Machine learning and related approaches in transcriptomics.

Data acquisition for transcriptomic studies used to be the bottleneck in the transcriptomic analytical pipeline. However, recent developments in transcriptome profiling technologies have increased researchers' ability to obtain data, resulting in a shift in focus to data analysis. Incorporating machine learning to traditional analytical methods allows the possibility of handling larger volumes of complex data more efficiently. Many bioinformaticians, especially those unfamiliar with ML in the study of human transcriptomics and complex biological systems, face a significant barrier stemming from their limited awareness of the current landscape of ML utilisation in this field. To address this gap, this review endeavours to introduce those individuals to the general types of ML, followed by a comprehensive range of more specific techniques, demonstrated through examples of their incorporation into analytical pipelines for human transcriptome investigations. Important computational aspects such as data pre-processing, task formulation, results (performance of ML models), and validation methods are encompassed. In hope of better practical relevance, there is a strong focus on studies published within the last five years, almost exclusively examining human transcriptomes, with outcomes compared with standard non-ML tools.

Protein-coding circular RNAs - mechanism, detection, and their role in cancer and neurodegenerative diseases.

Circular RNAs (circRNAs) are a unique class of non-coding RNAs and were originally thought to have no protein-coding potential due to their lack of a 5' cap and 3' poly(A) tail. However, recent studies have challenged this notion and revealed that some circRNAs have protein-coding potential. They have emerged as a key area of interest in cancer and neurodegeneration research as recent studies have identified several circRNAs that can produce functional proteins with important roles in cancer progression. The protein-coding potential of circRNAs is determined by the presence of an open reading frame (ORF) within the circular structure that can encode a protein. In some cases, the ORF can be translated into a functional protein despite the lack of traditional mRNA features. While the protein-coding potential of most circRNAs remains unclear, several studies have identified specific circRNAs that can produce functional proteins. Understanding the protein-coding potential of circRNAs is important for unravelling their biological functions and potential roles in disease. Our review provides comprehensive coverage of recent advances in the field of circRNA protein-coding capacity and its impact on cancer and neurodegenerative diseases pathogenesis and progression.

Serotonin receptor expression in hippocampus and temporal cortex of temporal lobe epilepsy patients by postictal generalized electroencephalographic suppression duration.

OBJECTIVE: Prolonged postictal generalized electroencephalographic suppression (PGES) is a potential biomarker for sudden unexpected death in epilepsy (SUDEP), which may be associated with dysfunctional autonomic responses and serotonin signaling. To better understand molecular mechanisms, PGES duration was correlated to 5HT1A and 5HT2A receptor protein expression and RNAseq from resected hippocampus and temporal cortex of temporal lobe epilepsy patients with seizures recorded in preoperative evaluation. METHODS: Analyses included 36 cases (age = 14-64 years, age at epilepsy onset = 0-51 years, epilepsy duration = 2-53 years, PGES duration = 0-93 s), with 13 cases in all hippocampal analyses. 5HT1A and 5HT2A protein was evaluated by Western blot and histologically in hippocampus (n = 16) and temporal cortex (n = 9). We correlated PGES duration to our previous RNAseq dataset for serotonin receptor expression and signaling pathways, as well as weighted gene correlation network analysis (WGCNA) to identify correlated gene clusters. RESULTS: In hippocampus, 5HT2A protein by Western blot positively correlated with PGES duration (p = .0024, R2  = .52), but 5HT1A did not (p = .87, R2  = .0020). In temporal cortex, 5HT1A and 5HT2A had lower expression and did not correlate with PGES duration. Histologically, PGES duration did not correlate with 5HT1A or 5HT2A expression in hippocampal CA4, dentate gyrus, or temporal cortex. RNAseq identified two serotonin receptors with expression that correlated with PGES duration in an exploratory analysis: HTR3B negatively correlated (p = .043, R2  = .26) and HTR4 positively correlated (p = .049, R2  = .25). WGCNA identified four modules correlated with PGES duration, including positive correlation with synaptic transcripts (p = .040, Pearson correlation r = .52), particularly potassium channels (KCNA4, KCNC4, KCNH1, KCNIP4, KCNJ3, KCNJ6, KCNK1). No modules were associated with serotonin receptor signaling. SIGNIFICANCE: Higher hippocampal 5HT2A receptor protein and potassium channel transcripts may reflect underlying mechanisms contributing to or resulting from prolonged PGES. Future studies with larger cohorts should assess functional analyses and additional brain regions to elucidate mechanisms underlying PGES and SUDEP risk.

Circular RNA Expression and Interaction Patterns Are Perturbed in Amyotrophic Lateral Sclerosis.

Circular RNAs (circRNAs) are a type of long noncoding RNA that are highly abundant and highly conserved throughout evolution and exhibit differential expression patterns in various tissue types in multiple diseases, including amyotrophic lateral sclerosis (ALS). The most well-known function of circRNAs is their ability to act as microRNA (miRNA) sponges. This entails circRNA binding to miRNA, which would otherwise target and degrade messenger RNA, thus affecting gene expression. This study analyzed ALS patient samples from three spinal cord regions to investigate circular transcriptome perturbation and circular RNA-microRNA-mRNA interactions. Using stringent statistical parameters, we identified 92 differentially expressed circRNAs across the spinal cord tissues with the aim of identifying specific circRNAs with biomarker potential. We also found evidence for differential expression of 37 linear RNAs possibly due to miRNA sequestration by circRNAs, thus revealing their potential as novel biomarkers and therapeutic candidates for ALS.

Investigation of Transcriptome Patterns in Endometrial Cancers from Obese and Lean Women.

Endometrial cancer is the most common gynaecological malignancy in developed countries. One of the largest risk factors for endometrial cancer is obesity. The aim of this study was to determine whether there are differences in the transcriptome of endometrial cancers from obese vs. lean women. Here we investigate the transcriptome of endometrial cancer between obese and lean postmenopausal women using rRNA-depleted RNA-Seq data from endometrial cancer tissues and matched adjacent non-cancerous endometrial tissues. Differential expression analysis identified 12,484 genes (6370 up-regulated and 6114 down-regulated) in endometrial cancer tissues from obese women, and 6219 genes (3196 up-regulated and 3023 down-regulated) in endometrial cancer tissues from lean women (adjusted p-value < 0.1). A gene ontology enrichment analysis revealed that the top 1000 up-regulated genes (by adjusted p-value) were enriched for growth and proliferation pathways while the top 1000 down-regulated genes were enriched for cytoskeleton restructure networks in both obese and lean endometrial cancer tissues. In this study, we also show perturbations in the expression of protein coding genes (HIST1H2BL, HIST1H3F, HIST1H2BH, HIST1H1B, TTK, PTCHD1, ASPN, PRELP, and CDH13) and the lncRNA MBNL1-AS1 in endometrial cancer tissues. Overall, this study has identified gene expression changes that are similar and also unique to endometrial cancers from obese vs. lean women. Furthermore, some of these genes may serve as prognostic biomarkers or, possibly, therapeutic targets for endometrial cancer.

Neural circular transcriptomes across mammalian species.

Circular RNAs (circRNAs) have recently attracted significant interest in the realm of science and the evolution of species. Given the lack of information available on circRNAs due to various barriers related to sequencing techniques and bioinformatics tools, little regarding their function is known. It has been predicted that circRNAs contribute to gene expression regulation, but aside from a few specific cases, this contention has yet to be proven. Although the role of circRNAs in evolution remains elusive, from the few studies that have shown circRNA conservation in mammalian species, tissue specificity in brain regions, and the abundance of circRNAs in the brains of various species, the concept is becoming more likely with much gravitas. The proposed functional role of circRNAs being gene regulators is of great interest and would provide a basis to further understand not only the functional capabilities of organisms, but also the evolution of mammalian species.

Cell type-specific circular RNA expression in human glial cells.

The circular transcriptome of human glial cells is an area of neuroscience that has not been thoroughly elucidated. Circular RNAs (circRNAs) have the potential to facilitate the understanding of vast, complex and unknown mechanisms derived from the human transcriptome, including elements of the human brain that are not known and the evolution of the human brain, the complexities of which are not well understood. Moreover, the glial cells have been determined to contribute to human brain evolution. This study presents the first comprehensive analysis of the human brain glia circRNA transcriptome, that is, astrocytes, microglia and oligodendrocytes. After stringent criteria applied to the detection of circRNAs, it was found that the circular transcriptomes of these glia are unique from one another, and hence might be indicative of distinct roles for circRNAs within the brain. This study found 265, 239 and 442 circRNAs comprising the unique circular transcriptome of astrocytes, microglia and oligodendrocytes, respectively. The most abundant circRNAs in these glial cell types are expressed by parent genes co-expressing linear RNAs in low abundance, suggesting spliceosome activity favorable to the back-splicing mechanism instead of canonical splicing activity.

Coding and non-coding transcriptome of mesial temporal lobe epilepsy: Critical role of small non-coding RNAs.

Our understanding of mesial temporal lobe epilepsy (MTLE), one of the most common form of drug-resistant epilepsy in humans, is derived mainly from clinical, imaging, and physiological data from humans and animal models. High-throughput gene expression studies of human MTLE have the potential to uncover molecular changes underlying disease pathogenesis along with novel therapeutic targets. Using RNA- and small RNA-sequencing in parrallel, we explored differentially expressed genes in the hippocampus and cortex of MTLE patients who had undergone surgical resection and non-epileptic controls. We identified differentially expressed genes in the hippocampus of MTLE patients and differentially expressed small RNAs across both the cortex and hippocampus. We found significant enrichment for astrocytic and microglial genes among up-regulated genes, and down regulation of neuron specific genes in the hippocampus of MTLE patients. The transcriptome profile of the small RNAs reflected disease state more robustly than mRNAs, even across brain regions which show very little pathology. While mRNAs segregated predominately by brain region for MTLE and controls, small RNAs segregated by disease state. In particular, our data suggest that specific miRNAs (e.g., let-7b-3p and let-7c-3p) may be key regulators of multiple pathways related to MTLE pathology. Further, we report a strong association of other small RNA species with MTLE pathology. As such we have uncovered novel elements that may contribute to the establishment and progression of MTLE pathogenesis and that could be leveraged as therapeutic targets.

Changes in circular RNA expression patterns during human foetal brain development.

Circular RNAs (circRNAs) are a recently identified class of long non-coding RNAs and their expression is regulated in a tissue- and developmental stage-specific manner. Recent studies indicate the potential regulatory role that circRNAs may have, particularly in the brain, where they are most abundant. This study aims to elucidate changes in circRNA patterns during human embryonic brain development. We detected a number of differentially expressed genes that showed distinct expression profiles for circular and linear transcripts despite having originated from the same genes, implicating a dichotomy in the regulation of these two RNA species. Altogether our study showed that circular and linear RNAs have independent expression patterns, and that circular transcriptomes from different developing stages have distinct characteristics in terms of transcript abundance and isoform diversity.

Early transcriptome changes in response to chemical long-term potentiation induced via activation of synaptic NMDA receptors in mouse hippocampal neurons.

Long term potentiation (LTP) is a form of synaptic plasticity. In the present study LTP was induced via activation of synaptic NMDA receptors in primary hippocampal neuron cultures from neonate mice and RNA was isolated for RNA sequencing at 20 min following LTP induction. RNA sequencing and differential expression testing was performed to determine the identity and abundance of protein-coding and non-coding RNAs in control and LTP induced neuron cultures. We show that expression levels of a small group of transcripts encoding proteins involved in negative regulation of gene expression (Adcyap1, Id3), protein translation (Rpl22L1), extracellular structure organization (Bgn), intracellular signalling (Ppm1H, Ntsr2, Cldn10) and protein citrullination (PAD2) are downregulated in the stimulated neurons. Our results suggest that the early stages of LTP are accompanied by the remodelling of the biosynthetic machinery, interactions with the extracellular matrix and intracellular signalling pathways at the transcriptional level.

Transcriptome landscape of long intergenic non-coding RNAs in endometrial cancer.

Endometrial cancer is the most common gynecological malignancy in the developed world. It is the fifth most common cancer and accounts for 4.8% of all cancers in women. Long intergenic non-coding RNAs (lincRNAs), a subclass of long non-coding RNAs, are pervasively transcribed throughout the human genome. OBJECTIVE: LincRNA expression patterns in endometrial cancer compared to normal healthy tissue are poorly characterised. In this study, the lincRNA transcriptome of endometrial cancers and adjacent normal endometrium from the same patients was sequenced and compared with transcriptomes of other gynaecologic malignancies including ovarian and cervical cancers. METHODS: RNA was isolated from malignant and adjacent non-affected endometrial tissue from 6 patients with low grade and stage Type I endometrial cancer. Subsequently, Illumina paired-end RNA sequencing was performed, followed by bioinformatics analysis, to determine differential transcriptome expression patterns. RESULTS: LINC00958 was upregulated in all three cancers, and four lincRNAs including LINC01480, LINC00645, LINC00891 and LINC00702 demonstrated exquisite specificity for malignant endometrium compared to normal endometrium while also distinguishing endometrial cancer from ovarian and cervical cancers. Furthermore, LINC01480 has features required to express a micropeptide. CONCLUSIONS: The lincRNAs, characterised in this study, represent high priority genes to be tested for functional significance in the pathogenesis and/or progression of endometrial cancer. Furthermore, lincRNAs have potential to be released into the bloodstream and therefore the four lincRNAs identified here may represent biomarkers for early detection of endometrial cancer without biopsy.

Transcriptional regulation of long-term potentiation.

Long-term potentiation (LTP), the persistent strengthening of synapses following high levels of stimulation, is a form of synaptic plasticity that has been studied extensively as a possible mechanism for learning and memory formation. The strengthening of the synapse that occurs during LTP requires cascades of complex molecular processes and the coordinated remodeling of pre-synaptic and post-synaptic neurons. Despite over four decades of research, our understanding of the transcriptional mechanisms and molecular processes underlying LTP remains incomplete. Identification of all the proteins and non-coding RNA transcripts expressed during LTP may provide greater insight into the molecular mechanisms involved in learning and memory formation.

Characterization of circular RNAs landscape in multiple system atrophy brain.

Circular RNAs (circRNAs) have been recently identified as a naturally occurring family of widespread and diverse endogenous non-coding RNAs that may regulate gene expression in mammals. They are unusually stable RNA molecules with cell type- or developmental stage-specific expression patterns. However, the role of circRNAs in pathology of complex disease is entirely unknown. Here, we report the specific circular transcriptome in the multiple system atrophy (MSA) brain as determined by RNA sequencing. Five circRNAs, namely IQCK, MAP4K3, EFCAB11, DTNA, and MCTP1, were identified and validated as specifically over-expressed in MSA frontal cortex. The expression levels of linear transcripts were not significantly altered and thus did not follow the pattern of their circular counterparts. Further analysis of expression of five MSA-specific circRNAs revealed their over-expression in the white matter of the MSA cortical tissue. Together, this is the first report describing perturbation of circular transcriptome in α-synucleinopathies.

Transcriptome analysis of grey and white matter cortical tissue in multiple system atrophy.

Multiple system atrophy (MSA) is a distinct member of a group of neurodegenerative diseases known as α-synucleinopathies, which are characterized by the presence of aggregated α-synuclein in the brain. MSA is unique in that the principal site for α-synuclein deposition is in the oligodendrocytes rather than neurons. The cause of MSA is unknown, and the pathogenesis of MSA is still largely speculative. Brain transcriptome perturbations during the onset and progression of MSA are mostly unknown. Using RNA sequencing, we performed a comparative transcriptome profiling analysis of the grey matter (GM) and white matter (WM) of the frontal cortex of MSA and control brains. The transcriptome sequencing revealed increased expression of the alpha and beta haemoglobin genes in MSA WM, decreased expression of the transthyretin (TTR) gene in MSA GM and numerous region-specific long intervening non-coding RNAs (lincRNAs). In contrast, we observed only moderate changes in the expression patterns of the α-synuclein (SNCA) gene, which confirmed previous observations by other research groups. Our study suggests that at the transcriptional level, MSA pathology may be related to increased iron levels in WM and perturbations of the non-coding fraction of the transcriptome.

Long intervening non-coding RNA 00320 is human brain-specific and highly expressed in the cortical white matter.

Pervasive transcription of the genome produces a diverse array of functional non-coding RNAs (ncRNAs). One particular class of ncRNAs, long intervening non-coding RNAs (lincRNAs) are thought to play a role in regulating gene expression and may be a major contributor to organism and tissue complexity. The human brain with its heterogeneous cellular make-up is a rich source of lincRNAs; however, the functions of the majority of lincRNAs are unknown. Recently, by completing RNA sequencing (RNA-Seq) of the human frontal cortex, we identified linc00320 as being highly expressed in the white matter compared to grey matter in multiple system atrophy (MSA) brain. Here, we further investigate the expression patterns of linc00320 and conclude that it is involved in specific brain regions rather than having involvement in the MSA disease process. We also show that the full-length linc00320 is only expressed in human brain tissue and not in other primates, suggesting that it may be involved in improved functional connectivity for higher human brain cognition.

Long noncoding RNAs in TDP-43 and FUS/TLS-related frontotemporal lobar degeneration (FTLD).

Frontotemporal lobar degeneration (FTLD) defines a spectrum of heterogeneous neurodegenerative disorders characterized by the progressive deterioration of the frontal and anterior temporal lobes of the brain. FTLD is histopathologically classified according to the presence of neuropathological protein aggregates. Two of the major pathologies, FTLD-TDP and FTLD-FUS, are characterized by the abnormal accumulation in cytoplasmic inclusions of RNA-binding proteins (RBPs) - TDP-43 and FUS/TLS, respectively. That suggests that a crucial common downstream pathway leading to cell death might involve the disruption of RNA-based mechanisms. Long noncoding RNAs have emerged as key regulators in the different layers of gene regulation. Increasing evidence suggests that long non-coding RNAs (lncRNAs) may have pivotal biological functions in the brain and, not surprisingly, they have been implicated with neurodegenerative diseases, like Alzheimer's and Parkinson's diseases. Recent studies report that FTLD/ALS-related proteins TDP-43 and FUS/TLS bind lncRNAs, and that several lncRNAs have binding sites for TDP-43 and/or FUS/TLS. These findings raise important questions about how TDP-43 and FUS/TLS pathologies can affect lncRNA-based mechanisms. One alternative is that TDP-43 and FUS/TLS regulate lncRNA transcription or transcript stability. In fact, it has been demonstrated that lncRNAs are dysregulated upon either depletion or unavailability of functional TDP-43 or FUS/TLS in a range of different models and diseases, including post-mortem samples from subjects with FTLD-TDP. The second alternative is that the binding to TDP-43 or FUS/TLS would enable lncRNAs to perform their cellular function. In this case, the unavailability of these RBPs would disrupt functional properties of lncRNAs, without necessarily altering their cellular levels. It has been experimentally demonstrated that the cellular function of some lncRNAs is strictly dependent on the direct binding to TDP-43 or FUS/TLS.

The alternative splicing of the apolipoprotein E gene is unperturbed in the brains of Alzheimer's disease patients.

The prevalence of Alzheimer's disease (AD) is increasing rapidly worldwide due to an ageing population and a lack of disease modifying therapeutics. In monogenic forms of AD mutations lead to the accumulation of neurotoxic peptides called beta-amyloid. Beta-amyloid accumulation is also postulated to precipitate sporadic AD although the pathogenesis of this common form remains largely unknown. The two leading risk factors for sporadic AD are ageing and the possession of the APOE epsilon 4 allele. Changes in APOE expression that are independent of the epsilon genotype have also been described in the AD brain including a recent RNA-Seq analysis that showed upregulation of a rare alternative splice isoform (APOE-005). To replicate these RNA-Seq findings we used quantitative reverse transcriptase polymerase chain reaction (RT-qPCR) to compare APOE-005 and total APOE expression in the superior temporal gyrus of 14 AD cases and 16 neurologically normal controls. In AD, this area shows prominent beta-amyloid deposition but few neurofibrillary tangles and only moderate neuronal loss. As poorer RNA quality among the AD cases was a likely confounder in this study, the analysis was repeated in a RIN-matched sub-cohort of 17 individuals. Contrary to the original RNA-Seq study, we found no difference in total APOE, APOE-005 or the common isoform, APOE-001, between AD cases and controls. Our findings are consistent with ApoE acting largely at the protein level to increase the risk for sporadic AD.

Recent advances in the investigation of fusion RNAs and their role in molecular pathology of cancer.

Gene fusions have had a significant role in the development of various types of cancer, oftentimes involved in oncogenic activities through dysregulation of gene expression or signalling pathways. Some cancer-associated chromosomal translocations can undergo backsplicing, resulting in fusion-circular RNAs, a more stable isoform immune to RNase degradation. This stability makes fusion circular RNAs a promising diagnostic biomarker for cancer. While the detection of linear fusion RNAs and their function in certain cancers have been described in literature, fusion circular RNAs lag behind due to their low abundance in cancer cells. This review highlights current literature on the role of linear and circular fusion transcripts in cancer, tools currently available for detecting of these chimeric RNAs and their function and how they play a role in tumorigenesis.

Investigation of the Circular Transcriptome in Alzheimer's Disease Brain.

Circular RNAs (circRNAs) are a subclass of non-coding RNAs which have demonstrated potential as biomarkers for Alzheimer's disease (AD). In this study, we conducted a comprehensive exploration of the circRNA transcriptome within AD brain tissues. Specifically, we assessed circRNA expression patterns in the dorsolateral prefrontal cortex collected from nine AD-afflicted individuals and eight healthy controls. Utilising two circRNA detection tools, CIRI2 and CIRCexplorer2, we detected thousands of circRNAs and performed a differential expression analysis. CircRNAs which exhibited statistically significantly differential expression were identified as AD-specific differentially expressed circRNAs. Notably, our investigation revealed 120 circRNAs with significant upregulation and 1325 circRNAs displaying significant downregulation in AD brains when compared to healthy brain tissue. Additionally, we explored the expression profiles of the linear RNA counterparts corresponding to differentially expressed circRNAs in AD-afflicted brains and discovered that the linear RNA counterparts exhibited no significant changes in the levels of expression. We used CRAFT tool to predict that circUBE4B had potential to target miRNA named as hsa-miR-325-5p, ultimately regulated CD44 gene. This study provides a comprehensive overview of differentially expressed circRNAs in the context of AD brains, underscoring their potential as molecular biomarkers for AD. These findings significantly enhance our comprehension of AD's underlying pathophysiological mechanisms, offering promising avenues for future diagnostic and therapeutic developments.

Bile acids mediate fructose-associated liver tumour growth in mice.

High fructose diets are associated with an increased risk of liver cancer. Previous studies in mice suggest increased lipogenesis is a key mechanism linking high fructose diets to liver tumour growth. However, these studies administered fructose to mice at supraphysiological levels. The aim of this study was to determine whether liver tumour growth and lipogenesis were altered in mice fed fructose at physiological levels. To test this, we injected male C57BL/6 mice with the liver carcinogen diethylnitrosamine and then fed them diets without fructose or fructose ranging from 10 to 20 % total calories. Results showed mice fed diets with ≥15 % fructose had significantly increased liver tumour numbers (2-4-fold) and total tumour burden (∼7-fold) vs mice fed no-fructose diets. However, fructose-associated tumour burden was not associated with lipogenesis. Conversely, unbiased metabolomic analyses revealed bile acids were elevated in the sera of mice fed a 15 % fructose diet vs mice fed a no-fructose diet. Using a syngeneic ectopic liver tumour model, we show that ursodeoxycholic acid, which decreases systemic bile acids, significantly reduced liver tumour growth in mice fed the 15 % fructose diet but not mice fed a no-fructose diet. These results point to a novel role for systemic bile acids in mediating liver tumour growth associated with a high fructose diet. Overall, our study shows fructose intake at or above normal human consumption (≥15 %) is associated with increased liver tumour numbers and growth and that modulating systemic bile acids inhibits fructose-associated liver tumour growth in mice.