Mouse Genome Research Articles

Methylation is maintained specifically at imprinting control regions but not other DMRs associated with imprinted genes in mice bearing a mutation in the Dnmt1 intrinsically disordered domain.

Differential methylation of imprinting control regions in mammals is essential for distinguishing the parental alleles from each other and regulating their expression accordingly. To ensure parent of origin-specific expression of imprinted genes and thereby normal developmental progression, the differentially methylated states that are inherited at fertilization must be stably maintained by DNA methyltransferase 1 throughout subsequent somatic cell division. Further epigenetic modifications, such as the acquisition of secondary regions of differential methylation, are dependent on the methylation status of imprinting control regions and are important for achieving the monoallelic expression of imprinted genes, but little is known about how imprinting control regions direct the acquisition and maintenance of methylation at these secondary sites. Recent analysis has identified mutations that reduce DNA methyltransferase 1 fidelity at some genomic sequences but not at others, suggesting that it may function differently at different loci. We examined the impact of the mutant DNA methyltransferase 1 P allele on methylation at imprinting control regions as well as at secondary differentially methylated regions and non-imprinted sequences. We found that while the P allele results in a major reduction in DNA methylation levels across the mouse genome, methylation is specifically maintained at imprinting control regions but not at their corresponding secondary DMRs. This result suggests that DNA methyltransferase 1 may work differently at imprinting control regions or that there is an alternate mechanism for maintaining methylation at these critical regulatory regions and that maintenance of methylation at secondary DMRs is not solely dependent on the methylation status of the ICR.

DeepSATA: A Deep Learning-Based Sequence Analyzer Incorporating the Transcription Factor Binding Affinity to Dissect the Effects of Non-Coding Genetic Variants.

Utilizing large-scale epigenomics data, deep learning tools can predict the regulatory activity of genomic sequences, annotate non-coding genetic variants, and uncover mechanisms behind complex traits. However, these tools primarily rely on human or mouse data for training, limiting their performance when applied to other species. Furthermore, the limited exploration of many species, particularly in the case of livestock, has led to a scarcity of comprehensive and high-quality epigenetic data, posing challenges in developing reliable deep learning models for decoding their non-coding genomes. The cross-species prediction of the regulatory genome can be achieved by leveraging publicly available data from extensively studied organisms and making use of the conserved DNA binding preferences of transcription factors within the same tissue. In this study, we introduced DeepSATA, a novel deep learning-based sequence analyzer that incorporates the transcription factor binding affinity for the cross-species prediction of chromatin accessibility. By applying DeepSATA to analyze the genomes of pigs, chickens, cattle, humans, and mice, we demonstrated its ability to improve the prediction accuracy of chromatin accessibility and achieve reliable cross-species predictions in animals. Additionally, we showcased its effectiveness in analyzing pig genetic variants associated with economic traits and in increasing the accuracy of genomic predictions. Overall, our study presents a valuable tool to explore the epigenomic landscape of various species and pinpoint regulatory deoxyribonucleic acid (DNA) variants associated with complex traits.

High-fat diet plus HNF1A variant promotes polyps by activating β-catenin in early-onset colorectal cancer.

The incidence of early-onset colorectal cancer (EO-CRC) is rising and is poorly understood. Lifestyle factors and altered genetic background possibly contribute. Here, we performed targeted exon sequencing of archived leukocyte DNA from 158 EO-CRC participants, which identified a missense mutation at p.A98V within the proximal DNA binding domain of Hepatic Nuclear Factor 1 α (HNF1AA98V, rs1800574). The HNF1AA98V exhibited reduced DNA binding. To test function, the HNF1A variant was introduced into the mouse genome by CRISPR/Cas9, and the mice were placed on either a high-fat diet (HFD) or high-sugar diet (HSD). Only 1% of the HNF1A mutant mice developed polyps on normal chow; however, 19% and 3% developed polyps on the HFD and HSD, respectively. RNA-Seq revealed an increase in metabolic, immune, lipid biogenesis genes, and Wnt/β-catenin signaling components in the HNF1A mutant relative to the WT mice. Mouse polyps and colon cancers from participants carrying the HNF1AA98V variant exhibited reduced CDX2 and elevated β-catenin proteins. We further demonstrated decreased occupancy of HNF1AA98V at the Cdx2 locus and reduced Cdx2 promoter activity compared with WT HNF1A. Collectively, our study shows that the HNF1AA98V variant plus a HFD promotes the formation of colonic polyps by activating β-catenin via decreasing Cdx2 expression.

Broad H3K4me3 domains: Maintaining cellular identity and their implication in super-enhancer hijacking.

The human and mouse genomes are complex from a genomic standpoint. Each cell has the same genomic sequence, yet a wide array of cell types exists due to the presence of a plethora of regulatory elements in the non-coding genome. Recent advances in epigenomic profiling have uncovered non-coding gene proximal promoters and distal enhancers of transcription genome-wide. Extension of promoter-associated H3K4me3 histone mark across the gene body, known as a broad H3K4me3 domain (H3K4me3-BD), is a signature of constitutive expression of cell-type-specific regulation and of tumour suppressor genes in healthy cells. Recently, it has been discovered that the presence of H3K4me3-BDs over oncogenes is a cancer-specific feature associated with their dysregulated gene expression and tumourigenesis. Moreover, it has been shown that the hijacking of clusters of enhancers, known as super-enhancers (SE), by proto-oncogenes results in the presence of H3K4me3-BDs over the gene body. Therefore, H3K4me3-BDs and SE crosstalk in healthy and cancer cells therefore represents an important mechanism to identify future treatments for patients with SE driven cancers.

#6741 TRANSCRIPTION FACTOR ATF3 EXACERBATES RENAL FIBROSIS THROUGH RECRUITING HISTONE ACETYLASES

Abstract Background and Aims Chronic kidney disease (CKD) threatens public health around the world. Renal fibrosis is a final common pathology in renal diseases of different etiologies. However, no efficient, broadly applicable anti-fibrotic therapies exist. Histone acetylation is an epigenetic modification involved in the regulation of multiple disease-specific gene expressions, balanced by histone acetyltransferases and histone deacetylase (HDAC) enzymes. Transcription factors (TF) could recruit histone-modifying enzymes and alter gene expression. But now, much remains unknown about the role of histone acetylation and whether there exits the regulator of it in fibrosis of CKD. Here, we for the first time explored a novel aspect of ATF3 promoting kidney fibrosis via recruiting the histone acetylases to fibrosis-related genes in CKD. Method We constructed a CKD mouse model by using adenine and oxonate. Serum were taken for the measurement of BUN and creatinine. Kidneys were harvested and processed for histological, sequencing, qRT-PCR, and western blotting analysis. RNA-seq and ChIP-seq (antibody: H3K27ac) of mouse kidneys were conducted for profiling the differential gene expression and H3K27ac. GO and KEGG analysis was performed for annotation of the function. Motif analysis was applied for searching the regulator of H3K27ac. Biopsy samples of patients with common CKD were used for immunohistochemical analysis. ATF3-KO mice were also conducted for defining the function of ATF3 in CKD. To profile the ATF3 binding site on chromatin, ChIP-seq of ATF3 in mice kidneys of normal control and CKD were also performed. Besides, the data contained ATF3 interact protein in databases (ChIP-atlas and BioGrid) and the results of protein interaction prediction were integrated for finding the protein-protein interaction of ATF3. TGF-β- stimulated TCMK-1 cell served as the cell model for further mechanism validation. CO-IP in cells clarifies that ATF3 could recruit the histone acetyltransferases. ChIP-qPCR of the fibrosis gene was performed to further verify the mechanism of ATF3-regulated gene expression. Results Firstly, compared with the normal control, histone modification H3K27ac repertoire was changed in chronic kidney disease. 11859 H3K27ac peaks specific to CKD, and 14996 H3K27ac peaks specific to healthy mice kidneys were found. The binding intensity of these regions has altered dramatically in the kidney of CKD. Pearson correlation analysis and function annotation showed that CKD-specific changes of H3K27ac were associated with fibrosis gene expression. Motif analysis showed that ATF3 in the top 10 TFs that may interact with H3K27ac. Among those TFs, ATF3 elicited the highest expression. Besides, ATF3 could co-localize with H3K27ac along the mice genome globally. By immunohistochemical staining analysis, we observed the upregulation of ATF3 in the kidneys of patients with biopsy-proven CKD. We found that knocking out the ATF3 could significantly reduce kidney damage in CKD mice. In parallel with improved morphological injuries, the level of kidney fibrosis gene expression was also significantly reduced in ATF3–/– mice. Furthermore, the H3K27ac on the mice chromatin altered along with the knockout of ATF3, and several fibrosis-related gene expressions which elevated in CKD were decreased by deficiency of ATF3. On the region of gene cd2, H3K27ac disappeared with the knockout of ATF3 in CKD. We found that ATF3 could interact with histone acetyltransferases (CBP, KAT7, P300) to promote the H3K27ac. Further evidenced by ChIP-qPCR of fibrosis gene cd2, we found that ATF3 could recruit the histone acetyltransferases on the fibrosis gene to promote the H3K27ac to provoke the gene expression in the cell model. Conclusion We profiled the landscape of H3K27ac in CKD, explored the epigenetic regulator of H3K27ac, and revealed that ATF3 promotes renal fibrosis in CKD. Notably, we for the first time explored a novel mechanism of ATF3 about recruiting histone acetyltransferases to regulate the H3K27ac on fibrosis gene. Our data highlighted that ATF3 might represent a potential therapeutic target against fibrotic kidney diseases.

MDB-30. IDENTIFICATION AND CHARACTERIZATION OF NOVEL TRANSCRIBED ULTRACONSERVED REGIONS AS LONG NON-CODING RNAS IN MEDULLOBLASTOMA

Abstract Medulloblastoma is the most common pediatric brain tumor. While the 5-year survival rate for medulloblastoma patients has improved with molecular subgrouping, children who undergo standard of care treatments frequently experience permanent neurological deficits. Additionally, under-characterized medulloblastoma subgroups lack targeted therapies, correlating to lower survival rates for these patients. To find new therapies, a better understanding of the molecular basis of medulloblastoma is needed. This work investigates the role of “Transcribed Ultra-Conserved Regions” (TUCRs) in medulloblastoma about which there is no published literature. TUCRs represent a distinct group of 481 putative lncRNAs that are 100% conserved across human, mouse, and rat genomes. These ultra-conserved elements change 20 times slower than the average genomic region and are therefore under intense negative selection within multiple species. Since their discovery, TUCRs have only been correlated to a handful of cancers and remain exceedingly underexamined within cancer research. Using RNA-sequencing and differential gene expression analysis of medulloblastoma tumors as compared to normal human cerebellum samples, we identified a multitude of differentially expressed TUCRs spanning across all TUCR genomic annotations. In vitro experimentation using several medulloblastoma established cell lines indicated significant upregulation of these TUCR transcripts. We are currently investigating the functional effects and mechanisms of action of select differentially expressed TUCRs in medulloblastoma. Altogether, our work is investigating the oncogenic and tumor suppressive nature of TUCRs. It represents the first study of TUCRs in medulloblastoma that will significantly advance our basic understanding of this highly understudied class of molecules.

Comparative analysis of bats and rodents’ genomes suggests a relation between non-LTR retrotransposons, cancer incidence, and ageing

The presence in nature of species showing drastic differences in lifespan and cancer incidence has recently increased the interest of the scientific community. In particular, the adaptations and the genomic features underlying the evolution of cancer-resistant and long-lived organisms have recently focused on transposable elements (TEs). In this study, we compared the content and dynamics of TE activity in the genomes of four rodent and six bat species exhibiting different lifespans and cancer susceptibility. Mouse, rat, and guinea pig genomes (short-lived and cancer-prone organisms) were compared with that of naked mole rat (Heterocephalus glaber) which is a cancer-resistant organism and the rodent with the longest lifespan. The long-lived bats of the genera Myotis, Rhinolophus, Pteropus and Rousettus were instead compared with Molossus molossus, which is one of the organisms with the shortest lifespan among the order Chiroptera. Despite previous hypotheses stating a substantial tolerance of TEs in bats, we found that long-lived bats and the naked mole rat share a marked decrease of non-LTR retrotransposons (LINEs and SINEs) accumulation in recent evolutionary times.

Characterization of Unidirectional Replication Forks in the Mouse Genome.

Origins of replication are genomic regions in which replication initiates in a bidirectional manner. Recently, a new methodology (origin-derived single-stranded DNA sequencing; ori-SSDS) was developed that allows the detection of replication initiation in a strand-specific manner. Reanalysis of the strand-specific data revealed that 18-33% of the peaks are non-symmetrical, suggesting a single direction of replication. Analysis of replication fork direction data revealed that these are origins of replication in which the replication is paused in one of the directions, probably due to the existence of a replication fork barrier. Analysis of the unidirectional origins revealed a preference of G4 quadruplexes for the blocked leading strand. Taken together, our analysis identified hundreds of genomic locations in which the replication initiates only in one direction, and suggests that G4 quadruplexes may serve as replication fork barriers in such places.

Genome-wide identification of endogenous retrovirus elements and their active transcription in mink genome.

Mammalian endogenous retroviruses (ERVs) are ancient retroviruses that have been integrated into genomes. ERVs were believed to be inactive until the discovery of ERV transcription in the mouse genome. However, the transcription level and function of ERV elements in mammalian genomes are not well understood. In this study, we performed the first genome-wide scanning of ERV loci in the American mink (Neogale vison) genome (NeoERV) followed by transcriptomic analysis to detect actively transcribed NeoERV elements. A total of 365,791 NeoERV loci were identified, and161,205 (44%) of these loci were found to be actively transcribed based on transcriptomic data from three types of tissues (amygdala, trachea and lung). More than one third of the actively transcribed NeoERV loci were tissue-specific. Furthermore, some of the active loci were associated with host gene transcription, and the level of NeoERV transcription was positively correlated with that of host genes, specifically when active loci were located in overlapped gene regions. An in-depth analysis of the envelope protein coding env gene showed that, in general, its transcription level was higher than that of NeoERVs, which is believed to be associated with host immunity.

Mouse B2 SINE elements function as IFN-inducible enhancers.

Regulatory networks underlying innate immunity continually face selective pressures to adapt to new and evolving pathogens. Transposable elements (TEs) can affect immune gene expression as a source of inducible regulatory elements, but the significance of these elements in facilitating evolutionary diversification of innate immunity remains largely unexplored. Here, we investigated the mouse epigenomic response to type II interferon (IFN) signaling and discovered that elements from a subfamily of B2 SINE (B2_Mm2) contain STAT1 binding sites and function as IFN-inducible enhancers. CRISPR deletion experiments in mouse cells demonstrated that a B2_Mm2 element has been co-opted as an enhancer driving IFN-inducible expression of Dicer1. The rodent-specific B2 SINE family is highly abundant in the mouse genome and elements have been previously characterized to exhibit promoter, insulator, and non-coding RNA activity. Our work establishes a new role for B2 elements as inducible enhancer elements that influence mouse immunity, and exemplifies how lineage-specific TEs can facilitate evolutionary turnover and divergence of innate immune regulatory networks.

Discovery and characterization of novel Cre-type tyrosine site-specific recombinases for advanced genome engineering.

Tyrosine-type site-specific recombinases (Y-SSRs) are versatile tools for genome engineering due to their ability to mediate excision, integration, inversion and exchange of genomic DNA with single nucleotide precision. The ever-increasing need for sophisticated genome engineering is driving efforts to identify novel SSR systems with intrinsic properties more suitable for particular applications. In this work, we develop a systematic computational workflow for annotation of putative Y-SSR systems and apply this pipeline to identify and characterize eight new naturally occurring Cre-type SSR systems. We test their activity in bacterial and mammalian cells and establish selectivity profiles for the new and already established Cre-type SSRs with regard to their ability to mutually recombine their target sites. These data form the basis for sophisticated genome engineering experiments using combinations of Y-SSRs in research fields including advanced genomics and synthetic biology. Finally, we identify putative pseudo-sites and potential off-targets for Y-SSRs in the human and mouse genome. Together with established methods for altering the DNA-binding specificity of this class of enzymes, this work should facilitate the use of Y-SSRs for future genome surgery applications.

CopMEM2: Robust and scalable maximum exact match finding.

Finding Maximum Exact Matches, i.e., matches between two strings that cannot be further extended to the left or right, is a classic string problem with applications in genome-to-genome comparisons. The existing tools rarely explicitly address the problem of MEM finding for a pair of very similar genomes, which may be computationally challenging. We present copMEM2, a multi-threaded implementation of its predecessor. Together with a few optimizations, including a carefully built predecessor query data structure and sort procedure selection, and taking care for highly similar data, copMEM2 allows to compute all MEMs of minimum length 50 between the human and mouse genomes in 59 s, using 10.40 GB of RAM and 12 threads, being at least a few times faster than its main contenders. On a pair of human genomes, hg18 and hg19, the results are 324 s and 16.57 GB, respectively. copMEM2 is available at https://github.com/wbieniec/copmem2. Supplementary data are available at Bioinformatics online.

Abstract 157: Inkiln is a Novel Long Noncoding RNA Promoting Vascular Smooth Muscle Inflammation Via Scaffolding Mkl1 and Usp10

Activation of vascular smooth muscle cells (VSMCs) inflammation is critical in initiation and progression of vascular disease. However, the role of human-specific long noncoding RNAs (lncRNAs) in VSMC inflammation is largely unknown. We performed Bulk RNA-seq in differentiated human VSMCs and revealed a novel human-specific lncRNA called IN flammatory M K L1 I nteracting L ong N oncoding RNA ( INKILN ). INKILN is barely expressed in contractile VSMCs or healthy vessels and induced in human atherosclerosis and abdominal aortic aneurysm. INKILN is transcriptionally activated through p65/NF-κB pathway. INKILN activates proinflammatory gene expression in cultured human VSMCs and in ex vivo cultured vessels. Mechanistically, INKILN physically binds to and stabilizes MKL1 protein, a key activator of VSMC inflammation through the p65/NF-κB pathway. In the IL1β-activated VSMCs, depletion of INKILN blocks the nuclear localization of both p65 and MKL1, and abolishes the physical interaction between these two proteins, resulting in the reduced transactvity of p65/NF-κB. Further, INKILN knockdown enhances MKL1 ubiquitination, likely through the reduced physical interaction with the deubiquitinating enzyme USP10. The in vivo study of INKILN has utilized a Bacterial Artificial Chromosome (BAC) INKILN -Transgenic (Tg) mouse line, due to the lack of non-conserved INKILN gene in the mouse genome. In the ligation-injured carotid arteries of BAC INKILN -Tg mice, human lncRNA INKILN is induced and has exacerbated neointimal formation. In conclusion, these findings elucidate an important pathway of VSMC inflammation involving an INKILN /MKL1/USP10 regulatory axis. Human BAC Tg mice offer a novel and physiologically relevant approach for investigating human-specific lncRNAs under vascular disease conditions.

Epithelial sodium channel (ENaC) in GtoPdb v.2023.1

Epithelial sodium channel (ENaC) in GtoPdb v.2023.1

ZAC in GtoPdb v.2023.1

The zinc-activated channel (ZAC, nomenclature as agreed by the NC-IUPHAR Subcommittee for the Zinc Activated Channel) is a member of the Cys-loop family that includes the nicotinic ACh, 5-HT3, GABAA and strychnine-sensitive glycine receptors [2, 3, 5]. The channel is likely to exist as a homopentamer of 4TM subunits that form an intrinsic cation selective channel equipermeable to Na+, K+ and Cs+, but impermeable to Ca2+ and Mg2+ [5]. ZAC displays constitutive activity that can be blocked by tubocurarine, TTFB and high concentrations of Ca2+ [5]. Although denoted ZAC, the channel is more potently activated by H+ and Cu2+, with greater and lesser efficacy than Zn2+, respectively [5]. Orthologs of the human ZACN gene are present in a wide range of mammalian genomes, but notably not in the mouse or rat genomes. [2, 3].

Lysophospholipid (S1P) receptors in GtoPdb v.2023.1

Sphingosine 1-phosphate (S1P) receptors (nomenclature as agreed by the NC-IUPHAR Subcommittee on Lysophospholipid receptors [96]) are activated by the endogenous lipid sphingosine 1-phosphate (S1P). Originally cloned as orphan members of the endothelial differentiation gene (edg) family [16, 123], the receptors are currently designated as S1P1R through S1P5R [73, 16, 123]. Their gene nomenclature has been codified as human S1PR1, S1PR2, etc. (HUGO Gene Nomenclature Committee, HGNC) and S1pr1, S1pr2, etc. for mice (Mouse Genome Informatics Database, MGI) to reflect species and receptor function. All S1P receptors (S1PRs) have been knocked-out in mice constitutively and in some cases, conditionally. S1PRs, particularly S1P1, are expressed throughout all mammalian organ systems. Ligand delivery occurs via two known carriers (or "chaperones"): albumin and HDL-bound apolipoprotein M (ApoM), the latter of which elicits biased agonist signaling by S1P1 in multiple cell types [18, 53]. The five S1PRs, two chaperones, and active cellular metabolism have complicated analyses of receptor ligand binding in native systems. Signaling pathways and physiological roles have been characterized through radioligand binding in heterologous expression systems, targeted deletion of the different S1PRs, and most recently, mouse models that report in vivo S1P1R activation [101, 103]. The structures of S1P1 [180, 69, 108, 184], S1P2 [32], S1P3[116, 187], and S1P5 [110, 185] are solved, and confirmed aspects of ligand binding, specificity, and receptor activation, determined previously through biochemical and genetic studies [69, 17]. fingolimod (FTY720), the first FDA-approved drug to target any of the lysophospholipid receptors, binds as a phosphorylated metabolite to four of the five S1PRs, and was the first oral therapy for multiple sclerosis (MS) [35]. Second-generation S1PR modulators siponimod, ozanimod, and ponesimod that target S1P1 and S1P5 are also FDA approved for the treatment of various MS forms [16, 123]. In 2021, ozanimod became the first S1PR modulator to be FDA approved for the treatment of ulcerative colitis [145]. The mechanisms of action of fingolimod and other S1PR-modulating drugs now in development include binding S1PRs in multiple organ systems, e.g., immune and nervous systems, although the precise nature of their receptor interactions requires clarification [141, 37, 63, 64].

Lysophospholipid (LPA) receptors in GtoPdb v.2023.1

Lysophosphatidic acid (LPA) receptors (nomenclature as agreed by the NC-IUPHAR Subcommittee on Lysophospholipid Receptors [62, 23, 91, 144]) are activated by the endogenous phospholipid LPA. The first receptor, LPA1, was identified as ventricular zone gene-1 (vzg-1) [46], This discovery represented the beginning of the de-orphanisation of members of the endothelial differentiation gene (edg) family, as other LPA and sphingosine 1-phosphate (S1P) receptors were found. Five additional LPA receptors (LPA2,3,4,5,6) have since been identified [91] and their gene nomenclature codified for human LPAR1, LPAR2, etc. (HUGO Gene Nomenclature Committee, HGNC) and Lpar1, Lpar2, etc. for mice (Mouse Genome Informatics Database, MGI) to reflect species and receptor function of their corresponding proteins. The crystal structure of LPA1 [17, 80, 2] and LPA6 [128] are solved and indicate that LPA accesses the extracellular binding pocket, consistent with its proposed delivery via autotaxin [17]. These studies have also implicated cross-talk with endocannabinoids via phosphorylated intermediates that can also activate these receptors. The binding affinities to LPA1 of unlabeled, natural LPA and anandamide phosphate (AEAp) were measured using backscattering interferometry (pKd = 9) [92, 115]. Utilization of this method indicated affinities that were 77-fold lower than when measured using radioactivity-based protocols [143]. Targeted deletion of LPA receptors has clarified signalling pathways and identified physiological and pathophysiological roles. Multiple groups have independently published validation of all six LPA receptors described in these tables, and further validation was achieved using a distinct read-out via a novel TGFα "shedding* assay [54]. LPA has been proposed to be a ligand for GPR35 [103], supported by a study revealing that LPA modulates macrophage function through GPR35 [60]. However chemokine (C-X-C motif) ligand 17 (CXCL17) is reported to be a ligand for GPR35/CXCR8 [85]. Moreover, LPA has also been described as an agonist for the transient receptor potential (Trp) ion channels TRPV1 [96] and TRPA1 [65]. All of these proposed non-GPCR receptor identities require confirmation and are not currently recognized as bona fide LPA receptors.

An unbiased AAV-STARR-seq screen revealing the enhancer activity map of genomic regions in the mouse brain in vivo

Enhancers are important cis-regulatory elements controlling cell-type specific expression patterns of genes. Furthermore, combinations of enhancers and minimal promoters are utilized to construct small, artificial promoters for gene delivery vectors. Large-scale functional screening methodology to construct genomic maps of enhancer activities has been successfully established in cultured cell lines, however, not yet applied to terminally differentiated cells and tissues in a living animal. Here, we transposed the Self-Transcribing Active Regulatory Region Sequencing (STARR-seq) technique to the mouse brain using adeno-associated-viruses (AAV) for the delivery of a highly complex screening library tiling entire genomic regions and covering in total 3 Mb of the mouse genome. We identified 483 sequences with enhancer activity, including sequences that were not predicted by DNA accessibility or histone marks. Characterizing the expression patterns of fluorescent reporters controlled by nine candidate sequences, we observed differential expression patterns also in sparse cell types. Together, our study provides an entry point for the unbiased study of enhancer activities in organisms during health and disease.

Transcriptome analyses of murine right and left maxilla-mandibular complex.

The objective of the study was to investigate differential gene expression between murine right and left maxilla-mandibular (MxMn) complexes. Wild-type (WT) C57BL/6 embryonic (E) day 14.5 (n = 3) and 18.5 (n = 3) murine embryos. The E14.5 and 18.5 embryos were harvested and hemi-sectioned the MxMn complexes into right and left halves in the mid-sagittal plane. We isolated total RNA using Trizol reagent and further purified using the RNA-easy kit (QIAGEN). We confirmed equal expression of house-keeping genes in right and left halves using RT-PCR and then performed paired-end whole mRNA sequencing in LC Sciences (Houston, TX) followed by differential transcript analyses (>1 or <-1 log fold change; p < .05; q < .05; and FPKM >0.5 in 2/3 samples). The Mouse Genome Informatics and Online Mendelian Inheritance in Man databases as well as gnomAD constraint scores were used to prioritize differentially expressed transcripts. There were 19 upregulated and 19 downregulated transcripts at E14.5 and 8 upregulated and 17 downregulated transcripts at E18.5 time-points. These differentially expressed transcripts were statistically significant and shown to be associated with craniofacial phenotypes in mouse models. These transcripts also have significant gnomAD constraint scores and are enriched in biological processes critical for embryogenesis. We identified significant differential expression of transcripts between E14.5 and 18.5 murine right and left MxMn complexes. These findings when extrapolated to humans, they may provide a biological basis for facial asymmetry. Further experiments are required to validate these findings in murine models with craniofacial asymmetry.

Deep learning and support vector machines for transcription start site identification

Recognizing transcription start sites is key to gene identification. Several approaches have been employed in related problems such as detecting translation initiation sites or promoters, many of the most recent ones based on machine learning. Deep learning methods have been proven to be exceptionally effective for this task, but their use in transcription start site identification has not yet been explored in depth. Also, the very few existing works do not compare their methods to support vector machines (SVMs), the most established technique in this area of study, nor provide the curated dataset used in the study. The reduced amount of published papers in this specific problem could be explained by this lack of datasets. Given that both support vector machines and deep neural networks have been applied in related problems with remarkable results, we compared their performance in transcription start site predictions, concluding that SVMs are computationally much slower, and deep learning methods, specially long short-term memory neural networks (LSTMs), are best suited to work with sequences than SVMs. For such a purpose, we used the reference human genome GRCh38. Additionally, we studied two different aspects related to data processing: the proper way to generate training examples and the imbalanced nature of the data. Furthermore, the generalization performance of the models studied was also tested using the mouse genome, where the LSTM neural network stood out from the rest of the algorithms. To sum up, this article provides an analysis of the best architecture choices in transcription start site identification, as well as a method to generate transcription start site datasets including negative instances on any species available in Ensembl. We found that deep learning methods are better suited than SVMs to solve this problem, being more efficient and better adapted to long sequences and large amounts of data. We also create a transcription start site (TSS) dataset large enough to be used in deep learning experiments.