Supplementary Figures S1-S2 and Supplementary Tables S2-S4.

Deep intron space harbors a diverse array of splicing regulatory elements that cooperate with better-known exon-proximal elements to enforce proper tissue-specific and development-specific pre-mRNA processing. Many deep intron elements have been highly conserved through vertebrate evolution, yet remain poorly annotated in the human genome. Recursive splicing exons (RS-exons) and intraexons promote noncanonical, multistep resplicing pathways in long introns, involving transient intermediate structures that are greatly underrepresented in RNA-seq datasets. Decoy splice sites and decoy exons act at a distance to inhibit splicing catalysis at annotated splice sites, with functional consequences such as exon skipping and intron retention. RNA:RNA bridges can juxtapose distant sequences within or across introns to activate deep intron splicing enhancers and silencers, to loop out exons to be skipped, or to select one member of a mutually exclusive set of exons. Similarly, protein bridges mediated by interactions among transcript-bound RNA binding proteins (RBPs) can modulate splicing outcomes. Experimental disruption of deep intron elements serving any of these functions can abrogate normal splicing, strongly suggesting that natural mutations of deep intron elements can do likewise to cause human disease. Understanding noncanonical splicing pathways and discovering deep intron regulatory signals, many of which map hundreds to many thousands of nucleotides from annotated splice junctions, is of great academic interest for basic scientists studying alternative splicing mechanisms. Hopefully, this knowledge coupled with increased analysis of deep intron sequences will also have important medical applications, as better interpretation of deep intron mutations may reveal new disease mechanisms and suggest new therapies. This article is categorized under: RNA Processing > Splicing Regulation/Alternative Splicing.

A translationally silent single nucleotide mutation in exon 44 (E44) of the von Willebrand factor (VWF) gene is associated with inefficient removal of intron 44 in a von Willebrand disease (VWD) patient. This intron retention (IR) event was previously attributed to reordered E44 secondary structure that sequesters the normal splice donor site. We propose an alternative mechanism: the mutation introduces a cryptic splice donor site that interferes with the function of the annotated site to favor IR. We evaluated both models using minigene splicing reporters engineered to vary in secondary structure and/or cryptic splice site content. Analysis of splicing efficiency in transfected K562 cells suggested that the mutation-generated cryptic splice site in E44 was sufficient to induce substantial IR. Mutations predicted to vary secondary structure at the annotated site also had modest effects on IR and shifted the balance of residual splicing between the cryptic site and annotated site, supporting competition among the sites. Further studies demonstrated that introduction of cryptic splice donor motifs at other positions in E44 did not promote IR, indicating that interference with the annotated site is context dependent. We conclude that mutant deep exon splice sites can interfere with proper splicing by inducing IR.

The decoy exon model has been proposed to regulate a subset of intron retention (IR) events involving predominantly larger introns (>1kb). Splicing reporter studies have shown that decoy splice sites are essential for activity, suggesting that decoys act by engaging intron-terminal splice sites and competing with cross-intron interactions required for intron excision. The decoy model predicts that antisense oligonucleotides blocking decoy splice sites in endogenous pre-mRNA should increase productive gene expression by reducing IR. Indeed, we now demonstrate that targeting a decoy 5′ splice site in the O-GlcNAc transferase (OGT) gene reduced IR from ∼80% to ∼20% in primary human erythroblasts, accompanied by increases in spliced OGT RNA and OGT protein expression. The remaining OGT IR was refractory to antisense treatment and might be mediated by independent mechanism(s). In contrast, other retained introns were strongly dependent on decoy function, since IR was nearly eliminated by antisense targeting of 5′ splice sites. Genes in the latter group encode the widely expressed splicing factor (SF3B1), and the erythroid-specific structural protein, alpha-spectrin (SPTA1). These results show that modulating decoy exon function can dramatically alter IR, and suggest that dynamic regulation of decoy exons could be a mechanism to fine tune gene expression post-transcriptionally in many cell types.

Protein 4.1N, a member of the protein 4.1 family, is highly expressed in the brain. But its function remains to be fully defined. Using 4.1N−/− mice, we explored the function of 4.1N in vivo. We show that 4.1N−/− mice were born at a significantly reduced Mendelian ratio and exhibited high mortality between 3 to 5 weeks of age. Live 4.1N−/− mice were smaller than 4.1N+/+ mice. Notably, while there were no significant differences in organ/body weight ratio for most of the organs, the testis/body and ovary/body ratio were dramatically decreased in 4.1N−/− mice, demonstrating selective effects of 4.1N deficiency on the development of the reproductive systems. Histopathology of the reproductive organs showed atrophy of both testis and ovary. Specifically, in the testis there is a lack of spermatogenesis, lack of leydig cells and lack of mature sperm. Similarly, in the ovary there is a lack of follicular development and lack of corpora lutea formation, as well as lack of secretory changes in the endometrium. Examination of pituitary glands revealed that the secretory granules were significantly decreased in pituitary glands of 4.1N−/− compared to 4.1N+/+. Moreover, while GnRH was expressed in both neuronal cell body and axons in the hypothalamus of 4.1N+/+ mice, it was only expressed in the cell body but not the axons of 4.1N-/- mice. Our findings uncover a novel role for 4.1N in the axis of hypothalamus-pituitary gland-reproductive system.

In this issue of Blood, Ji et al have identified a novel regulatory mechanism that ensures proper splicing of human alpha-globin pre-messenger RNA (pre-mRNA).

Diamond Blackfan Anemia Syndrome (DBA) is a rare, congenital bone marrow failure syndrome characterized by severe macrocytic anemia, most often diagnosed during infancy. Congenital anomalies and predisposition to cancer are also important features of DBA. Establishment of a molecular diagnosis in a patient with DBA is critical to determine treatment strategies (i.e. the identification of compatible related transplant donors), as well as developing reproductive strategies for genetically at risk families. The overwhelming majority (>98.75%) of DBA patients with a molecular diagnosis have mutations in a Ribosomal Protein (RP) gene. Targeted and exome sequencing (WES) strategies can identify RP mutations in >70% of DBA patients (Ulrisch et al. Am J Hum Genet. 2018). Single Nucleotide Polymorphism Comparative Genome Hybridization (SNP array) detects >30 kb deletions of RP genes (which cannot be identified by sequencing) in ~10% of DBA patients (Farrar et al. Blood. 2011), leaving ~20% of DBA patients without a molecular diagnosis. We hypothesized that smaller copy number variants (CNVs - either insertions or deletions) in RP genes that are below the limit of detection of SNP array are responsible for the remaining 20%. To test this hypothesis we collected DNA with informed consent for whole genome sequencing (WGS) analysis from 6 patients who had no mutations detected by WES or SNP array. On average, we aligned ~1x1010paired end reads of 250 base pairs for each patient (~83X coverage of the genome). The aligned sequences were analyzed for CNVs using two independent software packages. Delly analyzes the two ends of each sequence read and maps them to the current human reference genome. Read ends that map further apart than expected are flagged as potential CNVs. CNVkit estimates regions of copy loss by changes in average sequencing depth. Using relatively relaxed thresholds in Delly and CNVkit we identified ~100 candidate CNVs in each patient. We filtered out CNVs present in public databases and focused on those CNVs in the region of the RP genes. This analysis identified 2-5 potential RP gene associated CNVs in each patient. We designed PCR primers that flanked each putative CNV and confirmed at least one RP CNV in all 6 patient DNAs. At this time, the CNVs in two patients are in the process of evaluation. We have validated causative RP CNVs in the other 4 patients, representing one known and three novel DBA genes. One patient had a 464 bp deletion in 3rdintron of the RPL27 gene, which is mutated in other DBA patients. We hypothesized that the deletion caused a splicing defect. Using a mini gene in which the second intron of the gamma globin gene was replaced with the 3rdintron of either the wild type or mutant RPL27 gene, we showed that the mutant exon was not spliced. An alternative hypothesis, that the deletion removed an enhancer element, was also tested, but no enhancer activity was detected. We conclude that the RPL27exon deletion causes aberrant splicing leading to an unstable RPL27 mRNA and haploinsufficiency of RPL27. A second patient had a 3.5 kb deletion at the 3' end of the RPS5 gene, including the stop codon and poly A addition site. We hypothesized that the lack of the 3' processing signals would lead to an unstable mRNA. To test this hypothesis we generated MYC and FLAG tagged wildtype and 3' deleted RPS5 genes and co-transfected them into 293T cells. Regardless of the tag used, RT-PCR analysis showed a severe reduction in the mutant mRNA levels. Western Blot analysis demonstrated that only the wild type protein was expressed, leading to the conclusion that the RPS5 truncation led to an unstable RPS5 mRNA and haploinsufficiency of RPS5. A third patient had a 28 kb deletion that removes the RPS9 gene. shRNA knockdown of RPS9 mRNA in normal CD34+ cells inhibited erythroid differentiation, leading to the conclusion that RPS9 deficiency causes DBA. Finally, we observed a 3 bp insertion in exon 6 of the RPL14 gene. The deletion adds an alanine residue to a string of 10 alanines in the wild type allele. We confirmed the insertion by targeted sequence analysis of patient DNA. Our data show that WGS can identify small CNVs that cause DBA in at least 2/3 of patients who do not have a mutation detectable by other methods. We believe that WGS analysis following targeted sequencing, SNP array and WES can identify virtually all DBA mutations. With declining WGS costs, we recommend adding WGS to the molecular diagnostic pipeline for genetic testing of DBA. Disclosures Farrar: Novartis: Research Funding. Vlachos:Novartis: Other: Steering committee member.

The erythroid transcriptome is extensively remodeled during terminal erythropoiesis by dynamic changes in RNA splicing of cassette exons and retained introns. Mechanistic studies of these RNA processing networks will provide new insight into pathways that impact structure and function of the erythroid proteome during erythroblast differentiation. We previously showed that up-regulation of EPB41 exon 16 splicing imparts new functionality to the encoded protein, enhancing protein-protein interactions that mechanically strengthen the red cell membrane. More recently, RNA-seq analysis revealed that numerous erythroid transcripts exhibit up-regulation of intron retention (IR) events, some of which are controlled by a decoy exon-mediated mechanism that can reduce the output of translated mRNA so as to limit protein expression. Here we demonstrate that modulation of decoy-mediated IR quantitatively affects protein expression in primary human erythroid cells. We first studied the OGT gene (O-GlcNAc transferase), a key regulator of O-GlcNAC homeostasis. OGT expression responds to pharmacological inhibitors by regulating intron 4 retention, by a mechanism requiring a intronic splicing silencer (1) that functions as a decoy exon (2) to nonproductively engage annotated splice sites at the ends of the intron, thereby blocking excision and enforcing its retention. In erythroid CD34+ progenitors at day 7 of culture, we found that treatment with an OGT inhibitor (OSMI-1) reduced OGT IR and increased spliced OGT RNA and OGT protein. Conversely, an inhibitor of the antagonistic OGA enzyme (thiamet-G) induced greater OGT IR and reduced OGT protein expression. These results are similar to what was reported previously in established cell lines (1), and suggest that modulation of IR can vary mRNA and protein expression in primary cells >5-fold. To further explore the model, we independently blocked OGT IR in erythroid cultures by electroporation of an antisense morpholino directed against the OGT decoy exon 5' splice site. RT-qPCR and western blot analysis confirmed substantial reduction in IR, coupled with an increase in spliced RNA and elevated OGT protein expression, compared to control cells or cells treated with an irrelevant morpholino. The OGT-specific MO also substantially blocked IR induced by thiamet-G. These results show that pharmacological- or antisense-mediated alteration in IR can significantly change protein expression in primary erythroblast cultures. We propose that the abundance of IR transcripts in late erythropoiesis represents a widespread modulation of protein output by post-transcriptional pathways operating at the level of intron retention. Park SK, et al. (2017) Cell Rep. 20: 1088-99.Parra M et al. (2018) RNA 24: 1255-65. Disclosures No relevant conflicts of interest to declare.

During terminal erythropoiesis, the splicing machinery in differentiating erythroblasts executes a robust intron retention (IR) program that impacts expression of hundreds of genes. We studied IR mechanisms in the SF3B1 splicing factor gene, which expresses ∼50% of its transcripts in late erythroblasts as a nuclear isoform that retains intron 4. RNA-seq analysis of nonsense-mediated decay (NMD)-inhibited cells revealed previously undescribed splice junctions, rare or not detected in normal cells, that connect constitutive exons 4 and 5 to highly conserved cryptic cassette exons within the intron. Minigene splicing reporter assays showed that these cassettes promote IR. Genome-wide analysis of splice junction reads demonstrated that cryptic noncoding cassettes are much more common in large (>1 kb) retained introns than they are in small retained introns or in nonretained introns. Functional assays showed that heterologous cassettes can promote retention of intron 4 in the SF3B1 splicing reporter. Although many of these cryptic exons were spliced inefficiently, they exhibited substantial binding of U2AF1 and U2AF2 adjacent to their splice acceptor sites. We propose that these exons function as decoys that engage the intron-terminal splice sites, thereby blocking cross-intron interactions required for excision. Developmental regulation of decoy function underlies a major component of the erythroblast IR program.

Proper expression of the MDS-disease gene, SF3B1, ensures appropriate pre-mRNA splicing in erythroid progenitors and during terminal erythropoiesis. We previously showed that the SF3B1 gene is post-transcriptionally regulated in a differentiation stage-specific manner by intron retention (IR), such that ~50% of its transcripts in mature erythroblasts retain intron 4. Based on new mechanistic studies, we propose a model in which mostly unannotated and noncoding exons within intron 4 function as splicing decoys; i.e., they promote retention of intron 4 by interacting with, and blocking splice sites of, the adjacent exons 4 and 5. A total of six putative decoy exons were revealed via RT-PCR and RNA-seq analysis of RNA from erythroblasts treated with inhibitors of nonsense-mediated decay. That decoy exons have IR-promoting activity is suggested by several criteria. First, the frequency of interaction between constitutive exons 4 and 5 and putative decoy exons within intron 4, measured by the abundance of splice junctions in RNA-seq read data, is temporally correlated with levels of intron 4 retention during terminal erythropoiesis. Both IR and decoy splice junctions were low in early stage erythroblasts and much higher in mature erythroblasts. Second, selected decoy exons exhibited IR-promoting activity in the context of minigene splicing reporters expressing the exon 3-6 region of SF3B1 in transfected K562 cells. The wild type minigene reproduced the intron-specific retention phenotype, since it was fully spliced at introns 3 and 5 but exhibited substantial retention of intron 4, whereas deletion of decoy exon 4e, or mutation of its splice sites, substantially decreased IR. Third, RBP (RNA binding protein) cross-linking data from K562 cells show that 3‘ splice site factors including U2AF1 and U2AF2 can bind specifically to 3‘ splice sites of intron 4's decoy exons. Finally, several experiments showed that IR-promoting activity of decoy exons is a more general phenomenon that likely governs IR in other erythroid genes. We observed not only that SF3B1 intron 4 decoy exons could promote IR in heterologous contexts, but also that predicted decoy exons from other erythroblast transcripts could promote IR in the SF3B1 minigene. Apart from this experimental data, comparative genomics revealed that the SF3B1 decoy exons are extremely conserved among vertebrate genomes, with two of the exons being essentially identical from fish to humans. Together this data supports the hypothesis that a subset of up-regulated IR events in late erythroblasts are controlled by decoy exons that block productive splicing at the flanking exons. We propose that regulated IR is an important post-transcriptional mechanism for adjusting cellular splicing capacity during terminal erythropoiesis by regulating expression of key splicing factors such as SF3B1.

Hematopoietic ontogeny is characterized by distinct primitive and definitive erythroid lineages. Definitive erythroblasts mature and enucleate extravascularly and form a unique membrane skeleton, composed of spectrin, 4.1R-complex, and ankyrinR-complex components, to survive the vicissitudes of the adult circulation. However, little is known about the formation and composition of the membrane skeleton in primitive erythroblasts, which progressively mature while circulating in the embryonic bloodstream. We found that primary primitive erythroblasts express the major membrane skeleton genes present in similarly staged definitive erythroblasts, suggesting that the composition and formation of this membrane network is conserved in maturing primitive and definitive erythroblasts despite their respective intravascular and extravascular locations. Membrane deformability and stability of primitive erythroblasts, assayed by microfluidic studies and fluorescence imaged microdeformation, respectively, significantly increase prior to enucleation. These functional changes coincide with protein 4.1 R isoform switching and protein 4.1R-null primitive erythroblasts fail to establish normal membrane stability and deformability. We conclude that maturing primitive erythroblasts initially navigate the embryonic vasculature prior to establishing a deformable cytoskeleton, which is ultimately formed prior to enucleation. Formation of an erythroid-specific, protein 4.1R-dependent membrane skeleton is an important feature not only of definitive, but also of primitive, erythropoiesis in mammals.

The Rbfox genes encode an ancient family of sequence‐specific RNA binding proteins (RBPs) that are critical developmental regulators in multiple tissues including skeletal muscle, cardiac muscle, and brain. The hallmark of Rbfox proteins is a single high‐affinity RRM domain, highly conserved from insects to humans, that binds preferentially to UGCAUG motifs at diverse regulatory sites in pre‐mRNA introns, mRNA 3’UTRs, and pre‐miRNAs hairpin structures. Versatile regulatory circuits operate on Rbfox pre‐mRNA and mRNA to ensure proper expression of Rbfox1 protein isoforms, which then act on the broader transcriptome to regulate alternative splicing networks, mRNA stability and translation, and microRNA processing. Complex Rbfox expression is encoded in large genes encompassing multiple promoters and alternative splicing options that govern spatiotemporal expression of structurally distinct and tissue‐specific protein isoforms with different classes of RNA targets. Nuclear Rbfox1 is a candidate master regulator that binds intronic UGCAUG elements to impact splicing efficiency of target alternative exons, many in transcripts for other splicing regulators. Tissue‐specificity of Rbfox‐mediated alternative splicing is executed by combinatorial regulation through the integrated activity of Rbfox proteins and synergistic or antagonistic splicing factors. Studies in animal models show that Rbfox1‐related genes are critical for diverse developmental processes including germ cell differentiation and memory in Drosophila, neuronal migration and function in mouse brain, myoblast fusion and skeletal muscle function, and normal heart function. Finally, genetic and biochemical evidence suggest that aberrations in Rbfox‐regulated circuitry are risk factors for multiple human disorders, especially neurodevelopmental disorders including epilepsy and autism, and cardiac hypertrophy. WIREs RNA 2017, 8:e1398. doi: 10.1002/wrna.1398This article is categorized under: RNA Processing > Splicing Regulation/Alternative Splicing

Intron retention (IR) regulates hundreds of erythroid genes in a differentiation stage-specific manner during terminal erythropoiesis. Regulated genes include highly expressed RNA binding proteins (RBPs) such as SF3B1, as well as iron transporters (e.g., mitoferrins 1 and 2), and cytoskeletal proteins (e.g., alpha spectrin). Selected IR transcripts are relatively abundant; 25-50% of the above-mentioned transcripts can be polyadenylated, retained in the nucleus, and efficiently spliced at all but one or two introns, thus limiting the amount of translatable cytoplasmic mRNA. We are studying novel IR regulatory mechanisms involving both splice site strength and deeper intronic elements, using model introns that are differentially regulated in human erythroblasts. SF3B1, a key pre-RNA splicing factor implicated in MDS, exhibits dynamic regulation of intron 4, with low IR in proerythroblasts and high IR in mature orthochromatic erythroblasts. Intron 4 sequences include three ultraconserved elements that encode cryptic exons we term 4a, 4b, and 4c, two of which (4a and 4c) encode premature termination codons (PTCs) expected to induce nonsense-mediated decay (NMD) if spliced into SF3B1 transcripts. We hypothesize that these PTC exons can have an additional function as inefficiently spliced decoy(s), that is, their splice sites may interact with splice sites at the boundaries of intron 4 to form non-productive complexes that do not permit efficient splicing but instead prevent excision of the intron. This concept represents an extension of decoy models previously proposed by others to explain selected exon skipping events, and is supported by several recent findings: eCLIP (enhanced cross-linking and immunoprecipitation) data indicate strong binding of 3' splice site factors U2AF1 and U2AF2 at these exons; low levels of exon 4a and 4c splicing can be seen in NMD-inhibited cells; and deletion of exon 4c from minigene splicing reporters decreases IR. Besides SF3B1, we identified numerous other dynamically-regulated IR events encompassing cryptic PTC exons that bind 3' splice site factors. In contrast to the dynamic regulation of IR in SF3B1, IR in SLC25A37 (mitoferrin-1) and SLC25A28 (mitoferrin-2) is stably maintained at a high level throughout terminal erythropoiesis. Baseline IR levels for stably retained introns correlate with splice site strength, but are also influenced by deeper intronic elements. For example, SLC25A28 intron 2 contains intronic elements that appear to reduce IR, since blocking them with antisense morpholinos leads to substantially increased IR levels. These studies demonstrate that IR in major erythroid genes is regulated by sequences within the retained introns that can either increase or decrease retention, suggesting that multiple IR pathways are employed during terminal erythropoiesis to regulate gene expression.

Purpose of reviewErythroid progenitors must accurately and efficiently splice thousands of pre-mRNAs as the cells undergo extensive changes in gene expression and cellular remodeling during terminal erythropoiesis. Alternative splicing choices are governed by interactions between RNA binding proteins and cis-regulatory binding motifs in the RNA. This review will focus on recent studies that define the genome-wide scope of splicing in erythroblasts and discuss what is known about its regulation.Recent findingsRNA-seq analysis of highly purified erythroblast populations has revealed an extensive program of alternative splicing of both exons and introns. During normal erythropoiesis, stage-specific splicing transitions alter the structure and abundance of protein isoforms required for optimized red cell production. Mutation or deficiency of splicing regulators underlies hematopoietic disease in myelopdysplasia syndrome patients via disrupting the splicing program.SummaryErythroid progenitors execute an elaborate alternative splicing program that modulates gene expression posttranscriptionally, ultimately regulating the structure and function of the proteome in a differentiation stage-specific manner during terminal erythropoiesis. This program helps drive differentiation and ensure synthesis of the proper protein isoforms required to produce mechanically stable red cells. Mutation or deficiency of key splicing regulatory proteins disrupts the splicing program to cause disease.

Computational analysis of RNA-seq data from highly purified human erythroblasts has been instrumental in revealing changes in pre-mRNA splicing during terminal erythropoiesis. Here we report updated studies of intron retention (IR), a type of alternative splicing in which specific introns are retained in otherwise efficiently-processed transcripts, allowing post-transcriptional modulation of cellular mRNA levels. Differences in differentiation stage-specificity, degree of retention, nuclear/cytoplasmic localization, and sensitivity to nonsense-mediated decay (NMD) suggest the existence of multiple classes of erythroblast IR subject to distinct regulatory controls. Two clusters comprising ~470 "developmentally dynamic" introns in 354 genes exhibit more efficient splicing in proerythroblasts, but elevated intron retention in orthochromatic erythroblasts prior to enucleation. Dynamic regulation of late erythroblast IR parallels previously described splicing switches involving alternative exons. Gene ontology analysis revealed that the dynamic intron group is highly enriched in genes with RNA processing functions. Among these are several spliceosomal factors including SF3B1, a commonly mutated gene in myelodysplasia patients. We also identified several clusters of "developmentally stable" introns whose IR levels are not substantially modulated during erythropoiesis. Among this latter type are two clusters containing 294 introns that are enriched in functions related to metal ion binding. Key genes include mitoferrin-1 (SC25A37; IR~50%) and mitoferrin-2 (SLC25A28; IR~20-30%), mitochondrial iron importers essential for heme biosynthesis. We observed a correlation between splice site strength and percent IR among developmentally stable but not dynamic intron clusters, indicating that splicing regulatory mechanism(s) for the latter must require additional sequence features. A search for such features revealed that IR was significantly higher adjacent to alternative 'PTC' exons containing premature termination codons than it was adjacent to other exons; moreover, by direct RT-PCR analysis we discovered novel (unannotated) PTC exons in additional retained introns. The proposed role of PTC exons in IR is being studied experimentally using an array of minigene splicing reporter constructs. Finally, we noted that while specific IR events are erythroid specific, e.g., in the alpha spectrin gene SPTA1, computational analysis of public RNA-seq data demonstrated that most erythroblast IR events were also observed in granulocytes and in 16 other tissues surveyed by the human BodyMap project. Intron retention is likely to play critical roles in gene regulation in both hematological and non-hematological tissues.

We present a tool, keep me around (kma), a suite of python scripts and an R package that finds retained introns in RNA-Seq experiments and incorporates biological replicates to reduce the number of false positives when detecting retention events. kma uses the results of existing quantification tools that probabilistically assign multi-mapping reads, thus interfacing easily with transcript quantification pipelines. The data is represented in a convenient, database style format that allows for easy aggregation across introns, genes, samples, and conditions to allow for further exploratory analysis.

Erythroid RNAs, like their nonerythroid counterparts, are subject to post-transcriptional processing events that critically impact their coding capacity for the erythroid proteome. Previous studies have shown that differentiating human and mouse erythroblasts execute an extensive and dynamic alternative splicing program involving regulation of numerous alternative exons. Here we report that controlled excision of selected introns is also an important component of the erythroblast alternative splicing program. Intron retention (IR) patterns in differentiating human erythroblasts were determined via RNA-seq analysis of FACS-purified erythroblast populations. Comparison of IR among erythroblast populations and between erythroblasts and other hematopoietic cells suggests that regulation of IR occurs in a differentiation stage- and tissue-specific manner. For example, there was little overlap of intron retention events in erythroblasts with those reported in differentiating granulocytes. Moreover, the IR profile of proerythroblasts differed substantially from that in orthochromatic erythroblasts, with IR generally increasing in the more mature cells that are preparing for enucleation. IR in erythroblasts affected numerous genes functioning in RNA processing, iron homeostasis and heme biosynthesis, protein translation, and membrane properties. Mature erythroblasts exhibited retention of introns in several human disease genes including SF3B1, a splicing factor often mutated in myelodysplasia; TFR2, encoding transferrin receptor 2 that is mutated in a form of hemochromatosis; and FUS, an RNA binding protein implicated in ALS. Inspection of intronic RNA-seq reads in >60 genes with IR revealed that single or multiple introns can be retained within a transcript; however, other introns within the same genes, and indeed the great majority of introns in erythroblast-expressed genes, are efficiently spliced with minimal or no IR. Retained introns may be flanked by either constitutively or alternatively spliced exons, suggesting different regulatory mechanisms. Ongoing studies will explore whether IR in some transcripts might function to down-regulate gene expression by introduction of premature termination codons that induce nonsense-mediated decay, or alternatively, whether IR transcripts could represent a reserve of nearly-completed mRNAs that can be processed in response to appropriate physiological stimuli. In sum, these results suggest that a highly regulated IR program plays an important role in erythroid differentiation. Disclosures No relevant conflicts of interest to declare.

Alternative pre-messenger RNA splicing remodels the human transcriptome in a spatiotemporal manner during normal development and differentiation. Here we explored the landscape of transcript diversity in the erythroid lineage by RNA-seq analysis of five highly purified populations of morphologically distinct human erythroblasts, representing the last four cell divisions before enucleation. In this unique differentiation system, we found evidence of an extensive and dynamic alternative splicing program encompassing genes with many diverse functions. Alternative splicing was particularly enriched in genes controlling cell cycle, organelle organization, chromatin function and RNA processing. Many alternative exons exhibited differentiation-associated switches in splicing efficiency, mostly in late-stage polychromatophilic and orthochromatophilic erythroblasts, in concert with extensive cellular remodeling that precedes enucleation. A subset of alternative splicing switches introduces premature translation termination codons into selected transcripts in a differentiation stage-specific manner, supporting the hypothesis that alternative splicing-coupled nonsense-mediated decay contributes to regulation of erythroid-expressed genes as a novel part of the overall differentiation program. We conclude that a highly dynamic alternative splicing program in terminally differentiating erythroblasts plays a major role in regulating gene expression to ensure synthesis of appropriate proteome at each stage as the cells remodel in preparation for production of mature red cells.