The cellular concentrations of splicing factors (SFs) are critical for controlling alternative splicing. Most serine and arginine-enriched (SR) protein SFs regulate their own concentration via a homeostatic feedback mechanism that involves regulation of inclusion of non-coding 'poison exons' (PEs) that target transcripts for nonsense-mediated decay. The importance of SR protein PE splicing during animal development is largely unknown despite PE ultra-conservation across animal genomes. To address this, we used mouse genetics to disrupt an ultra-conserved PE in the Tra2b gene encoding the SR protein Tra2β. Focussing on germ cell development, we found that Tra2b PE deletion causes azoospermia due to catastrophic cell death during meiotic prophase. Failure to proceed through meiosis was associated with increased Tra2b expression sufficient to drive aberrant Tra2β protein hyper-responsive splice patterns. Although critical for meiotic prophase, Tra2b PE deletion spared earlier mitotically active germ cells, even though these still required Tra2b gene function. Our data indicate that PE splicing control prevents the accumulation of toxic levels of Tra2β protein that are incompatible with meiotic prophase. This unexpected connection with male fertility helps explain Tra2b PE ultra-conservation and indicates the importance of evaluating PE function in animal models.

Sam68 and SLM2 are paralog RNA binding proteins (RBPs) expressed in the cerebral cortex and display similar splicing activities. However, their relative functions during cortical development are unknown. We found that these RBPs exhibit an opposite expression pattern during development. Sam68 expression declines postnatally while SLM2 increases after birth, and this developmental pattern is reinforced by hierarchical control of Sam68 expression by SLM2. Analysis of Sam68:Slm2 double knockout (Sam68:Slm2dko) mice revealed hundreds of exons that respond to joint depletion of these proteins. Moreover, parallel analysis of single and double knockout cortices indicated that exons regulated mainly by SLM2 are characterized by a dynamic splicing pattern during development, whereas Sam68-dependent exons are spliced at relatively constant rates. Dynamic splicing of SLM2-sensitive exons is completely suppressed in the Sam68:Slm2dko developing cortex. Sam68:Slm2dko mice die perinatally with defects in neurogenesis and in neuronal differentiation, and develop a hydrocephalus, consistent with splicing alterations in genes related to these biological processes. Thus, our study reveals that developmental control of separate Sam68 and Slm2 paralog genes encoding homologous RBPs enables the orchestration of a dynamic splicing program needed for brain development and viability, while ensuring a robust redundant mechanism that supports proper cortical development.

Previously, we showed that the germ cell-specific nuclear protein RBMXL2 represses cryptic splicing patterns during meiosis and is required for male fertility (Ehrmann et al., 2019). Here, we show that in somatic cells the similar yet ubiquitously expressed RBMX protein has similar functions. RBMX regulates a distinct class of exons that exceed the median human exon size. RBMX protein-RNA interactions are enriched within ultra-long exons, particularly within genes involved in genome stability, and repress the selection of cryptic splice sites that would compromise gene function. The RBMX gene is silenced during male meiosis due to sex chromosome inactivation. To test whether RBMXL2 might replace the function of RBMX during meiosis we induced expression of RBMXL2 and the more distantly related RBMY protein in somatic cells, finding each could rescue aberrant patterns of RNA processing caused by RBMX depletion. The C-terminal disordered domain of RBMXL2 is sufficient to rescue proper splicing control after RBMX depletion. Our data indicate that RBMX and RBMXL2 have parallel roles in somatic tissues and the germline that must have been conserved for at least 200 million years of mammalian evolution. We propose RBMX family proteins are particularly important for the splicing inclusion of some ultra-long exons with increased intrinsic susceptibility to cryptic splice site selection.

High levels of transcription and alternative splicing are recognized hallmarks of gene expression in the testis and largely driven by cells in meiosis. Because of this, the male meiosis stage of the cell cycle is often viewed as having a relatively permissive environment for gene expression. In this review, we highlight recent findings that identify the RNA binding protein RBMXL2 as essential for male meiosis. RBMXL2 functions as a "guardian of the transcriptome" that protects against the use of aberrant (or "cryptic") splice sites that would disrupt gene expression. This newly discovered protective role during meiosis links with a wider field investigating mechanisms of cryptic splicing control that protect neurons from amyotrophic lateral sclerosis and Alzheimer's disease. We discuss how the mechanism repressing cryptic splicing patterns during meiosis evolved, and why it may be essential for sperm production and male fertility.

Background: Androgen steroid hormones are key drivers of prostate cancer. Previous work has shown that androgens can drive the expression of alternative mRNA isoforms as well as transcriptional changes in prostate cancer cells. Yet to what extent androgens control alternative mRNA isoforms and how these are expressed and differentially regulated in prostate tumours is unknown. Methods: Here we have used RNA-Seq data to globally identify alternative mRNA isoform expression under androgen control in prostate cancer cells, and profiled the expression of these mRNA isoforms in clinical tissue. Results: Our data indicate androgens primarily switch mRNA isoforms through alternative promoter selection. We detected 73 androgen regulated alternative transcription events, including utilisation of 56 androgen-dependent alternative promoters, 13 androgen-regulated alternative splicing events, and selection of 4 androgen-regulated alternative 3′ mRNA ends. 64 of these events are novel to this study, and 26 involve previously unannotated isoforms. We validated androgen dependent regulation of 17 alternative isoforms by quantitative PCR in an independent sample set. Some of the identified mRNA isoforms are in genes already implicated in prostate cancer (including LIG4, FDFT1 and RELAXIN), or in genes important in other cancers (e.g. NUP93 and MAT2A). Importantly, analysis of transcriptome data from 497 tumour samples in the TGCA prostate adenocarcinoma (PRAD) cohort identified 13 mRNA isoforms Open Peer Review Reviewer Status Invited Reviewers 1 2 3 version 1 03 Aug 2018 report report report Sebastian Oltean, University of Exeter, Exeter, UK 1. Cyril F. Bourgeois , University of Lyon, Lyon, France 2. Jennifer Byrne , The Children's Hospital at Westmead, Westmead, Australia 3. Any reports and responses or comments on the article can be found at the end of the article. Page 1 of 35 F1000Research 2018, 7:1189 Last updated: 21 AUG 2021

Previously we showed that the germline-specific RNA binding protein RBMXL2 is essential for male meiosis where it represses cryptic splicing patterns (1). Here we find that its ubiquitously expressed paralog RBMX helps underpin human genome stability by preventing non-productive splicing. In particular, RBMX blocks selection of aberrant splice and polyadenylation sites within some ultra-long exons that would interfere with genes needed for normal replication fork activity. Target exons include within the ETAA1 ( Ewings Tumour Associated 1 ) gene, where RBMX collaborates with its interaction partner Tra2β to enable full-length exon inclusion by blocking selection of an aberrant 3’ splice site. Our data reveal a novel group of RNA processing targets potently repressed by RBMX, and help explain why RBMX is associated with gene expression networks in cancer, replication and sensitivity to genotoxic drugs.

RBMX is a ubiquitously expressed nuclear RNA binding protein that is encoded by a gene on the X chromosome. RBMX belongs to a small protein family with additional members encoded by paralogs on the mammalian Y chromosome and other chromosomes. These RNA binding proteins are important for normal development, and also implicated in cancer and viral infection. At the molecular level RBMX family proteins contribute to splicing control, transcription and genome integrity. Establishing what endogenous genes and pathways are controlled by RBMX and its paralogs will have important implications for understanding chromosome biology, DNA repair and mammalian development. Here we review what is known about this family of RNA binding proteins, and identify important current questions about their functions.

Male germ cells of all placental mammals express an ancient nuclear RNA binding protein of unknown function called RBMXL2. Here we find that deletion of the retrogene encoding RBMXL2 blocks spermatogenesis. Transcriptome analyses of age-matched deletion mice show that RBMXL2 controls splicing patterns during meiosis. In particular, RBMXL2 represses the selection of aberrant splice sites and the insertion of cryptic and premature terminal exons. Our data suggest a Rbmxl2 retrogene has been conserved across mammals as part of a splicing control mechanism that is fundamentally important to germ cell biology. We propose that this mechanism is essential to meiosis because it buffers the high ambient concentrations of splicing activators, thereby preventing poisoning of key transcripts and disruption to gene expression by aberrant splice site selection.

Prostate is the most frequent cancer in men. Prostate cancer progression is driven by androgen steroid hormones, and delayed by androgen deprivation therapy (ADT). Androgens control transcription by stimulating androgen receptor (AR) activity, yet also control pre-mRNA splicing through less clear mechanisms. Here we find androgens regulate splicing through AR-mediated transcriptional control of the epithelial-specific splicing regulator ESRP2. Both ESRP2 and its close paralog ESRP1 are highly expressed in primary prostate cancer. Androgen stimulation induces splicing switches in many endogenous ESRP2-controlled mRNA isoforms, including splicing switches correlating with disease progression. ESRP2 expression in clinical prostate cancer is repressed by ADT, which may thus inadvertently dampen epithelial splice programmes. Supporting this, treatment with the AR antagonist bicalutamide (Casodex) induced mesenchymal splicing patterns of genes including FLNB and CTNND1. Our data reveals a new mechanism of splicing control in prostate cancer with important implications for disease progression.

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Male germ cells of all placental mammals express an ancient nuclear RNA binding protein of unknown function called RBMXL2. Here we find that deletion of the retrogene encoding RBMXL2 blocks spermatogenesis. Transcriptome analyses of age-matched deletion mice show that RBMXL2 controls splicing patterns during meiosis. In particular, RBMXL2 represses the selection of aberrant splice sites and the insertion of cryptic and premature terminal exons. Our data suggest a Rbmxl2 retrogene has been conserved across mammals as part of a splicing control mechanism that is fundamentally important to germ cell biology. We propose that this mechanism is essential to meiosis because it buffers the high ambient concentrations of splicing activators, thereby preventing poisoning of key transcripts and disruption to gene expression by aberrant splice site selection. https://doi.org/10.7554/eLife.39304.001 eLife digest In humans and other mammals, a sperm from a male fuses with an egg cell from a female to produce an embryo that may ultimately grow into a new individual. Sperm and egg cells are made when certain cells in the body divide in a process called meiosis. Many proteins are required for meiosis to happen and these proteins are made using instructions provided by genes, which are made of a molecule called DNA. The DNA within a gene is transcribed to make molecules of ribonucleic acid (or RNA for short). The cell then modifies many of these RNAs in a process called splicing before using them as templates to make proteins. During splicing, segments of RNA known as introns are discarded and other segments termed exons are joined together. Some exons may also be removed from RNAs in different combinations to create different proteins from the same gene. A protein called RBMXL2 is able to bind to RNA molecules and is only made during and after meiosis in humans and most other mammals. RBMXL2 can also bind to other proteins that are known to be involved in controlling splicing of RNAs, but its role in splicing remains unclear. To address this question, Ehrmann et al. studied the gene that encodes the RBMXL2 protein in mice. Removing this gene prevented male mice from being able to make sperm. Further experiments using a technique called RNA sequencing showed that the RBMXL2 protein helps to ensure that splicing happens correctly by preventing bits of exons and introns in mouse genes from being rearranged. These findings suggest that the gene encoding RBMXL2 is part of a splicing control mechanism that is important for making sperm and egg cells. The work of Ehrmann et al. could eventually help some couples understand why they have problems conceiving children. Male infertility is poorly understood, and not knowing its causes can harm the mental health of affected men. Furthermore, these findings may help researchers to understand the role of a closely related protein called RBMY that has also been linked to infertility in men, but is much more difficult to study. https://doi.org/10.7554/eLife.39304.002 Introduction It has been a long recognised hallmark of mammalian gene expression patterns that there are extremely high levels of transcription and transcriptome complexity in testicular cells (de la Grange et al., 2010; Licatalosi, 2016; Soumillon et al., 2013; Pan et al., 2008; Wang et al., 2008; Clark et al., 2007; Grosso et al., 2008; Yeo et al., 2004). These high gene expression levels are thought to result from epigenetic changes that favour relaxed patterns of gene expression during meiosis – the unique form of division used to generate sperm and eggs (Soumillon et al., 2013). Many nuclear RNA-binding proteins are differentially expressed during and immediately after meiosis (Grellscheid et al., 2011; Schmid et al., 2013). These include the nuclear RNA binding protein RBMXL2 (also known as hnRNP GT) that is expressed only during and immediately after male meiosis, but not in the preceding spermatogonial cells (Ehrmann et al., 2008) (Figure 1A). Consistent with an important function in germ cell development, genetic studies have identified point mutations within infertile men in the human chromosome 11 RBMXL2 gene (Westerveld et al., 2004). A further connection to male infertility is that RBMXL2 belongs to the same gene family as RBMY, which was historically the first human Y chromosome infertility gene identified in the search for the AZF (AZOOSPERMIA FACTOR) gene (Ma et al., 1993). Figure 1 with 2 supplements see all Download asset Open asset Creation of a mouse model which does not express the testis-specific RNA binding protein RBMXL2. (A) The RBMXL2 gene evolved via retrotransposition of RBMX early in mammalian evolution. The cladogram shows three members of the gene family that derived from an ancestral RBMX gene, summarising their global functions, expression, and RNA binding sites of their encoded proteins. (B) Modular structure of the RBMXL2 protein, showing the known protein interaction with Tra2β. (C) Western blot confirming no RBMXL2 protein is made within 2 replicate samples of Rbmxl2-/- adult testis, compared to wild type (Rbmxl2+/+) and heterozygous (Rbmxl2+/-) adult testes. https://doi.org/10.7554/eLife.39304.003 RBMXL2 evolved ~65 million years ago via retrotransposition of an mRNA from the X chromosome located RBMX gene (Figure 1A). An RBMXL2 gene is found in all placental mammals, which is consistent with a fundamental role in germ cell biology. This role remains to be identified, but RBMXL2 protein has an N-terminal RNA Recognition Motif (abbreviated RRM, Figure 1B). RBMX and RBMY proteins are also nuclear RNA binding proteins that are very similar to RBMXL2 (73.2% and 36.8% overall identity to RBMX and RBMY, respectively; 93.7% and 77.2% identity within the RRMs, Figure 1—figure supplement 1). The RNA binding specificity of RBMXL2 protein is unknown, but both RBMX and RBMY proteins bind to AA dinucleotide-containing RNA sequences (Cléry et al., 2011; Moursy et al., 2014; Nasim et al., 2003). The RBMXL2, RBMX and RBMY proteins interact with and modulate the splicing activity of Tra2β and SR proteins in vitro (Figure 1B) (Cléry et al., 2011; Moursy et al., 2014; Liu et al., 2009; Nasim et al., 2003; Elliott et al., 2000a), suggesting a role in splicing control. Maintaining proper ratios of mRNA splice isoforms can be critical in normal development (Kalsotra and Cooper, 2011), where changes in isoforms can have effects on encoded proteins ranging from major to subtle. Alternative splicing is known to be critical for germ cell development. For example, deletion of the splicing regulator protein PTBP2 within germ cells affects mRNA isoforms important for cell-cell communication with Sertoli cells (Hannigan et al., 2017). Some alternative splicing events in the testis are conserved between humans and mice so may control fundamental aspects of germ cell biology (Schmid et al., 2013). However many alternative splicing patterns are not conserved between humans and mice (Kan et al., 2005). RBMX protein is also reported to control transcription (Takemoto et al., 2007), affect DNA double strand repair and mitotic sister chromatid cohesion (Adamson et al., 2012; Matsunaga et al., 2012), and to bind to m6A methylated RNA (Liu et al., 2017). The function of RBMXL2 and why this RNA binding protein has been conserved across placental mammals are not known. A major factor limiting understanding of endogenous RBMXL2 functions has been the absence of a reliable mouse model. Development of a mouse model is also critical to test the importance of this wider family of RNA binding proteins in germ cell development. Men carrying the AZFb deletion on their Y chromosomes are missing RBMY genes and undergo meiotic arrest. However, it is unclear if RBMY loss is causing this phenotype, because the deletion interval encompasses several other genes which could contribute to male infertility (Elliott, 2000; Vogt et al., 1996). Within the meiotic and immediately post-meiotic germ cells which express RBMXL2, the RBMX and RBMY genes are transcriptionally inactivated within a heterochromatic structure called the XY body (Wang, 2004). Meiosis thus provides a genetically tractable window to probe RBMXL2 function where there should be no redundancy effects possible with either RBMX or RBMY. Hence to discover what RBMXL2 does in the germline we have made a conditional Rbmxl2 gene knockout mouse. Analysis of this knockout mouse reveals that RBMXL2 protein is essential for meiosis and has a major role in protecting the meiotic transcriptome from aberrant selection of cryptic splice sites that are normally ignored by the spliceosome. Our data suggest this fundamentally important process operates so efficiently in meiosis that it has been previously undetected, yet is critical to avoid male infertility caused by aberrant splicing of key meiotic transcripts. Results RBMXL2 protein is essential for male fertility To test how important RBMXL2 protein is for male germline development we made a conditional mouse model in which we flanked the Rbmxl2 open reading frame with LoxP sites. Since Rbmxl2 is only expressed in the testis (Elliott et al., 2000b), we chose to delete the entire Rbmxl2 open reading frame to create a null allele. We achieved this by crossing our conditional model with a mouse strain expressing Cre recombinase under control of the ubiquitous Pgk promoter (experimental details are provided in the Materials and methods). We confirmed deletion of this genomic region in homozygous Rbmxl2 gene knockout (Rbmxl2-/-) mice by Southern blotting (Figure 1—figure supplement 2) and the specific absence of the 50 KDa RBMXL2 protein from knockout testes by Western blotting (Figure 1C). Rbmxl2-/- mice developed comparably to their wild type littermates but their testes were much smaller (Figure 2A and B). This small testis phenotype correlated with a severe disruption of testicular histology. Adult Rbmxl2-/- mice contained cells undergoing meiosis but almost no post-meiotic cells (Figure 2C, meiotic spermatocytes are abbreviated Spc, and post-meiotic round spermatids are abbreviated Rtd). The epididymis dissected from Rbmxl2-/- mice were completely devoid of sperm. Four wild type mice tested had an average of 7.07 ± 1.39×106 epididymal sperm ml−1, compared with zero in four Rbmxl2-/- mice (Figure 2D and E). No effect on female fertility was observed in Rbmxl2-/- mice (not shown). Figure 2 with 4 supplements see all Download asset Open asset RBMXL2 protein is important for mouse germ cells to progress past diplotene into metaphase I of meiosis. (A) Testis morphologies of adult wild type and homozygous knockout (Rbmxl2-/-) mice (scale above testes shown in millimeters). (B) Testis:body weight ratios of different genotype mice. P values were calculated using a t test. (C) Micrographs of haemotoxylin-stained testis sections from wild type and Rbmxl2-/- mice. Abbreviations: Spg, spermatogonia; Spc, spermatocyte; Rtd, round spermatid; Spd, elongating spermatid. Scale bar = 20 um. (D) Epididymis of wild type and Rbmxl2-/- mice stained with haemotoxylin and eosin. Asterisk indicates lumen (note lumen is empty in Rbmxl2-/- section indicative of azoospermia, but contains multiple cells in the wild type). Scale bar = 20 um. (E) Epididymal sperm counts of wild type and Rbmxl2-/- mice (n = 4 of each genotype). The error bar represents SEM. (F) Meiotic prophase I stages detected in wild type and Rbmxl2-/- testis chromosome spreads stained for SYCP3 (pseudocoloured red) and SYCP1 (pseudocoloured green); and DAPI and SYCP3 (pseudocoloured blue and red respectively). No metaphase I nuclei were identified in the Rbmxl2-/- testes. Quantitative data are provided in Figure 2—figure supplement 2. https://doi.org/10.7554/eLife.39304.006 Very rarely a few round spermatids were observed in adult Rbmxl2-/- testis sections (Figure 2—figure supplement 1A), although no elongated spermatids were detected. The presence of these round spermatids indicates that meiosis can occasionally complete in the absence of RBMXL2 protein. This is consistent with loss of RBMXL2 causing either (1) a developmental block in meiosis from which a few cells can escape; or alternatively (2) a slow attrition effect, in which the Rbmxl2-/- testis phenotype is caused by post-meiotic stages dying in the adult testis. To differentiate between these two possibilities we histologically analysed testes at 21 days postpartum (21dpp), at which point germ cells in the wild type (Rbmxl2 +/+) mice have just started to enter the post-meiotic round spermatid stage. At 21dpp there were still significantly fewer round spermatids in the Rbmxl2-/- sections compared to wild type (Figure 2—figure supplement 1B), even though there would have been less time for round spermatid cell death compared with the adult. This result thus supports a strong meiotic block in Rbmxl2-/- mice, with occasional completion of meiosis rather than a gradual attrition of round spermatids. Most Rbmxl2-/- germ cells do not progress past meiotic diplotene The above data demonstrate that an Rbmxl2 gene, which is conserved in all placental mammals and specifically expressed in male meiosis, is essential for mouse spermatogenesis somewhere within meiotic prophase. Staining of nuclear spreads with antibodies specific to the meiotic chromosome proteins SYCP1 and SYCP3 more precisely showed that Rbmxl2-/- mice arrest germ cell development during the diplotene substage of meiotic prophase (Figure 2F and Figure 2—figure supplement 2). Metaphase I nuclei were only detected in wild type mice and not in Rbmxl2-/- mice (N = 3 wild type and N = 3 Rbmxl2-/-testes, 180 and 138 spermatocyte nuclei scored for wild type and Rbmxl2-/- respectively). No significant differences in the frequency of earlier stages of meiotic prophase were detected between Rbmxl2-/- and wild type testes. The presence of germ cell populations up to diplotene in Rbmxl2-/- testes was further confirmed by analysis of histological sections stained with Periodic Acid Schiff (PAS) (Figure 2—figure supplement 3A–D). Staining of adult testis sections with PAS or antibodies specific to the apoptotic marker activated Caspase three further showed that adult Rbmxl2-/- germ cells die via apoptosis (Figure 2—figure supplement 3E–F). Analysis of diplotene nuclear spreads revealed further abnormalities in the Rbmxl2-/- testis. There was a decreased number of H3K9me3-marked centromere clusters (Takada et al., 2011), with each individual cluster also containing more centromeres than in wild type testis (Figure 2—figure supplement 4A–C). H3K9me3 staining was also present at the sex body in a proportion (82%) of wild type diplotene spermatocytes, but essentially absent in mutant diplotene spermatocytes (1.5% of diplotene spermatocytes stained, 66 Rbmxl2-/- and 66 wild type nuclei scored, n = 3 for both) (Figure 2—figure supplement 4D). These defects in centromere clustering and H3K9me3 modification of the sex body were still detectable but less severe at pachytene (Figure 2—figure supplement 4D). Asynapsis was rarely observed in either control or mutant pachytene stage spermatocytes, indicating that the defect causing diplotene arrest does not significantly impact either synaptonemal complex formation or meiotic homology searching (Figure 2—figure supplement 4E) (N = 3, 109 and 94 nuclei scored for wild type and Rbmxl2-/- respectively). In summary, the above mouse phenotype showed that deletion of the ancient RNA binding protein RBMXL2 induces progressive defects in male mouse meiotic prophase. These defects culminate with a major block during diplotene that prevents entry into metaphase I, but with defects becoming already apparent during pachytene. Gene expression analysis of age matched wild type and Rbmxl2-/- testes Since RBMXL2 is a nuclear RNA binding protein we predicted that the above phenotype could be associated with a primary molecular defect in generating or processing RNAs in the Rbmxl2-/- testis. To test this we analysed wild type and Rbmxl2-/- testes using RNAseq. Based on the knockout phenotype above, and because the adult wild type testis contains additional more advanced germ cells that are missing from the adult Rbmxl2-/- testis, we analysed testes at 18 days post partum (18dpp) during the first synchronised wave of mouse spermatogenesis. Such 18dpp wild type mouse testes contain germ cells between spermatogonia all the way through to diplotene, with 60% of cells engaged in meiotic prophase, but no post-meiotic cells (Bellvé et al., 1977). Gene expression analysis of this RNAseq data (Anders and Huber, 2010) showed overall patterns of transcription were similar between wild type and Rbmxl2-/- 18dpp testes (Figure 3A, and Figure 3—source data 1). Only 45 genes showed a fold change greater or equal to two between the wild type and Rbmxl2-/- testes, with an adjusted p value of less than 0.05, and only 23 of these changes were for known protein coding genes (Figure 3—source data 1). The strongest difference in overall gene expression between the wild type and Rbmxl2-/- backgrounds was for the Rbmxl2-/- gene itself, as expected since the Rbmxl2-/- gene is deleted from the knockout mouse. We also detected strong expression changes within the Fsip2 gene, particularly for Fsip2 exon 16 and downstream exons that were expressed only in the wild type background (Figure 3—figure supplement 1A and B). Mutations in Fsip2 correlate with defects in human sperm flagella – cellular structures which develop after meiosis (Martinez et al., 2018). RNAseq analysis detected more subtle expression changes in Cul4a, Slc9c1 and Tex15, each of which are required for mouse male fertility (Smith et al., 2018) (Figure 3—source data 1) (Kopanja et al., 2011). Genes encoding transcription factors (Myf6 and Nxn) and a signalling protein (Cyr61) that controls apoptosis (Jun and Lau, 2011) also changed expression. Mouse phenotype information at the Mouse Genome Database (MGD) (Smith et al., 2018) indicate that each of these latter three genes have important roles in normal development but not specifically of the testis (Figure 3—source data 2). Sixteen (36% of the total detected) gene expression changes in the Rbmxl2-/- testis were for predicted non coding RNAs of unknown function (Figure 3—source data 1). Figure 3 with 3 supplements see all Download asset Open asset RBMXL2 protein expression controls splicing patterns of important genes during meiosis. (A) MAplot showing gene expression levels in 18dpp testis transcriptomes and how they change between wild type and Rbmxl2-/- mice. Genes with more than a 2-fold change in gene expression, and an adjusted p value of less than 0.05 are shown as red dots. All other genes are shown as black dots. This data given in full within Figure 3—source data 1. (B) Scatterplot showing splicing changes between the wild-type and Rbmxl2-/- testis detected by RNA-seq. The scatterplot shows expected percentage of exonic segment inclusion, or E(PSI) for wild-type and Rbmxl2-/- testis detected by RNA-seq using the bioinformatics programme MAJIQ (Vaquero-Garcia et al., 2016). Splicing changes in some genes are named and arrowed. High-confidence splicing changes (P(|ΔPSI| > V)>95%) are marked in dark red for a predicted change of V = 20%, and dark yellow for a predicted change of V = 10%, and are also listed in Figure 3—source data 2 All other quantified splice changes are in yellow (21,280 events examined). Dashed lines indicate ΔPSI of +/- 10%. (C) Left pie chart: Proportions of genes identified by RNAseq analysis to have defects at the splicing and transcriptional levels. Right pie chart: Proportions of phenotypes reported by whole gene knockout annotated for genes detected to have splicing defects in the absence of RBMXL2 protein (Smith et al., 2018) (see also Figure 3—source data 6). (D) The top 10 most frequently recovered pentamers after HITS-CLIP for RBMXL2 in the adult mouse testis (AA dinucleotides are shown in red). https://doi.org/10.7554/eLife.39304.011 Figure 3—source data 1 Complete list of gene expression changes between the 18dpp testes of wild type and Rbmxl2-/- testes detected by DESeq. https://doi.org/10.7554/eLife.39304.015 Download elife-39304-fig3-data1-v1.xlsx Figure 3—source data 2 MGI mouse phenotypes reported by the Mouse Genome Database (Smith et al., 2018) after whole gene knockout for RBMXL2 transcription targets. https://doi.org/10.7554/eLife.39304.016 Download elife-39304-fig3-data2-v1.xlsx Figure 3—source data 3 Complete list of 237 high-confidence, mis-regulated local splicing variations in 186 genes between wild type and Rbmxl2-/- testes detected by Majiq (Vaquero-Garcia et al., 2016). https://doi.org/10.7554/eLife.39304.017 Download elife-39304-fig3-data3-v1.xlsx Figure 3—source data 4 PCR primers used to analyse splicing events by RT-PCR that change between wild type and Rbmxl2-/- testes. Some genes have more than one regulated event. The coloured cells represent events in a gene also occurring in duplicate on the list. https://doi.org/10.7554/eLife.39304.018 Download elife-39304-fig3-data4-v1.xlsx Figure 3—source data 5 Gene Ontology (GO) terms for RBMXL2 splicing targets identified. GO terms were obtained from the Mouse Genome Database (Smith et al., 2018). https://doi.org/10.7554/eLife.39304.019 Download elife-39304-fig3-data5-v1.xlsx Figure 3—source data 6 MGI mouse phenotypes reported after whole gene knockout for RBMXL2 splicing targets. Phenotypes were obtained from the Mouse Genome Database (Smith et al., 2018). https://doi.org/10.7554/eLife.39304.020 Download elife-39304-fig3-data6-v1.xlsx Figure 3—source data 7 List of RBMXL2 target genes that are already implicated in mouse development. Summary of GO terms and MGI phenotypes of genes that are regulated at the splicing level by RBMXL2 (compiled from Figure 3—source datas 5 and 6). https://doi.org/10.7554/eLife.39304.021 Download elife-39304-fig3-data7-v1.docx RBMXL2 protein controls splicing patterns during meiosis We carried out further bioinformatic analysis to specifically search for mis-regulated splicing events in the Rbmxl2-/- testes (Vaquero-Garcia et al., 2016). A total of 237 high-confidence, mis-regulated local splicing variations were identified in 186 genes (Figure 3B, and Figure 3—source data 3). Using RT-PCR to distinguish splice isoforms 27 of these splicing changes were experimentally tested, validating 23/27 splice isoform switches (Figure 3—source data 4). Some genes had more than one splicing event controlled by RBMXL2 (e.g. the Catsperb gene had three events). Gene Ontology (GO) analysis showed that a number of the genes regulated at the splicing level by RBMXL2 have established roles in spermatogenesis, meiosis and germ cell development (Figure 3—source data 5). However, amongst the complete set of regulated genes there was no significant enrichment of particular GO terms. This is consistent with RBMXL2 regulating splicing of a functionally diverse group of genes. Analysis of knockout phenotypes provided by the MGD (Smith et al., 2018) indicated that whole gene deletion of 25/186 RBMXL2-regulated genes cause male infertility. These latter target genes must thus have a key role in germ cell development (Figure 3C, Figure 3—source data 6, Figure 3—source data 7). Genetic deletion of some other RBMXL2 target genes cause either embryonic or neonatal lethality (33/186 genes), developmental (19/36 genes) or pleiotropic defects (43/186 genes). Cell type mis-splicing of these latter genes during meiosis could thus cause severe phenotypic effects on germ cell biology. Fourteen (33%) of the genes originally identified to have changed overall expression levels between wild type and Rbmxl2-/- testes also changed splicing patterns (Figure 3C, and Figure 3—source datas 1 and 3). These included splice variants in Esco1 (encoding a protein involved in chromatid cohesion), Slc39a8 (that encodes the transporter protein responsible for cadmium toxicity in the testis) (Dalton et al., 2005); and the Slc9c1 gene (annotated on the MGD as essential for male fertility [Smith et al., 2018]). The RNA binding specificity of RBMXL2 protein was unknown. Thus we used high throughput sequencing cross linking immunoprecipitation (HITS-CLIP) to enable us to correlate splicing changes detected within the Rbmxl2-/- 18dpp mouse testis with global RBMXL2 protein-RNA interactions (Grellscheid et al., 2011). Antibodies specific to mouse RBMXL2 immuno-precipitated a radiolabelled RNA protein adduct of the known size of RBMXL2 protein (50 KDa) after treatment with high concentrations of RNase (Figure 3—figure supplement 2A) (Elliott et al., 2000b). Lower concentrations of RNase were used to retrieve an average tag length of 40 nucleotides. Enriched motif analysis of the sequenced RBMXL2 CLIP tags showed that each of the top 10 5-mers contained the dinucleotide AA (Figure 3D). Interestingly, AA is also the dinucleotide bound by RBMX (Moursy et al., 2014), which is consistent with over 90% shared sequence identity within the RRMs of these proteins (Figure 1—figure supplement 1). These intragenic cross-linked sites mapped to the mouse genome most frequently within introns, consistent with RBMXL2 being involved in nuclear RNA-processing events (Figure 3—figure supplement 2B). Next we searched for CLIP tags mapping to regions near the splicing events controlled by RBMXL2, which would suggest direct regulation. We observed enrichment of RBMXL2 binding both within alternative exon sequences and in regions proximal to regulated splice sites compared to non-regulated events (Figure 3—figure supplement 3A). Consistent with these enriched binding occurrences, motif maps of the top pentamers identified by CLIP (Figure 3D) showed regions of enrichment proximal to regulated versus non-regulated splice junctions (Figure 3—figure supplement 3B). RBMXL2 protein represses splicing of exons that would compromise meiotic gene expression Overall, the above bioinformatics analysis indicated an accumulation of defective splice isoform patterns in the Rbmxl2-/- testis. MAJIQ captures local splicing variations (LSVs) that involve both known and un-annotated (de-novo) splice sites, junctions and exons. These LSVs can correspond to both classical binary splicing events and more complex events involving three or more junctions. Since classical binary events (e.g. skipped exons) are easier to inspect and visualise with the RNAseq reads on the UCSC genome browser, we focused on the binary events for further investigation. This showed that 60% of the 87 most easily visualised classical splicing events controlled by RBMXL2 involve the altered selection of de-novo splice sites (defined as not currently annotated on the most recent build of the Mus musculus genome GRCm38/mm10, Figure 3—source data 3). These de-novo splicing variations controlled by RBMXL2 are putatively cryptic events as they insert novel internal exons that were flanked by consensus GT-AG 5' and 3' splice sites. Moreover, these de-novo splice sites showed low expected percent spliced in (E(PSI)) values in wild type 18 dpp testes and across a panel of twelve mouse tissues, further suggesting they are cryptic events only included in the absence of RBMXL2 and not in wild type mice (Figure 4—figure supplement 1). Detailed investigation indicated 84% of such cryptic exons were either not multiples of three or introduced stop codons into their mRNAs, meaning their insertion into mRNAs would disrupt protein reading frames and interfere with meiotic protein expression. Splicing inclusion of cryptic terminal exons could also severely impact patterns of meiotic gene expression. We detected high inclusion of a cryptic terminal exon within the 5′ UTR of the Kdm4d gene (Figure 4A and B, note also increased splicing of an already annotated upstream Kdm4d alternative exon within the Rbmxl2-/- testis). We confirmed splicing inclusion of this Kdm4d cryptic exon in the Rbmxl2-/- testes using RT-PCR (primer positions are shown in Figure 4A). Kdm4d encodes a histone demethylase protein that is important for normal patterns of germ cell apoptosis in the testis (Iwamori et al., 2011). Exon two contains the entire CDS (Coding DNA Sequence, Figure 4A) of the Kdm4d gene. Analysis of exon junction read numbers using Sashimi plots confirmed different nuclear processing pathways are used for Kdm4d in wild type and knockout testes (Figure 4—figure supplement 2A). There were 14-fold more exon junction reads connecting Kdm4d exon one to the cryptic exon (140 reads), when compared to exon junction reads joining the cryptic exon to Kdm4d coding exon 2. This pattern is consistent with splicing inclusion of the cryptic exon being connected with use of an associated polyA site. Consistent with this, some individual RNAseq reads extended past the 5′ splice site of the Kdm4d cryptic exon and then terminated downstream following a consensus polyadenylation site (AATAAA) sequence (Figure 4C). Both bioinformatics analysis (Figure 3—source data 1) a

Background: Androgen steroid hormones are key drivers of prostate cancer. Previous work has shown that androgens can drive the expression of alternative mRNA isoforms as well as transcriptional changes in prostate cancer cells. Yet to what extent androgens control alternative mRNA isoforms and how these are expressed and differentially regulated in prostate tumours is unknown. Methods: Here we have used RNA-Seq data to globally identify alternative mRNA isoform expression under androgen control in prostate cancer cells, and profiled the expression of these mRNA isoforms in clinical tissue. Results: Our data indicate androgens primarily switch mRNA isoforms through alternative promoter selection. We detected 73 androgen regulated alternative transcription events, including utilisation of 56 androgen-dependent alternative promoters, 13 androgen-regulated alternative splicing events, and selection of 4 androgen-regulated alternative 3′ mRNA ends. 64 of these events are novel to this study, and 26 involve previously unannotated isoforms. We validated androgen dependent regulation of 17 alternative isoforms by quantitative PCR in an independent sample set. Some of the identified mRNA isoforms are in genes already implicated in prostate cancer (including LIG4, FDFT1 and RELAXIN), or in genes important in other cancers (e.g. NUP93 and MAT2A). Importantly, analysis of transcriptome data from 497 tumour samples in the TGCA prostate adenocarcinoma (PRAD) cohort identified 13 mRNA isoforms (including TPD52, TACC2 and NDUFV3) that are differentially regulated in localised prostate cancer relative to normal tissue, and 3 (OSBPL1A, CLK3 and TSC22D3) which change significantly with Gleason grade and tumour stage. Conclusions: Our findings dramatically increase the number of known androgen regulated isoforms in prostate cancer, and indicate a highly complex response to androgens in prostate cancer cells that could be clinically important.

Background: Circular RNAs (circRNAs) are predominantly derived from protein coding genes, and some can act as microRNA sponges or transcriptional regulators. Changes in circRNA levels have been identified during human development which may be functionally important, but lineage-specific analyses are currently lacking. To address this, we performed RNAseq analysis of human embryonic stem (ES) cells differentiated for 90 days towards 3D laminated retina. Results: A transcriptome-wide increase in circRNA expression, size, and exon count was observed, with circRNA levels reaching a plateau by day 45. Parallel statistical analyses, controlling for sample and locus specific effects, identified 239 circRNAs with expression changes distinct from the transcriptome-wide pattern, but these all also increased in abundance over time. Surprisingly, circRNAs derived from long non-coding RNAs (lncRNAs) were found to account for a significantly larger proportion of transcripts from their loci of origin than circRNAs from coding genes. The most abundant, circRMST:E12-E6, showed a > 100X increase during differentiation accompanied by an isoform switch, and accounts for > 99% of RMST transcripts in many adult tissues. The second most abundant, circFIRRE:E10-E5, accounts for > 98% of FIRRE transcripts in differentiating human ES cells, and is one of 39 FIRRE circRNAs, many of which include multiple unannotated exons. Conclusions: Our results suggest that during human ES cell differentiation, changes in circRNA levels are primarily globally controlled. They also suggest that RMST and FIRRE, genes with established roles in neurogenesis and topological organisation of chromosomal domains respectively, are processed as circular lncRNAs with only minor linear species.

Cell migration drives cell invasion and metastatic progression in prostate cancer and is a major cause of mortality and morbidity. However the mechanisms driving cell migration in prostate cancer patients are not fully understood. We previously identified the cancer-associated cell migration protein Tetraspanin 1 (TSPAN1) as a clinically relevant androgen regulated target in prostate cancer. Here we find that TSPAN1 is acutely induced by androgens, and is significantly upregulated in prostate cancer relative to both normal prostate tissue and benign prostate hyperplasia (BPH). We also show for the first time, that TSPAN1 expression in prostate cancer cells controls the expression of key proteins involved in cell migration. Stable upregulation of TSPAN1 in both DU145 and PC3 cells significantly increased cell migration and induced the expression of the mesenchymal markers SLUG and ARF6. Our data suggest TSPAN1 is an androgen-driven contributor to cell survival and motility in prostate cancer.

SLM2 and Sam68 are splicing regulator paralogs that usually overlap in function, yet only SLM2 and not Sam68 controls the Neurexin2 AS4 exon important for brain function. Herein we find that SLM2 and Sam68 similarly bind to Neurexin2 premRNA, both within the mouse cortex and in vitro. Protein domainswap experiments identify a region including the STAR domain that differentiates SLM2 and Sam68 activity in splicing target selection, and confirm that this is not established via the variant amino acids involved in RNA contact. However, far fewer SLM2 and Sam68 RNA binding sites flank the Neurexin2 AS4 exon, compared with those flanking the Neurexin1 and Neurexin3 AS4 exons under joint control by both Sam68 and SLM2. Doubling binding site numbers switched paralog sensitivity, by placing the Neurexin2 AS4 exon under joint splicing control by both Sam68 and SLM2. Our data support a model where the density of shared RNA binding sites around a target exon, rather than different paralog-specific proteinRNA binding sites, controls functional target specificity between SLM2 and Sam68 on the Neurexin2 AS4 exon. Similar models might explain differential control by other splicing regulators within families of paralogs with indistinguishable RNA binding sites.

STAR (signal transduction and activation of RNA) proteins regulate splicing of target genes that have roles in neural connectivity, survival and myelination in the vertebrate nervous system. These regulated splicing targets include mRNAs such as the Neurexins (Nrxn), SMN2 (survival of motor neuron) and MAG (myelin-associated glycoprotein). Recent work has made it possible to identify and validate STAR protein splicing targets in vivo by using genetically modified mouse models. In this review, we will discuss the importance of STAR protein splicing targets in the CNS (central nervous system).

The brain is made up of trillions of synaptic connections that together form neural networks needed for normal brain function and behavior. SLM2 is a member of a conserved family of RNA binding proteins, including Sam68 and SLM1, that control splicing of Neurexin1-3 pre-mRNAs. Whether SLM2 affects neural network activity is unknown. Here, we find that SLM2 levels are maintained by a homeostatic feedback control pathway that predates the divergence of SLM2 and Sam68. SLM2 also controls the splicing of Tomosyn2, LysoPLD/ATX, Dgkb, Kif21a, and Cask, each of which are important for synapse function. Cortical neural network activity dependent on synaptic connections between SLM2-expressing-pyramidal neurons and interneurons is decreased in Slm2-null mice. Additionally, these mice are anxious and have a decreased ability to recognize novel objects. Our data reveal a pathway of SLM2 homeostatic auto-regulation controlling brain network activity and behavior.

Tra2 proteins regulate pre-mRNA splicing in vertebrates and invertebrates, and are involved in important processes ranging from brain development in mice to sex determination in fruitflies. In structure Tra2 proteins contain two RS domains (domains enriched in arginine and serine residues) flanking a central RRM (RNA recognition motif). Understanding the mechanisms of how Tra2 proteins work to control splicing is one of the key requirements to understand their biology. In the present article, we review what is known about how Tra2 proteins regulate splicing decisions in mammals and fruitflies.

The splicing regulator proteins SRSF1 (also known as ASF/SF2) and SRSF3 (also known as SRP20) belong to the SR family of proteins and can be upregulated in cancer. The SRSF1 gene itself is amplified in some cancer cells, and cancer-associated changes in the expression of MYC also increase SRSF1 gene expression. Increased concentrations of SRSF1 protein promote prooncogenic splicing patterns of a number of key regulators of cell growth. Here, we review the evidence that upregulation of the SR-related Tra2β protein might have a similar role in cancer cells. The TRA2B gene encoding Tra2β is amplified in particular tumours including those of the lung, ovary, cervix, stomach, head, and neck. Both TRA2B RNA and Tra2β protein levels are upregulated in breast, cervical, ovarian, and colon cancer, and Tra2β expression is associated with cancer cell survival. The TRA2B gene is a transcriptional target of the protooncogene ETS-1 which might cause higher levels of expression in some cancer cells which express this transcription factor. Known Tra2β splicing targets have important roles in cancer cells, where they affect metastasis, proliferation, and cell survival. Tra2β protein is also known to interact directly with the RBMY protein which is implicated in liver cancer.

The RNA binding protein T-STAR was created following a gene triplication 520-610 million years ago, which also produced its two parologs Sam68 and SLM-1. Here we have created a T-STAR null mouse to identify the endogenous functions of this RNA binding protein. Mice null for T-STAR developed normally and were fertile, surprisingly, given the high expression of T-STAR in the testis and the brain, and the known infertility and pleiotropic defects of Sam68 null mice. Using a transcriptome-wide search for splicing targets in the adult brain, we identified T-STAR protein as a potent splicing repressor of the alternatively spliced segment 4 (AS4) exons from each of the Neurexin1-3 genes, and exon 23 of the Stxbp5l gene. T-STAR protein was most highly concentrated in forebrain-derived structures like the hippocampus, which also showed maximal Neurexin1-3 AS4 splicing repression. In the absence of endogenous T-STAR protein, Nrxn1-3 AS4 splicing repression dramatically decreased, despite physiological co-expression of Sam68. In transfected cells Neurexin3 AS4 alternative splicing was regulated by either T-STAR or Sam68 proteins. In contrast, Neurexin2 AS4 splicing was only regulated by T-STAR, through a UWAA-rich response element immediately downstream of the regulated exon conserved since the radiation of bony vertebrates. The AS4 exons in the Nrxn1 and Nrxn3 genes were also associated with distinct patterns of conserved UWAA repeats. Consistent with an ancient mechanism of splicing control, human T-STAR protein was able to repress splicing inclusion of the zebrafish Nrxn3 AS4 exon. Although Neurexin1-3 and Stxbp5l encode critical synaptic proteins, T-STAR null mice had no detectable spatial memory deficits, despite an almost complete absence of AS4 splicing repression in the hippocampus. Our work identifies T-STAR as an ancient and potent tissue-specific splicing regulator that uses a concentration-dependent mechanism to co-ordinately regulate regional splicing patterns of the Neurexin1-3 AS4 exons in the mouse brain.

Meiosis requires conserved transcriptional changes, but it is not known whether there is a corresponding set of RNA splicing switches. Here, we used RNAseq of mouse testis to identify changes associated with the progression from mitotic spermatogonia to meiotic spermatocytes. We identified ∼150 splicing switches, most of which affect conserved protein-coding exons. The expression of many key splicing regulators changed in the course of meiosis, including downregulation of polypyrimidine tract binding protein (PTBP1) and heterogeneous nuclear RNP A1, and upregulation of nPTB, Tra2β, muscleblind, CELF proteins, Sam68 and T-STAR. The sequences near the regulated exons were significantly enriched in target sites for PTB, Tra2β and STAR proteins. Reporter minigene experiments investigating representative exons in transfected cells showed that PTB binding sites were critical for splicing of a cassette exon in the Ralgps2 mRNA and a shift in alternative 5′ splice site usage in the Bptf mRNA. We speculate that nPTB might functionally replace PTBP1 during meiosis for some target exons, with changes in the expression of other splicing factors helping to establish meiotic splicing patterns. Our data suggest that there are substantial changes in the determinants and patterns of alternative splicing in the mitotic-to-meiotic transition of the germ cell cycle.