Search

RNA editing, the endogenous modification of nucleic acids, is known to be altered in genes with important neurological function in schizophrenia (SCZ). However, the global profile and molecular functions of disease-associated RNA editing remain unclear. Here, we analyzed RNA editing in postmortem brains of four SCZ cohorts and uncovered a significant and reproducible trend of hypoediting in patients of European descent. We report a set of SCZ-associated editing sites via WGCNA analysis, shared across cohorts. Using massively parallel reporter assays and bioinformatic analyses, we observed that differential 3′ untranslated region (3′UTR) editing sites affecting host gene expression were enriched for mitochondrial processes. Furthermore, we characterized the impact of two recoding sites in the mitofusin 1 ( MFN1 ) gene and showed their functional relevance to mitochondrial fusion and cellular apoptosis. Our study reveals a global reduction of editing in SCZ and a compelling link between editing and mitochondrial function in the disease.

Supplementary Figures S1-11 and Tables S1-2 are included in this file

RNA editing modifies single nucleotides of RNAs, regulating primary protein structure and protein abundance. In recent years, the diversity of proteins and complexity of gene regulation associated with RNA editing dysregulation has been increasingly appreciated in oncology. Large-scale shifts in editing have been observed in bulk tumors across various cancer types. However, RNA editing in single cells and individual cell types within tumors has not been explored. By profiling editing in single cells from lung adenocarcinoma biopsies, we found that the increased editing trend of bulk lung tumors was unique to cancer cells. Elevated editing levels were observed in cancer cells resistant to targeted therapy, and editing sites associated with drug response were enriched. Consistent with the regulation of antiviral pathways by RNA editing, higher editing levels in cancer cells were associated with reduced antitumor innate immune response, especially levels of natural killer cell infiltration. In addition, the level of RNA editing in cancer cells was positively associated with somatic point mutation burden. This observation motivated the definition of a new metric, RNA editing load, reflecting the amount of RNA mutations created by RNA editing. Importantly, in lung cancer, RNA editing load was a stronger predictor of patient survival than DNA mutations. This study provides the first single cell dissection of editing in cancer and highlights the significance of RNA editing load in cancer prognosis.Significance:RNA editing analysis in single lung adenocarcinoma cells uncovers RNA mutations that correlate with tumor mutation burden and cancer innate immunity and reveals the amount of RNA mutations that strongly predicts patient survival.See related commentary by Luo and Liang, p. 351

RNA splicing plays a critical role in post-transcriptional gene regulation. Exponential expansion of intron length poses a challenge for accurate splicing. Little is known about how cells prevent inadvertent and often deleterious expression of intronic elements due to cryptic splicing. In this study, we identify hnRNPM as an essential RNA binding protein that suppresses cryptic splicing through binding to deep introns, preserving transcriptome integrity. Long interspersed nuclear elements (LINEs) harbor large amounts of pseudo splice sites in introns. hnRNPM preferentially binds at intronic LINEs and represses LINE-containing pseudo splice site usage for cryptic splicing. Remarkably, a subgroup of the cryptic exons can form long dsRNAs through base-pairing of inverted Alu transposable elements scattered in between LINEs and trigger interferon immune response, a well-known antiviral defense mechanism. Notably, these interferon-associated pathways are found to be upregulated in hnRNPM-deficient tumors, which also exhibit elevated immune cell infiltration. These findings unveil hnRNPM as a guardian of transcriptome integrity. Targeting hnRNPM in tumors may be used to trigger an inflammatory immune response thereby boosting cancer surveillance.

RNA editing modifies single nucleotides of RNAs, regulating primary protein structure and protein abundance. In recent years, the diversity of proteins and complexity of gene regulation associated with RNA editing dysregulation has been increasingly appreciated in oncology. Large-scale shifts in editing have been observed in bulk tumors across various cancer types. However, RNA editing in single cells and individual cell types within tumors has not been explored. By profiling editing in single cells from lung adenocarcinoma biopsies, we found that the increased editing trend of bulk lung tumors was unique to cancer cells. Elevated editing levels were observed in cancer cells resistant to targeted therapy, and editing sites associated with drug response were enriched. Consistent with the regulation of antiviral pathways by RNA editing, higher editing levels in cancer cells were associated with reduced antitumor innate immune response, especially levels of natural killer cell infiltration. In addition, the level of RNA editing in cancer cells was positively associated with somatic point mutation burden. This observation motivated the definition of a new metric, RNA editing load, reflecting the amount of RNA mutations created by RNA editing. Importantly, in lung cancer, RNA editing load was a stronger predictor of patient survival than DNA mutations. This study provides the first single cell dissection of editing in cancer and highlights the significance of RNA editing load in cancer prognosis. Significance: RNA editing analysis in single lung adenocarcinoma cells uncovers RNA mutations that correlate with tumor mutation burden and cancer innate immunity and reveals the amount of RNA mutations that strongly predicts patient survival. See related commentary by Luo and Liang, p. 351

To identify genetic pathways for addiction, we analyzed intravenous self-administration of cocaine or saline in a panel of 84 inbred and recombinant inbred mouse strains over 10 days. We integrated the behavior data with RNA-Seq data from the medial frontal cortex and nucleus accumbens from 41 strains. The self-administration of cocaine and saline showed distinct genetic bases. We maximized power to map loci for cocaine intake by using a linear mixed model to account for this longitudinal phenotype while correcting for population structure. A total of 15 unique significant loci were identified in the genome-wide association study (GWAS). A transcriptome-wide association study (TWAS) highlighted theTrpv2ion channel as a key locus for cocaine self-administration from the GWAS. In addition, 17 genes supplementary to the GWAS were identified includingArhgef26, Slc18b1andSlco5a1. We found numerous instances where alternate splice site selection or RNA editing altered transcript abundance. Our work emphasizes the importance ofTrpv2, a known cannabinoid receptor, for the response to cocaine as well as identifying further relevant loci.

The zebra finch (Taeniopygia guttata), a representative oscine songbird species, has been widely studied to investigate behavioral neuroscience, most notably the neurobiological basis of vocal learning, a rare trait shared in only a few animal groups including humans. In 2019, an updated zebra finch genome annotation (bTaeGut1_v1.p) was released from the Ensembl database and is substantially more comprehensive than the first version published in 2010. In this study, we utilized the publicly available RNA-seq data generated from Illumina-based short-reads and PacBio single-molecule real-time (SMRT) long-reads to assess the bird transcriptome. To analyze the high-throughput RNA-seq data, we adopted a hybrid bioinformatic approach combining short and long-read pipelines. From our analysis, we added 220 novel genes and 8,134 transcript variants to the Ensembl annotation, and predicted a new proteome based on the refined annotation. We further validated 18 different novel proteins by using mass-spectrometry data generated from zebra finch caudal telencephalon tissue. Our results provide additional resources for future studies of zebra finches utilizing this improved bird genome annotation and proteome.

Human stem-cell-derived models provide the promise of accelerating our understanding of brain disorders, but not knowing whether they possess the ability to mature beyond mid- to late-fetal stages potentially limits their utility. We leveraged a directed differentiation protocol to comprehensively assess maturation in vitro. Based on genome-wide analysis of the epigenetic clock and transcriptomics, as well as RNA editing, we observe that three-dimensional human cortical organoids reach postnatal stages between 250 and 300 days, a timeline paralleling in vivo development. We demonstrate the presence of several known developmental milestones, including switches in the histone deacetylase complex and NMDA receptor subunits, which we confirm at the protein and physiological levels. These results suggest that important components of an intrinsic in vivo developmental program persist in vitro. We further map neurodevelopmental and neurodegenerative disease risk genes onto in vitro gene expression trajectories to provide a resource and webtool (Gene Expression in Cortical Organoids, GECO) to guide disease modeling.

Alternative splicing is an RNA processing mechanism that affects most genes in human, contributing to disease mechanisms and phenotypic diversity. The regulation of splicing involves an intricate network of cis-regulatory elements and trans-acting factors. Due to their high sequence specificity, cis-regulation of splicing can be altered by genetic variants, significantly affecting splicing outcomes. Recently, multiple methods have been applied to understanding the regulatory effects of genetic variants on splicing. However, it is still challenging to go beyond apparent association to pinpoint functional variants. To fill in this gap, we utilized large-scale data sets of the Genotype-Tissue Expression (GTEx) project to study genetically modulated alternative splicing (GMAS) via identification of allele-specific splicing events. We demonstrate that GMAS events are shared across tissues and individuals more often than expected by chance, consistent with their genetically driven nature. Moreover, although the allelic bias of GMAS exons varies across samples, the degree of variation is similar across tissues versus individuals. Thus, genetic background drives the GMAS pattern to a similar degree as tissue-specific splicing mechanisms. Leveraging the genetically driven nature of GMAS, we developed a new method to predict functional splicing-altering variants, built upon a genotype-phenotype concordance model across samples. Complemented by experimental validations, this method predicted >1000 functional variants, many of which may alter RNA-protein interactions. Lastly, 72% of GMAS-associated SNPs were in linkage disequilibrium with GWAS-reported SNPs, and such association was enriched in tissues of relevance for specific traits/diseases. Our study enables a comprehensive view of genetically driven splicing variations in human tissues.

Single-cell RNA sequencing (scRNA-seq) data contain rich information at the gene, transcript, and nucleotide levels. Most analyses of scRNA-seq have focused on gene expression profiles, and it remains challenging to extract nucleotide variants and isoform-specific information. Here, we present scAllele, an integrative approach that detects single-nucleotide variants, insertions, deletions, and their allelic linkage with splicing patterns in scRNA-seq. We demonstrate that scAllele achieves better performance in identifying nucleotide variants than other commonly used tools. In addition, the read-specific variant calls by scAllele enables allele-specific splicing analysis, a unique feature not afforded by other methods. Applied to a lung cancer scRNA-seq dataset, scAllele identified variants with strong allelic linkage to alternative splicing, some of which are cancer specific and enriched in cancer-relevant pathways. scAllele represents a versatile tool to uncover multilayer information and previously unidentified biological insights from scRNA-seq data.

Using third-generation sequencers, long-read RNA-seq is increasingly applied in transcriptomic studies given its major advantage in characterizing full-length transcripts. A number of methods have been developed to analyze this new type of data for transcript isoforms and their abundance. Another application, which is significantly under-explored, is to identify and analyze single nucleotide variants (SNVs) in the RNA. Identification of SNVs, such as genetic mutations or RNA editing sites, is fundamental to many biomedical questions. In long-read RNA-seq, SNV analysis presents significant challenges, due to the well-known relatively high error rates of the third-generation sequencers. Here, we present the first study to detect and analyze RNA editing sites in long-read RNA-seq. Our new method, L-GIREMI, effectively handles sequencing errors and biases in the reads, and uses a model-based approach to score RNA editing sites. Applied to PacBio long-read RNA-seq data, L-GIREMI affords a high accuracy in RNA editing identification. In addition, the unique advantage of long reads allowed us to uncover novel insights about RNA editing occurrences in single molecules and double-stranded RNA (dsRNA) structures. L-GIREMI provides a valuable means to study RNA nucleotide variants in long-read RNA-seq.

Post-transcriptional regulation plays a key role in transcriptome reprogramming during cardiac pathogenesis. In previous studies, we have identified that cardiac enriched RNA-binding protein, RBFox1 plays key role in cardiac hypertrophy through mRNA alternative splicing regulation in nuclei. However, RBFox1 gene also generates a cytosolic isoform (RBFox1c), suggesting additional functions of post-transcriptional regulation in heart. In adult heart, RBFox1c mRNA constituted ~ 40% of total RBFox1 level but was significantly repressed in pressure-overloaded failing mouse heart. Using CRISPR-Cas9 technology, we have established an isoform specific RBFox1c-cKO mouse. At baseline inactivation of RBFox1c led to decreased cardiac function along with induction of cardiac fibrosis. RBFox1c-cKO mice also showed macrophages infiltration into myocardium post 7days MI. In contrast, restoration of RBFox1c expression in adult intact hearts significantly reduced cardiac fibrosis post stress. RNA-seq analyses in RBFox1c expressing cardiomyocytes showed that RBFox1c specifically suppressed the expression of pro-inflammatory genes. Secondly, CLIP-Seq analysis and targeted RNA-IP showed that RBFox1c could directly interact with inflammatory pathway mRNAs. These results suggested the inflammatory mRNAs are direct downstream targets regulated by RBFox1c. Using both in vitro cultured cardiomyocytes and intact mouse hearts, we demonstrated that expression of RBFox1c reduces pro-inflammatory mRNA expression at baseline and upon hypertrophy stimulation. Lastly, we characterized the interactome of RBFox1c through proteomic analysis and found RBFox1c specifically interacted with a component of the RNA NMD machinery-Upf1. RBFox1c interaction with Upf1 in cardiomyocytes was diminished upon hypertrophic stress. Furthermore, by inactivation of Upf1 via siRNA, we demonstrated that RBFox1c mediated repression of proinflammatory genes was Upf1 dependent.RBFox1 regulates cardiac transcriptome reprogramming in two post-transcriptional processes via distinct isoforms. While the RBFox1n regulates RNA splicing, the RBFox1c functions through targeted mRNA repression of proinflammatory genes by recruitment of Upf1 mediated RNA degradation.

Lipin 1 regulates cellular lipid homeostasis through roles in glycerolipid synthesis (through phosphatidic acid phosphatase activity) and transcriptional coactivation. Lipin 1-deficient individuals exhibit episodic disease symptoms that are triggered by metabolic stress, such as stress caused by prolonged fasting. We sought to identify critical lipin 1 activities during fasting. We determined that lipin 1 deficiency induces widespread alternative mRNA splicing in liver during fasting, much of which is normalized by refeeding. The role of lipin 1 in mRNA splicing was largely independent of its enzymatic function. We identified interactions between lipin 1 and spliceosome proteins, as well as a requirement for lipin 1 to maintain homeostatic levels of spliceosome small nuclear RNAs and specific RNA splicing factors. In fasted Lpin1-/- liver, we identified a correspondence between alternative splicing of phospholipid biosynthetic enzymes and dysregulated phospholipid levels; splicing patterns and phospholipid levels were partly normalized by feeding. Thus, lipin 1 influences hepatic lipid metabolism through mRNA splicing, as well as through enzymatic and transcriptional activities, and fasting exacerbates the deleterious effects of lipin 1 deficiency on metabolic homeostasis.

Transcriptomic analyses of postmortem brains have begun to elucidate molecular abnormalities in autism spectrum disorder (ASD). However, a crucial pathway involved in synaptic development, RNA editing, has not yet been studied on a genome-wide scale. Here we profiled global patterns of adenosine-to-inosine (A-to-I) editing in a large cohort of postmortem brains of people with ASD. We observed a global bias for hypoediting in ASD brains, which was shared across brain regions and involved many synaptic genes. We show that the Fragile X proteins FMRP and FXR1P interact with RNA-editing enzymes (ADAR proteins) and modulate A-to-I editing. Furthermore, we observed convergent patterns of RNA-editing alterations in ASD and Fragile X syndrome, establishing this as a molecular link between these related diseases. Our findings, which are corroborated across multiple data sets, including dup15q (genomic duplication of 15q11.2-13.1) cases associated with intellectual disability, highlight RNA-editing dysregulation in ASD and reveal new mechanisms underlying this disorder.

BackgroundRNA editing generates modifications to the RNA sequences, thereby increasing protein diversity and shaping various layers of gene regulation. Recent studies have revealed global shifts in editing levels across many cancer types, as well as a few specific mechanisms implicating individual sites in tumorigenesis or metastasis. However, most tumor-associated sites, predominantly in noncoding regions, have unknown functional relevance.ResultsHere, we carry out integrative analysis of RNA editing profiles between epithelial and mesenchymal tumors, since epithelial-mesenchymal transition is a key paradigm for metastasis. We identify distinct editing patterns between epithelial and mesenchymal tumors in seven cancer types using TCGA data, an observation further supported by single-cell RNA sequencing data and ADAR perturbation experiments in cell culture. Through computational analyses and experimental validations, we show that differential editing sites between epithelial and mesenchymal phenotypes function by regulating mRNA abundance of their respective genes. Our analysis of RNA-binding proteins reveals ILF3 as a potential regulator of this process, supported by experimental validations. Consistent with the known roles of ILF3 in immune response, epithelial-mesenchymal differential editing sites are enriched in genes involved in immune and viral processes. The strongest target of editing-dependent ILF3 regulation is the transcript encoding PKR, a crucial player in immune and viral response.ConclusionsOur study reports widespread differences in RNA editing between epithelial and mesenchymal tumors and a novel mechanism of editing-dependent regulation of mRNA abundance. It reveals the broad impact of RNA editing in cancer and its relevance to cancer-related immune pathways.

In the mammalian heart, the left ventricle (LV) rapidly becomes more dominant in size and function over the right ventricle (RV) after birth. The molecular regulators responsible for this chamber-specific differential growth are largely unknown. We found that cardiomyocytes in the neonatal mouse RV had lower proliferation, more apoptosis, and a smaller average size compared with the LV. This chamber-specific growth pattern was associated with a selective activation of p38 mitogen-activated protein kinase (MAPK) activity in the RV and simultaneous inactivation in the LV. Cardiomyocyte-specific deletion of both the Mapk14 and Mapk11 genes in mice resulted in loss of p38 MAPK expression and activity in the neonatal heart. Inactivation of p38 activity led to a marked increase in cardiomyocyte proliferation and hypertrophy but diminished cardiomyocyte apoptosis, specifically in the RV. Consequently, the p38-inactivated hearts showed RV-specific enlargement postnatally, progressing to pulmonary hypertension and right heart failure at the adult stage. Chamber-specific p38 activity was associated with differential expression of dual-specific phosphatases (DUSPs) in neonatal hearts, including DUSP26. Unbiased transcriptome analysis revealed that IRE1α/XBP1-mediated gene regulation contributed to p38 MAPK-dependent regulation of neonatal cardiomyocyte proliferation and binucleation. These findings establish an obligatory role of DUSP/p38/IRE1α signaling in cardiomyocytes for chamber-specific growth in the postnatal heart.

Alternative splicing is an RNA processing mechanism that affects most genes in human, contributing to disease mechanisms and phenotypic diversity. The regulation of splicing involves an intricate network of cis-regulatory elements and trans-acting factors. Due to their high sequence specificity, cis-regulation of splicing can be altered by genetic variants, significantly affecting splicing outcomes. Recently, multiple methods have been applied to understanding the regulatory effects of genetic variants on splicing. However, it is still challenging to go beyond apparent association to pinpoint functional variants. To fill in this gap, we utilized large-scale datasets of the Genotype-Tissue Expression (GTEx) project to study genetically-modulated alternative splicing (GMAS) via identification of allele-specific splicing events. We demonstrate that GMAS events are shared across tissues and individuals more often than expected by chance, consistent with their genetically driven nature. Moreover, although the allelic bias of GMAS exons varies across samples, the degree of variation is similar across tissues vs. individuals. Thus, genetic background drives the GMAS pattern to a similar degree as tissue-specific splicing mechanisms. Leveraging the genetically driven nature of GMAS, we developed a new method to predict functional splicing-altering variants, built upon a genotype-phenotype concordance model across samples. Complemented by experimental validations, this method predicted >1000 functional variants, many of which may alter RNA-protein interactions. Lastly, 72% of GMAS-associated SNPs were in linkage disequilibrium with GWAS-reported SNPs, and such association was enriched in tissues of relevance for specific traits/diseases. Our study enables a comprehensive view of genetically driven splicing variations in human tissues.

Motivation RNA-sequencing (RNA-seq) enables global identification of RNA-editing sites in biological systems and disease. A salient step in many studies is to identify editing sites that statistically associate with treatment (e.g. case versus control) or covary with biological factors, such as age. However, RNA-seq has technical features that incumbent tests (e.g. t-test and linear regression) do not consider, which can lead to false positives and false negatives. Results In this study, we demonstrate the limitations of currently used tests and introduce the method, RNA-editing tests (REDITs), a suite of tests that employ beta-binomial models to identify differential RNA editing. The tests in REDITs have higher sensitivity than other tests, while also maintaining the type I error (false positive) rate at the nominal level. Applied to the GTEx dataset, we unveil RNA-editing changes associated with age and gender, and differential recoding profiles between brain regions. Availability and implementation REDITs are implemented as functions in R and freely available for download at https://github.com/gxiaolab/REDITs. The repository also provides a code example for leveraging parallelization using multiple cores.

microRNAs (miRNAs) are small non-coding RNAs that play critical roles in gene regulation. The presence of miRNAs in extracellular biofluids is increasingly recognized. However, most previous characterization of extracellular miRNAs focused on their overall expression levels. Alternative sequence isoforms and modifications of miRNAs were rarely considered in the extracellular space. Here, we developed a highly accurate bioinformatic method, called miNTA, to identify 3’ non-templated additions (NTAs) of miRNAs using small RNA-sequencing data. Using miNTA, we conducted an in-depth analysis of miRNA 3’ NTA profiles in 1047 extracellular RNA-sequencing data sets of 4 types of biofluids. This analysis identified hundreds of miRNAs with 3’ uridylation or adenylation, with the former being more prevalent. Among these miRNAs, up to 53% (22%) had an average 3’ uridylation (adenylation) level of at least 10% in a specific biofluid. Strikingly, we found that 3’ uridylation levels enabled segregation of different types of biofluids, more effectively than overall miRNA expression levels. This observation suggests that 3’ NTA levels possess fluid-specific information relatively robust to batch effects. In addition, we observed that extracellular miRNAs with 3’ uridylations are enriched in processes related to angiogenesis, apoptosis, and inflammatory response, and this type of modification may stabilize base-pairing between miRNAs and their target genes. Together, our study provides a comprehensive landscape of miRNA NTAs in human biofluids, which paves way for further biomarker discoveries. The insights generated in our work built a foundation for future functional, mechanistic, and translational discoveries.

Recent studies revealed global shifts in RNA editing, the modification of RNA sequences, across many cancers. Besides a few sites implicated in tumorigenesis or metastasis, most tumor-associated sites, predominantly in noncoding regions, have unknown function. Here, we characterize editing profiles between epithelial (E) and mesenchymal (M) phenotypes in seven cancer types, as epithelial-mesenchymal transition (EMT) is a key paradigm for metastasis. We observe distinct editing patterns between E and M tumors and EMT induction upon loss of ADAR enzymes in cultured cells. E-M differential sites are highly enriched in genes involved in immune and viral processes, some of which regulate mRNA abundance of their respective genes. We identify a novel mechanism in which ILF3 preferentially stabilizes edited transcripts. Among editing-dependent ILF3 targets is the transcript encoding PKR, a crucial player in immune response. Our study demonstrates the broad impact of RNA editing in cancer and relevance of editing to cancer-related immune pathways.

The essential biological cofactor heme is synthesized in cells in the Fe(II) form. Oxidized Fe(III) heme is specifically required for processing primary transcripts of microRNAs (pri-miRNAs) by the RNA-binding protein DGCR8, a core component of the Microprocessor complex. It is unknown how readily available Fe(III) heme is in the largely reducing environment in human cells and how changes in cellular Fe(III) heme availability alter microRNA (miRNA) expression. Here we address the first question by characterizing DGCR8 mutants with various degrees of deficiency in heme-binding. We observed a strikingly simple correlation between Fe(III) heme affinityin vitroand the Microprocessor activity in HeLa cells, with the heme affinity threshold for activation estimated to be between 0.6-5 pM under typical cell culture conditions. The threshold is strongly influenced by cellular heme synthesis and uptake. We suggest that the threshold reflects a labile Fe(III) heme pool in cells. Based on our understanding of DGCR8 mutants, we reanalyzed recently reported miRNA sequencing data and conclude that heme is generally required for processing canonical pri-miRNAs, that heme modulates the specificity of Microprocessor, and that cellular heme level and differential DGCR8 heme occupancy alter the expression of distinct groups of miRNAs in a hierarchical fashion. Overall, our study provides the first glimpse of a labile Fe(III) heme pool important for a fundamental physiological function and reveal principles governing how Fe(III) heme modulates miRNA maturation at a genomic scale. We also discuss potential states and biological significance of the labile Fe(III) heme pool.

Alcohol (ethanol, EtOH) consumption during pregnancy can result in fetal alcohol spectrum disorders (FASDs), which are characterized by prenatal and postnatal growth restriction and craniofacial dysmorphology. Recently, cell-derived extracellular vesicles, including exosomes and microvesicles containing several species of RNAs (exRNAs), have emerged as a mechanism of cell-to-cell communication. However, EtOH’s effects on the biogenesis and function of non-coding exRNAs during fetal development have not been explored. Therefore, we studied the effects of maternal EtOH exposure on the composition of exosomal RNAs in the amniotic fluid (AF) using rat fetal alcohol exposure (FAE) model. Through RNA-Seq analysis we identified and verified AF exosomal miRNAs with differential expression levels specifically associated with maternal EtOH exposure. Uptake of purified FAE AF exosomes by rBMSCs resulted in significant alteration of molecular markers associated with osteogenic differentiation of rBMSCs. We also determined putative functional roles for AF exosomal miRNAs (miR-199a-3p, miR-214-3p and let-7g) that are dysregulated by FAE in osteogenic differentiation of rBMSCs. Our results demonstrate that FAE alters AF exosomal miRNAs and that exosomal transfer of dysregulated miRNAs has significant molecular effects on stem cell regulation and differentiation. Our results further suggest the usefulness of assessing molecular alterations in AF exRNAs to study the mechanisms of FAE teratogenesis that should be further investigated by using an in vivo model.

BACKGROUND It was recently discovered that abundant and stable extracellular RNA (exRNA) species exist in bodily fluids. Saliva is an emerging biofluid for biomarker development for noninvasive detection and screening of local and systemic diseases. Use of RNA-Sequencing (RNA-Seq) to profile exRNA is rapidly growing; however, no single preparation and analysis protocol can be used for all biofluids. Specifically, RNA-Seq of saliva is particularly challenging owing to high abundance of bacterial contents and low abundance of salivary exRNA. Given the laborious procedures needed for RNA-Seq library construction, sequencing, data storage, and data analysis, saliva-specific and optimized protocols are essential. METHODS We compared different RNA isolation methods and library construction kits for long and small RNA sequencing. The role of ribosomal RNA (rRNA) depletion also was evaluated. RESULTS The miRNeasy Micro Kit (Qiagen) showed the highest total RNA yield (70.8 ng/mL cell-free saliva) and best small RNA recovery, and the NEBNext library preparation kits resulted in the highest number of detected human genes [5649–6813 at 1 reads per kilobase RNA per million mapped (RPKM)] and small RNAs [482–696 microRNAs (miRNAs) and 190–214 other small RNAs]. The proportion of human RNA-Seq reads was much higher in rRNA-depleted saliva samples (41%) than in samples without rRNA depletion (14%). In addition, the transfer RNA (tRNA)-derived RNA fragments (tRFs), a novel class of small RNAs, were highly abundant in human saliva, specifically tRF-4 (4%) and tRF-5 (15.25%). CONCLUSIONS Our results may help in selection of the best adapted methods of RNA isolation and small and long RNA library constructions for salivary exRNA studies.