cryptic donor (NM_000518.5(HBB):c.79G > A) to provide a simple barometer for the earliest signs of space constraint, via cryptic donor activation. For clinical evaluation of intronic deletions, we assert D-BP lengths > 60 nt present low mis-splicing risk while space constraint increases exponentially with D-BP lengths < 55 nt, with critical risk and profound splicing abnormalities with D-BP lengths < 50 nt.

Loss-of-function mutation in PRMT9 causes abnormal synapse development by dysregulation of RNA alternative splicing

Protein arginine methyltransferase 9 (PRMT9) is a recently identified member of the PRMT family, yet its biological function remains largely unknown. Here, by characterizing an intellectual disability associated PRMT9 mutation (G189R) and establishing a Prmt9 conditional knockout (cKO) mouse model, we uncover an important function of PRMT9 in neuronal development. The G189R mutation abolishes PRMT9 methyltransferase activity and reduces its protein stability. Knockout of Prmt9 in hippocampal neurons causes alternative splicing of ~1900 genes, which likely accounts for the aberrant synapse development and impaired learning and memory in the Prmt9 cKO mice. Mechanistically, we discover a methylation-sensitive protein-RNA interaction between the arginine 508 (R508) of the splicing factor 3B subunit 2 (SF3B2), the site that is exclusively methylated by PRMT9, and the pre-mRNA anchoring site, a cis-regulatory element that is critical for RNA splicing. Additionally, using human and mouse cell lines, as well as an SF3B2 arginine methylation-deficient mouse model, we provide strong evidence that SF3B2 is the primary methylation substrate of PRMT9, thus highlighting the conserved function of the PRMT9/SF3B2 axis in regulating pre-mRNA splicing.

Self-supervised learning on millions of primary RNA sequences from 72 vertebrates improves sequence-based RNA splicing prediction

Language models pretrained by self-supervised learning (SSL) have been widely utilized to study protein sequences, while few models were developed for genomic sequences and were limited to single species. Due to the lack of genomes from different species, these models cannot effectively leverage evolutionary information. In this study, we have developed SpliceBERT, a language model pretrained on primary ribonucleic acids (RNA) sequences from 72 vertebrates by masked language modeling, and applied it to sequence-based modeling of RNA splicing. Pretraining SpliceBERT on diverse species enables effective identification of evolutionarily conserved elements. Meanwhile, the learned hidden states and attention weights can characterize the biological properties of splice sites. As a result, SpliceBERT was shown effective on several downstream tasks: zero-shot prediction of variant effects on splicing, prediction of branchpoints in humans, and cross-species prediction of splice sites. Our study highlighted the importance of pretraining genomic language models on a diverse range of species and suggested that SSL is a promising approach to enhance our understanding of the regulatory logic underlying genomic sequences.

Exonic splicing code and coordination of divalent metals in proteins

Exonic sequences contain both protein-coding and RNA splicing information but the interplay of the protein and splicing code is complex and poorly understood. Here, we have studied traditional and auxiliary splicing codes of human exons that encode residues coordinating two essential divalent metals at the opposite ends of the Irving-Williams series, a universal order of relative stabilities of metal-organic complexes. We show that exons encoding Zn2+-coordinating amino acids are supported much less by the auxiliary splicing motifs than exons coordinating Ca2+. The handicap of the former is compensated by stronger splice sites and uridine-richer polypyrimidine tracts, except for position -3 relative to 3' splice junctions. However, both Ca2+ and Zn2+ exons exhibit close-to-constitutive splicing in multiple tissues, consistent with their critical importance for metalloprotein function and a relatively small fraction of expendable, alternatively spliced exons. These results indicate that constraints imposed by metal coordination spheres on RNA splicing have been efficiently overcome by the plasticity of exon-intron architecture to ensure adequate metalloprotein expression.

Exploiting public databases of genomic variation to quantify evolutionary constraint on the branch point sequence in 30 plant and animal species

The branch point sequence is a degenerate intronic heptamer required for the assembly of the spliceosome during pre-mRNA splicing. Disruption of this motif may promote alternative splicing and eventually cause phenotype variation. Despite its functional relevance, the branch point sequence is not included in most genome annotations. Here, we predict branch point sequences in 30 plant and animal species and attempt to quantify their evolutionary constraints using public variant databases. We find an implausible variant distribution in the databases from 16 of 30 examined species. Comparative analysis of variants from whole-genome sequencing shows that variants submitted from exome sequencing or false positive variants are widespread in public databases and cause these irregularities. We then investigate evolutionary constraint with largely unbiased public variant databases in 14 species and find that the fourth and sixth position of the branch point sequence are more constrained than coding nucleotides. Our findings show that public variant databases should be scrutinized for possible biases before they qualify to analyze evolutionary constraint.

Genetic alterations in the BCR-ABL1 fusion gene related to imatinib resistance in chronic myeloid leukemia

Use of the potent tyrosine kinase inhibitor imatinib as the first-line treatment in chronic myeloid leukemia (CML) has decreased mortality from 20% to 2%. Approximately 30% of CML patients experience imatinib resistance, however, largely because of point mutations in the kinase domain of the BCR-ABL1 fusion gene. The aim of this study was to use next-generation sequencing (NGS) to identify mutations related to imatinib resistance. The study included 22 patients diagnosed with CML and experiencing no clinical response to imatinib. Total RNA was used for cDNA synthesis, with amplification of a fragment encompassing the BCR-ABL1 kinase domain using a nested-PCR approach. Sanger and NGS were applied to detect genetic alterations. HaplotypeCaller was used for variant calling, and STAR-Fusion software was applied for fusion breakpoint identification. After sequencing analysis, F311I, F317L, and E450K mutations were detected respectively in three different participants, and in another two patients, single nucleotide variants in BCR (rs9608100, rs140506, rs16802) and ABL1 (rs35011138) were detected. Eleven patients carried e14a2 transcripts, nine had e13a2 transcripts, and both transcripts were identified in one patient. One patient had co-expression of e14a2 and e14a8 transcripts. The results identify candidate single nucleotide variants and co-expressed BCR-ABL1 transcripts in cellular resistance to imatinib.

A progeroid syndrome caused by a deep intronic variant in TAPT1 is revealed by RNA/SI-NET sequencing

Exome sequencing has introduced a paradigm shift for the identification of germline variations responsible for Mendelian diseases. However, non-coding regions, which make up 98% of the genome, cannot be captured. The lack of functional annotation for intronic and intergenic variants makes RNA-seq a powerful companion diagnostic. Here, we illustrate this point by identifying six patients with a recessive Osteogenesis Imperfecta (OI) and neonatal progeria syndrome. By integrating homozygosity mapping and RNA-seq, we delineated a deep intronic TAPT1 mutation (c.1237-52 G>A) that segregated with the disease. Using SI-NET-seq, we document that TAPT1's nascent transcription was not affected in patients' fibroblasts, indicating instead that this variant leads to an alteration of pre-mRNA processing. Predicted to serve as an alternative splicing branchpoint, this mutation enhances TAPT1 exon 12 skipping, creating a protein-null allele. Additionally, our study reveals dysregulation of pathways involved in collagen and extracellular matrix biology in disease-relevant cells. Overall, our work highlights the power of transcriptomic approaches in deciphering the repercussions of non-coding variants, as well as in illuminating the molecular mechanisms of human diseases.

The evolution of pre-mRNA splicing and its machinery revealed by reduced extremophilic red algae

The Cyanidiales are a group of mostly thermophilic and acidophilic red algae that thrive near volcanic vents. Despite their phylogenetic relationship, the reduced genomes of Cyanidioschyzon merolae and Galdieria sulphuraria are strikingly different with respect to pre-mRNA splicing, a ubiquitous eukaryotic feature. Introns are rare and spliceosomal machinery is extremely reduced in C. merolae, in contrast to G. sulphuraria. Previous studies also revealed divergent spliceosomes in the mesophilic red alga Porphyridium purpureum and the red algal derived plastid of Guillardia theta (Cryptophyta), along with unusually high levels of unspliced transcripts. To further examine the evolution of splicing in red algae, we compared C. merolae and G. sulphuraria, investigating splicing levels, intron position, intron sequence features, and the composition of the spliceosome. In addition to identifying 11 additional introns in C. merolae, our transcriptomic analysis also revealed typical eukaryotic splicing in G. sulphuraria, whereas most transcripts in C. merolae remain unspliced. The distribution of intron positions within their host genes was examined to provide insight into patterns of intron loss in red algae. We observed increasing variability of 5' splice sites and branch donor regions with increasing intron richness. We also found these relationships to be connected to reductions in and losses of corresponding parts of the spliceosome. Our findings highlight patterns of intron and spliceosome evolution in related red algae under the pressures of genome reduction.

Computational prediction of human deep intronic variation

BACKGROUND

The adoption of whole-genome sequencing in genetic screens has facilitated the detection of genetic variation in the intronic regions of genes, far from annotated splice sites. However, selecting an appropriate computational tool to discriminate functionally relevant genetic variants from those with no effect is challenging, particularly for deep intronic regions where independent benchmarks are scarce.

RESULTS

In this study, we have provided an overview of the computational methods available and the extent to which they can be used to analyze deep intronic variation. We leveraged diverse datasets to extensively evaluate tool performance across different intronic regions, distinguishing between variants that are expected to disrupt splicing through different molecular mechanisms. Notably, we compared the performance of SpliceAI, a widely used sequence-based deep learning model, with that of more recent methods that extend its original implementation. We observed considerable differences in tool performance depending on the region considered, with variants generating cryptic splice sites being better predicted than those that potentially affect splicing regulatory elements. Finally, we devised a novel quantitative assessment of tool interpretability and found that tools providing mechanistic explanations of their predictions are often correct with respect to the ground - information, but the use of these tools results in decreased predictive power when compared to black box methods.

CONCLUSIONS

Our findings translate into practical recommendations for tool usage and provide a reference framework for applying prediction tools in deep intronic regions, enabling more informed decision-making by practitioners.

SPiP: Splicing Prediction Pipeline, a machine learning tool for massive detection of exonic and intronic variant effects on mRNA splicing

Modeling splicing is essential for tackling the challenge of variant interpretation as each nucleotide variation can be pathogenic by affecting pre-mRNA splicing via disruption/creation of splicing motifs such as 5'/3' splice sites, branch sites, or splicing regulatory elements. Unfortunately, most in silico tools focus on a specific type of splicing motif, which is why we developed the Splicing Prediction Pipeline (SPiP) to perform, in one single bioinformatic analysis based on a machine learning approach, a comprehensive assessment of the variant effect on different splicing motifs. We gathered a curated set of 4616 variants scattered all along the sequence of 227 genes, with their corresponding splicing studies. The Bayesian analysis provided us with the number of control variants, that is, variants without impact on splicing, to mimic the deluge of variants from high-throughput sequencing data. Results show that SPiP can deal with the diversity of splicing alterations, with 83.13% sensitivity and 99% specificity to detect spliceogenic variants. Overall performance as measured by area under the receiving operator curve was 0.986, better than SpliceAI and SQUIRLS (0.965 and 0.766) for the same data set. SPiP lends itself to a unique suite for comprehensive prediction of spliceogenicity in the genomic medicine era. SPiP is available at: https://sourceforge.net/projects/splicing-prediction-pipeline/.

Genome-wide detection of human variants that disrupt intronic branchpoints

Pre-messenger RNA splicing is initiated with the recognition of a single-nucleotide intronic branchpoint (BP) within a BP motif by spliceosome elements. Forty-eight rare variants in 43 human genes have been reported to alter splicing and cause disease by disrupting BP. However, until now, no computational approach was available to efficiently detect such variants in massively parallel sequencing data. We established a comprehensive human genome-wide BP database by integrating existing BP data and generating new BP data from RNA sequencing of lariat debranching enzyme DBR1-mutated patients and from machine-learning predictions. We characterized multiple features of BP in major and minor introns and found that BP and BP-2 (two nucleotides upstream of BP) positions exhibit a lower rate of variation in human populations and higher evolutionary conservation than the intronic background, while being comparable to the exonic background. We developed BPHunter as a genome-wide computational approach to systematically and efficiently detect intronic variants that may disrupt BP recognition. BPHunter retrospectively identified 40 of the 48 known pathogenic BP variants, in which we summarized a strategy for prioritizing BP variant candidates. The remaining eight variants all create AG-dinucleotides between the BP and acceptor site, which is the likely reason for missplicing. We demonstrated the practical utility of BPHunter prospectively by using it to identify a novel germline heterozygous BP variant of STAT2 in a patient with critical COVID-19 pneumonia and a novel somatic intronic 59-nucleotide deletion of ITPKB in a lymphoma patient, both of which were validated experimentally. BPHunter is publicly available from https://hgidsoft.rockefeller.edu/BPHunter and https://github.com/casanova-lab/BPHunter.

Prevalence, parameters, and pathogenic mechanisms for splice-altering acceptor variants that disrupt the AG exclusion zone

Predicting the pathogenicity of acceptor splice-site variants outside the essential AG is challenging, due to high sequence diversity of the extended splice-site region. Critical analysis of 24,445 intronic extended acceptor splice-site variants reported in ClinVar and the Leiden Open Variation Database (LOVD) demonstrates 41.9% of pathogenic variants create an AG dinucleotide between the predicted branchpoint and acceptor (AG-creating variants in the AG exclusion zone), 28.4% result in loss of a pyrimidine at the -3 position, and 15.1% result in loss of one or more pyrimidines in the polypyrimidine tract. Pathogenicity of AG-creating variants was highly influenced by their position. We define a high-risk zone for pathogenicity: > 6 nucleotides downstream of the predicted branchpoint and >5 nucleotides upstream from the acceptor, where 93.1% of pathogenic AG-creating variants arise and where naturally occurring AG dinucleotides are concordantly depleted (5.8% of natural AGs). SpliceAI effectively predicts pathogenicity of AG-creating variants, achieving 95% sensitivity and 69% specificity. We highlight clinical examples showing contrasting mechanisms for mis-splicing arising from AG variants: (1) cryptic acceptor created; (2) splicing silencer created: an introduced AG silences the acceptor, resulting in exon skipping, intron retention, and/or use of an alternative existing cryptic acceptor; and (3) splicing silencer disrupted: loss of a deep intronic AG activates inclusion of a pseudo-exon. In conclusion, we establish AG-creating variants as a common class of pathogenic extended acceptor variant and outline factors conferring critical risk for mis-splicing for AG-creating variants in the AG exclusion zone, between the branchpoint and acceptor.

Dynamic Transcriptional Landscape of Grass Carp (Ctenopharyngodon idella) Reveals Key Transcriptional Features Involved in Fish Development

A high-quality baseline transcriptome is a valuable resource for developmental research as well as a useful reference for other studies. We gathered 41 samples representing 11 tissues/organs from 22 important developmental time points within 197 days of fertilization of grass carp eggs in order to systematically examine the role of lncRNAs and alternative splicing in fish development. We created a high-quality grass carp baseline transcriptome with a completeness of up to 93.98 percent by combining strand-specific RNA sequencing and single-molecule real-time RNA sequencing technologies, and we obtained temporal expression profiles of 33,055 genes and 77,582 transcripts during development and tissue differentiation. A family of short interspersed elements was preferentially expressed at the early stage of zygotic activation in grass carp, and its possible regulatory components were discovered through analysis. Additionally, after thoroughly analyzing alternative splicing events, we discovered that retained intron (RI) alternative splicing events change significantly in both zygotic activation and tissue differentiation. During zygotic activation, we also revealed the precise regulatory characteristics of the underlying functional RI events.

Performance evaluation of computational methods for splice-disrupting variants and improving the performance using the machine learning-based framework

A critical challenge in genetic diagnostics is the assessment of genetic variants associated with diseases, specifically variants that fall out with canonical splice sites, by altering alternative splicing. Several computational methods have been developed to prioritize variants effect on splicing; however, performance evaluation of these methods is hampered by the lack of large-scale benchmark datasets. In this study, we employed a splicing-region-specific strategy to evaluate the performance of prediction methods based on eight independent datasets. Under most conditions, we found that dbscSNV-ADA performed better in the exonic region, S-CAP performed better in the core donor and acceptor regions, S-CAP and SpliceAI performed better in the extended acceptor region and MMSplice performed better in identifying variants that caused exon skipping. However, it should be noted that the performances of prediction methods varied widely under different datasets and splicing regions, and none of these methods showed the best overall performance with all datasets. To address this, we developed a new method, machine learning-based classification of splice sites variants (MLCsplice), to predict variants effect on splicing based on individual methods. We demonstrated that MLCsplice achieved stable and superior prediction performance compared with any individual method. To facilitate the identification of the splicing effect of variants, we provided precomputed MLCsplice scores for all possible splice sites variants across human protein-coding genes (http://39.105.51.3:8090/MLCsplice/). We believe that the performance of different individual methods under eight benchmark datasets will provide tentative guidance for appropriate method selection to prioritize candidate splice-disrupting variants, thereby increasing the genetic diagnostic yield.

Identification of an elusive spliceogenic MYBPC3 variant in an otherwise genotype-negative hypertrophic cardiomyopathy pedigree

The finding of a genotype-negative hypertrophic cardiomyopathy (HCM) pedigree with several affected members indicating a familial origin of the disease has driven this study to discover causative gene variants. Genetic testing of the proband and subsequent family screening revealed the presence of a rare variant in the MYBPC3 gene, c.3331−26T>G in intron 30, with evidence supporting cosegregation with the disease in the family. An analysis of potential splice-altering activity using several splicing algorithms consistently yielded low scores. Minigene expression analysis at the mRNA and protein levels revealed that c.3331−26T>G is a spliceogenic variant with major splice-altering activity leading to undetectable levels of properly spliced transcripts or the corresponding protein. Minigene and patient mRNA analyses indicated that this variant induces complete and partial retention of intron 30, which was expected to lead to haploinsufficiency in carrier patients. As most spliceogenic MYBPC3 variants, c.3331−26T>G appears to be non-recurrent, since it was identified in only two additional unrelated probands in our large HCM cohort. In fact, the frequency analysis of 46 known splice-altering MYBPC3 intronic nucleotide substitutions in our HCM cohort revealed 9 recurrent and 16 non-recurrent variants present in a few probands (≤ 4), while 21 were not detected. The identification of non-recurrent elusive MYBPC3 spliceogenic variants that escape detection by in silico algorithms represents a challenge for genetic diagnosis of HCM and contributes to solving a fraction of genotype-negative HCM cases.

From the comparative study of a circRNA originating from an mammalian ATXN2L intron to understanding the genesis of intron lariat-derived circRNAs

Circular intronic RNAs (ciRNAs) are still unexplored regarding mechanisms for their emergence. We considered the ATXN2L intron lariat-derived circular RNA (ciRNA-ATXN2L) as an opportunity to conduct a cross-species examination of ciRNA genesis. To this end, we investigated 207 datasets from 4 tissues and from 13 mammalian species. While in eight species, ciRNA-ATXN2L was never detected, in pigs and rabbits, ciRNA-ATXN2L was expressed in all tissues and sometimes at very high levels. Bovine tissues were an intermediate case and in macaques and cats, only ciRNA-ATXN2L traces were detected. The pattern of ciRNA-ATXN2L restricted to only five species is not related to a particular evolution of intronic sequences. To empower our analysis, we considered 221 additional introns including 80 introns where a lariat-derived ciRNA was previously described. The primary driver of micro-ciRNA genesis (< 155 nt as ciRNA-ATXN2L) appears to be the absence of a canonical "A" (i.e. a "tnA" located in the usual branching region) to build the lariat around this adenosine. The balance between available "non canonical-A" (no ciRNA genesis) and "non-A" (ciRNA genesis) for use as a branch point to build the lariat could modify the expression level of ciRNA-ATXN2L. In addition, the rare localization of the 2'-5' bond in an open RNA secondary structure could also negatively affect the lifetime of ciRNAs (macaque ciRNA-ATXN2L). Our analyses suggest that ciRNA-ATXN2L is likely a functionless splice remnant. This study provides a better understanding of the ciRNAs origin, especially drivers for micro ciRNA genesis.

Piwi/piRNAs control food intake by promoting neuropeptide F expression in locusts

Animal feeding, which directly affects growth and metabolism, is an important physiological process. However, the contribution of PIWI proteins and PIWI‐interacting RNAs (piRNAs) to the regulatory mechanism of animal feeding is unknown. Here, we report a novel function of Piwi and piRNAs in regulating food intake in locusts. Our study shows that the locust can serve as a representative species for determining PIWI function in insects. Knockdown of Piwi1 expression suppresses anabolic processes and reduces food consumption and body weight. The reduction in food intake by knockdown of Piwi1 expression results from decreased expression of neuropeptide NPF1 in a piRNA‐dependent manner. Mechanistically, intronic piRNAs might enhance RNA splicing of NPF1 by preventing hairpin formation at the branch point sites. These results suggest a novel nuclear PIWI/piRNA‐mediated mechanism that controls food intake in the locust nervous system.

The splicing effect of variants at branchpoint elements in cancer genes

PURPOSE

Branchpoint elements are required for intron removal, and variants at these elements can result in aberrant splicing. We aimed to assess the value of branchpoint annotations generated from recent large-scale studies to select branchpoint-abrogating variants, using hereditary cancer genes as model.

METHODS

We identified branchpoint elements in 119 genes associated with hereditary cancer from 3 genome-wide experimentally-inferred and 2 predicted branchpoint data sets. We then identified variants that occur within branchpoint elements from public databases. We compared conservation, unique variant observations, and population frequencies at different nucleotides within branchpoint motifs. Finally, selected minigene assays were performed to assess the splicing effect of variants at branchpoint elements within mismatch repair genes.

RESULTS

There was poor overlap between predicted and experimentally-inferred branchpoints. Our analysis of cancer genes suggested that variants at -2 nucleotide, -1 nucleotide, and branchpoint positions in experimentally-inferred canonical motifs are more likely to be clinically relevant. Minigene assay data showed the -2 nucleotide to be more important to branchpoint motif integrity but also showed fluidity in branchpoint usage.

CONCLUSION

Data from cancer gene analysis suggest that there are few high-risk alleles that severely impact function via branchpoint abrogation. Results of this study inform a general scheme to prioritize branchpoint motif variants for further study.

Comprehensive profiling of mRNA splicing indicates that GC content signals altered cassette exon inclusion in Ewing sarcoma

Ewing sarcoma (EwS) is a small round blue cell tumor and is the second most frequent pediatric bone cancer. 85% of EwS tumors express the fusion oncoprotein EWS-FLI1, the product of a t(11;22) reciprocal translocation. Prior work has indicated that transcription regulation alone does not fully describe the oncogenic capacity of EWS-FLI1, nor does it provide an effective means to stratify patient tumors. Research using EwS cell lines and patient samples has suggested that EWS-FLI1 also disrupts mRNA biogenesis. In this work we both describe the underlying characteristics of mRNA that are aberrantly spliced in EwS tumor samples as well as catalogue mRNA splicing events across other pediatric tumor types. Here, we also use short- and long-read sequencing to identify cis-factors that contribute to splicing profiles we observe in Ewing sarcoma. Our analysis suggests that GC content upstream of cassette exons is a defining factor of mRNA splicing in EwS. We also describe specific splicing events that discriminate EwS tumor samples from the assumed cell of origin, human mesenchymal stem cells derived from bone marrow (hMSC-BM). Finally, we identify specific splicing factors PCBP2, RBMX, and SRSF9 by motif enrichment and confirm findings from tumor samples in EwS cell lines.

The intronic branch point sequence is under strong evolutionary constraint in the bovine and human genome

The branch point sequence is a cis-acting intronic motif required for mRNA splicing. Despite their functional importance, branch point sequences are not routinely annotated. Here we predict branch point sequences in 179,476 bovine introns and investigate their variability using a catalogue of 29.4 million variants detected in 266 cattle genomes. We localize the bovine branch point within a degenerate heptamer "nnyTrAy". An adenine residue at position 6, that acts as branch point, and a thymine residue at position 4 of the heptamer are more strongly depleted for mutations than coding sequences suggesting extreme purifying selection. We provide evidence that mutations affecting these evolutionarily constrained residues lead to alternative splicing. We confirm evolutionary constraints on branch point sequences using a catalogue of 115 million SNPs established from 3,942 human genomes of the gnomAD database.

Infertility due to defective sperm flagella caused by an intronic deletion in DNAH17 that perturbs splicing

Artificial insemination in pig (Sus scrofa domesticus) breeding involves the evaluation of the semen quality of breeding boars. Ejaculates that fulfill predefined quality requirements are processed, diluted and used for inseminations. Within short time, eight Swiss Large White boars producing immotile sperm that had multiple morphological abnormalities of the sperm flagella were noticed at a semen collection center. The eight boars were inbred on a common ancestor suggesting that the novel sperm flagella defect is a recessive trait. Transmission electron microscopy cross-sections revealed that the immotile sperm had disorganized flagellar axonemes. Haplotype-based association testing involving microarray-derived genotypes at 41,094 SNPs of six affected and 100 fertile boars yielded strong association (P = 4.22 × 10-15) at chromosome 12. Autozygosity mapping enabled us to pinpoint the causal mutation on a 1.11 Mb haplotype located between 3,473,632 and 4,587,759 bp. The haplotype carries an intronic 13-bp deletion (Chr12:3,556,401-3,556,414 bp) that is compatible with recessive inheritance. The 13-bp deletion excises the polypyrimidine tract upstream exon 56 of DNAH17 (XM_021066525.1: c.8510-17_8510-5del) encoding dynein axonemal heavy chain 17. Transcriptome analysis of the testis of two affected boars revealed that the loss of the polypyrimidine tract causes exon skipping which results in the in-frame loss of 89 amino acids from DNAH17. Disruption of DNAH17 impairs the assembly of the flagellar axoneme and manifests in multiple morphological abnormalities of the sperm flagella. Direct gene testing may now be implemented to monitor the defective allele in the Swiss Large White population and prevent the frequent manifestation of a sterilizing sperm tail disorder in breeding boars.

Learning the Regulatory Code of Gene Expression

Data-driven machine learning is the method of choice for predicting molecular phenotypes from nucleotide sequence, modeling gene expression events including protein-DNA binding, chromatin states as well as mRNA and protein levels. Deep neural networks automatically learn informative sequence representations and interpreting them enables us to improve our understanding of the regulatory code governing gene expression. Here, we review the latest developments that apply shallow or deep learning to quantify molecular phenotypes and decode the cis-regulatory grammar from prokaryotic and eukaryotic sequencing data. Our approach is to build from the ground up, first focusing on the initiating protein-DNA interactions, then specific coding and non-coding regions, and finally on advances that combine multiple parts of the gene and mRNA regulatory structures, achieving unprecedented performance. We thus provide a quantitative view of gene expression regulation from nucleotide sequence, concluding with an information-centric overview of the central dogma of molecular biology.

Variant effect on splicing regulatory elements, branchpoint usage, and pseudoexonization: Strategies to enhance bioinformatic prediction using hereditary cancer genes as exemplars

It is possible to estimate the prior probability of pathogenicity for germline disease gene variants based on bioinformatic prediction of variant effect/s. However, routinely used approaches have likely led to the underestimation and underreporting of variants located outside donor and acceptor splice site motifs that affect messenger RNA (mRNA) processing. This review presents information about hereditary cancer gene germline variants, outside native splice sites, with experimentally validated splicing effects. We list 95 exonic variants that impact splicing regulatory elements (SREs) in BRCA1, BRCA2, MLH1, MSH2, MSH6, and PMS2. We utilized a pre-existing large-scale BRCA1 functional data set to map functional SREs, and assess the relative performance of different tools to predict effects of 283 variants on such elements. We also describe rare examples of intronic variants that impact branchpoint (BP) sites and create pseudoexons. We discuss the challenges in predicting variant effect on BP site usage and pseudoexonization, and suggest strategies to improve the bioinformatic prioritization of such variants for experimental validation. Importantly, our review and analysis highlights the value of considering impact of variants outside donor and acceptor motifs on mRNA splicing and disease causation.

Assessment of branch point prediction tools to predict physiological branch points and their alteration by variants

BACKGROUND

Branch points (BPs) map within short motifs upstream of acceptor splice sites (3'ss) and are essential for splicing of pre-mature mRNA. Several BP-dedicated bioinformatics tools, including HSF, SVM-BPfinder, BPP, Branchpointer, LaBranchoR and RNABPS were developed during the last decade. Here, we evaluated their capability to detect the position of BPs, and also to predict the impact on splicing of variants occurring upstream of 3'ss.

RESULTS

We used a large set of constitutive and alternative human 3'ss collected from Ensembl (n = 264,787 3'ss) and from in-house RNAseq experiments (n = 51,986 3'ss). We also gathered an unprecedented collection of functional splicing data for 120 variants (62 unpublished) occurring in BP areas of disease-causing genes. Branchpointer showed the best performance to detect the relevant BPs upstream of constitutive and alternative 3'ss (99.48 and 65.84% accuracies, respectively). For variants occurring in a BP area, BPP emerged as having the best performance to predict effects on mRNA splicing, with an accuracy of 89.17%.

CONCLUSIONS

Our investigations revealed that Branchpointer was optimal to detect BPs upstream of 3'ss, and that BPP was most relevant to predict splicing alteration due to variants in the BP area.

Myc/Max dependent intronic long antisense noncoding RNA, EVA1A-AS, suppresses the expression of Myc/Max dependent anti-proliferating gene EVA1A in a U2 dependent manner

The Myc gene has been implicated in the pathogenesis of most types of human cancerous tumors. Myc/Max activates large numbers of pro-tumor genes; however it also induces anti-proliferation genes. When anti-proliferation genes are activated by Myc, cancer cells can only survive if they are downregulated. Hepatocellular carcinoma (HCC) specific intronic long noncoding antisense (lnc-AS) RNA, the EVA1A-AS gene, is located within the second intron (I2) of the EVA1A gene (EVA-1 homolog A) that encodes an anti-proliferation factor. Indeed, EVA1A, but not EVA1A-AS, is expressed in normal liver. Depletion of EVA1A-AS suppressed cell proliferation of HepG2 cells by upregulation of EVA1A. Overexpression of EVA1A caused cell death at the G2/M phase via microtubule catastrophe. Furthermore, suppressed EVA1A expression levels are negatively correlated with differentiation grade in 365 primary HCCs, while EVA1A-AS expression levels are positively correlated with patient survival. Notably, both EVA1A and EVA1A-AS were activated by the Myc/Max complex. Eva1A-AS is transcribed in the opposite direction near the 3′splice site of EVA1A I2. The second intron did not splice out in a U2 dependent manner and EVA1A mRNA is not exported. Thus, the Myc/Max dependent anti-proliferating gene, EVA1A, is controlled by Myc/Max dependent anti-sense noncoding RNA for HCC survival.

Cooperative evolution of two different TEs results in lineage-specific novel transcripts in the BLOC1S2 gene

BACKGROUND

The BLOC1S2 gene encodes the multifunctional protein BLOS2, a shared subunit of two lysosomal trafficking complexes: i) biogenesis of lysosome-related organelles complex-1 and i) BLOC-1-related complex. In our previous study, we identified an intriguing unreported transcript of the BLOC1S2 gene that has a novel exon derived from two transposable elements (TEs), MIR and AluSp. To investigate the evolutionary footprint and molecular mechanism of action of this transcript, we performed PCR and RT-PCR experiments and sequencing analyses using genomic DNA and RNA samples from humans and various non-human primates.

RESULTS

The results showed that the MIR element had integrated into the genome of our common ancestor, specifically in the BLOC1S2 gene region, before the radiation of all primate lineages and that the AluSp element had integrated into the genome of our common ancestor, fortunately in the middle of the MIR sequences, after the divergence of Old World monkeys and New World monkeys. The combined MIR and AluSp sequences provide a 3' splice site (AG) and 5' splice site (GT), respectively, and generate the Old World monkey-specific transcripts. Moreover, branch point sequences for the intron removal process are provided by the MIR and AluSp combination.

CONCLUSIONS

We show for the first time that sequential integration into the same location and sequence divergence events of two different TEs generated lineage-specific transcripts through sequence collaboration during primate evolution.

Computational Methods for the Pharmacogenetic Interpretation of Next Generation Sequencing Data

Up to half of all patients do not respond to pharmacological treatment as intended. A substantial fraction of these inter-individual differences is due to heritable factors and a growing number of associations between genetic variations and drug response phenotypes have been identified. Importantly, the rapid progress in Next Generation Sequencing technologies in recent years unveiled the true complexity of the genetic landscape in pharmacogenes with tens of thousands of rare genetic variants. As each individual was found to harbor numerous such rare variants they are anticipated to be important contributors to the genetically encoded inter-individual variability in drug effects. The fundamental challenge however is their functional interpretation due to the sheer scale of the problem that renders systematic experimental characterization of these variants currently unfeasible. Here, we review concepts and important progress in the development of computational prediction methods that allow to evaluate the effect of amino acid sequence alterations in drug metabolizing enzymes and transporters. In addition, we discuss recent advances in the interpretation of functional effects of non-coding variants, such as variations in splice sites, regulatory regions and miRNA binding sites. We anticipate that these methodologies will provide a useful toolkit to facilitate the integration of the vast extent of rare genetic variability into drug response predictions in a precision medicine framework.