Functional organization of the transcriptome in human brain

Network Analysis of a Membrane-Enriched Brain Proteome across Stages of Alzheimer's Disease

Previous systems-based proteomic approaches have characterized alterations in protein co-expression networks of unfractionated asymptomatic (AsymAD) and symptomatic Alzheimer’s disease (AD) brains. However, it remains unclear how sample fractionation and sub-proteomic analysis influences the organization of these protein networks and their relationship to clinicopathological traits of disease. In this proof-of-concept study, we performed a systems-based sub-proteomic analysis of membrane-enriched post-mortem brain samples from pathology-free control, AsymAD, and AD brains (n = 6 per group). Label-free mass spectrometry based on peptide ion intensity was used to quantify the 18 membrane-enriched fractions. Differential expression and weighted protein co-expression network analysis (WPCNA) were then used to identify and characterize modules of co-expressed proteins most significantly altered between the groups. We identified a total of 27 modules of co-expressed membrane-associated proteins. In contrast to the unfractionated proteome, these networks did not map strongly to cell-type specific markers. Instead, these modules were principally organized by their associations with a wide variety of membrane-bound compartments and organelles. Of these, the mitochondrion was associated with the greatest number of modules, followed by modules linked to the cell surface compartment. In addition, we resolved networks with strong associations to the endoplasmic reticulum, Golgi apparatus, and other membrane-bound organelles. A total of 14 of the 27 modules demonstrated significant correlations with clinical and pathological AD phenotypes. These results revealed that the proteins within individual compartments feature a heterogeneous array of AD-associated expression patterns, particularly during the preclinical stages of disease. In conclusion, this systems-based analysis of the membrane-associated AsymAD brain proteome yielded a unique network organization highly linked to cellular compartmentalization. Further study of this membrane-associated proteome may reveal novel insight into the complex pathways governing the earliest stages of disease.

Identification of candidate genes and miRNAs associated with neuropathic pain induced by spared nerve injury

Neuropathic pain (NP) is a complex, chronic pain condition caused by injury or dysfunction affecting the somatosensory nervous system. This study aimed to identify crucial genes and miRNAs involved in NP. Microarray data (access number GSE91396) were downloaded from the Gene Expression Omnibus (GEO). Murine RNA-seq samples from three brain regions [nucleus accumbens, (NAc); medial prefrontal cortex, (mPFC) and periaqueductal gray, (PAG)] were compared between the spared nerve injury (SNI) model and a sham surgery. After data normalization, differentially expressed RNAs were screened using the limma package and functional enrichment analysis was performed with Database for Annotation, Visualization and Integrated Discovery. The microRNA (miRNA/miR)-mRNA regulatory network and miRNA-target gene-pathway regulatory network were constructed using Cytoscape software. A total of 2,776 differentially expressed RNAs (219 miRNAs and 2,557 mRNAs) were identified in the SNI model compared with the sham surgery group. A total of two important modules (red and turquoise module) were found to be related to NP using weighed gene co-expression network analysis (WGCNA) for the 2,325 common differentially expressed RNAs in three brain regions. The differentially expressed genes (DEGs) in the miRNA-mRNA regulatory network were significantly enriched in 21 Gene Ontology terms and five pathways. A total of four important DEGs (CXCR2, IL12B, TNFSF8 and GRK1) and five miRNAs (miR-208a-5p, miR-7688-3p, miR-344f-3p, miR-135b-3p and miR-135a-2-3p) were revealed according to the miRNA-target gene-pathway regulatory network to be related to NP. Four important DEGs (CXCR2, IL12B, TNFSF8 and GRK1) and five miRNAs (miR-208a-5p, miR-7688-3p, miR-344f-3p, miR-135b-3p and miR-135a-2-3p) were differentially expressed in SNI, indicating their plausible roles in NP pathogenesis.

MiR-190a potentially ameliorates postoperative cognitive dysfunction by regulating Tiam1

BACKGROUND

Postoperative cognitive dysfunction (POCD) affects a large number of post-surgery patients, especially for the elderly. However, the etiology of this neurocognitive disorder is largely unknown. Even if several studies have reported a small number of miRNAs as the essential modulatory factors in POCD, these findings are still rather limited. The aim of current study was to screen the POCD-related miRNAs in the hippocampus tissues and investigate the target genes of differentially expressed miRNAs and their biological functions underlying POCD pathophysiology.

METHODS

The miRNA microarray was used to find the abnormal expression of miRNAs in the hippocampus tissues from the POCD model mice to normal mice (Discovery cohort, 3 vs 3). The nominal significant results were validated in an independent sample of hippocampus tissues of 10 mice based on same miRNA microarray (Replication cohort, 5 vs 5). Expression level of the most significantly abnormal miRNA was further validated by real-time quantitative polymerase chain reaction (PCR). To determine the expression pattern among miRNAs and genes and detect the interactions, we conducted a weighted gene co-expression network analysis (WGCNA) in the miRNAs and genes expression data from hippocampus tissue of wild type mice (n = 24). The target genes of miRNAs were predicted using the miRWalk3.0 software. Furthermore, we used the ClueGO software to decipher the pathways network and reveal the biological functions of target genes of miRNAs.

RESULTS

We found that nine miRNAs showed significant associations with POCD in both datasets. Among these miRNAs, mmu-miR-190a-3p was the most significant one. By performing WGCNA analysis, we found 25 co-expression modules, of which mmu-miR-190a-3p was significantly anti-correlated with red module. Moreover, in the red module, 314 genes were significantly enriched in four pathways such as axon guidance and calcium signaling pathway, which are well-documented to be associated with psychiatric disorders and brain development. Also, 169 of the 314 genes were highly correlated with mmu-miR-190a-3p, and four genes (Sphkap, Arhgef25, Tiam1, and Ntrk3) had putative binding sites at 3'-UTR of mmu-miR-190a-3p. Based on protein-protein network analysis, we detected that Tiam1 was a central gene regulated by the mmu-miR-190a-3p.

CONCLUSIONS

Taken together, we conclude that mmu-miR-190a-3p is involved in the etiology of POCD and may serve as a novel predictive indicator for POCD.

Functional Geometry of Human Connectomes

Mapping the brain imaging data to networks, where nodes represent anatomical brain regions and edges indicate the occurrence of fiber tracts between them, has enabled an objective graph-theoretic analysis of human connectomes. However, the latent structure on higher-order interactions remains unexplored, where many brain regions act in synergy to perform complex functions. Here we use the simplicial complexes description of human connectome, where the shared simplexes encode higher-order relationships between groups of nodes. We study consensus connectome of 100 female (F-connectome) and of 100 male (M-connectome) subjects that we generated from the Budapest Reference Connectome Server v3.0 based on data from the Human Connectome Project. Our analysis reveals that the functional geometry of the common F&M-connectome coincides with the M-connectome and is characterized by a complex architecture of simplexes to the 14th order, which is built in six anatomical communities, and linked by short cycles. The F-connectome has additional edges that involve different brain regions, thereby increasing the size of simplexes and introducing new cycles. Both connectomes contain characteristic subjacent graphs that make them 3/2-hyperbolic. These results shed new light on the functional architecture of the brain, suggesting that insightful differences among connectomes are hidden in their higher-order connectivity.

Cross-Species Co-analysis of Prefrontal Cortex Chronic Ethanol Transcriptome Responses in Mice and Monkeys

Despite recent extensive genomic and genetic studies on behavioral responses to ethanol, relatively few new therapeutic targets for the treatment of alcohol use disorder have been validated. Here, we describe a cross-species genomic approach focused on identifying gene networks associated with chronic ethanol consumption. To identify brain mechanisms underlying a chronic ethanol consumption phenotype highly relevant to human alcohol use disorder, and to elucidate potential future therapeutic targets, we conducted a genomic study in a non-human primate model of chronic open-access ethanol consumption. Microarray analysis of RNA expression in anterior cingulate and subgenual cortices from rhesus macaques was performed across multiple cohorts of animals. Gene networks correlating with ethanol consumption or showing enrichment for ethanol-regulated genes were identified, as were major ethanol-related hub genes within these networks. A subsequent consensus module analysis was used to co-analyze monkey data with expression data from a chronic intermittent ethanol vapor-exposure and consumption model in C57BL/6J mice. Ethanol-related gene networks conserved between primates and rodents were enriched for genes involved in discrete biological functions, including; myelination, synaptic transmission, chromatin modification, Golgi apparatus function, translation, cellular respiration, and RNA processing. The myelin-related network, in particular, showed strong correlations with ethanol consumption behavior and displayed marked network reorganization between control and ethanol-drinking animals. Further bioinformatics analysis revealed that these networks also showed highly significant overlap with other ethanol-regulated gene sets. Altogether, these studies provide robust primate and rodent cross-species validation of gene networks associated with chronic ethanol consumption. Our results also suggest potential novel focal points for future therapeutic interventions in alcohol use disorder.

iSyTE 2.0: a database for expression-based gene discovery in the eye

Although successful in identifying new cataract-linked genes, the previous version of the database iSyTE (integrated Systems Tool for Eye gene discovery) was based on expression information on just three mouse lens stages and was functionally limited to visualization by only UCSC-Genome Browser tracks. To increase its efficacy, here we provide an enhanced iSyTE version 2.0 (URL: http://research.bioinformatics.udel.edu/iSyTE) based on well-curated, comprehensive genome-level lens expression data as a one-stop portal for the effective visualization and analysis of candidate genes in lens development and disease. iSyTE 2.0 includes all publicly available lens Affymetrix and Illumina microarray datasets representing a broad range of embryonic and postnatal stages from wild-type and specific gene-perturbation mouse mutants with eye defects. Further, we developed a new user-friendly web interface for direct access and cogent visualization of the curated expression data, which supports convenient searches and a range of downstream analyses. The utility of these new iSyTE 2.0 features is illustrated through examples of established genes associated with lens development and pathobiology, which serve as tutorials for its application by the end-user. iSyTE 2.0 will facilitate the prioritization of eye development and disease-linked candidate genes in studies involving transcriptomics or next-generation sequencing data, linkage analysis and GWAS approaches.

A single-nuclei RNA sequencing study of Mendelian and sporadic AD in the human brain

Background

Alzheimer’s disease (AD) is the most common form of dementia. This neurodegenerative disorder is associated with neuronal death and gliosis heavily impacting the cerebral cortex. AD has a substantial but heterogeneous genetic component, presenting both Mendelian and complex genetic architectures. Using bulk RNA-seq from the parietal lobes and deconvolution methods, we previously reported that brains exhibiting different AD genetic architecture exhibit different cellular proportions. Here, we sought to directly investigate AD brain changes in cell proportion and gene expression using single-cell resolution.

Methods

We generated unsorted single-nuclei RNA sequencing data from brain tissue. We leveraged the tissue donated from a carrier of a Mendelian genetic mutation, PSEN1 p.A79V, and two family members who suffer from sporadic AD, but do not carry any autosomal mutations. We evaluated alternative alignment approaches to maximize the titer of reads, genes, and cells with high quality. In addition, we employed distinct clustering strategies to determine the best approach to identify cell clusters that reveal neuronal and glial cell types and avoid artifacts such as sample and batch effects. We propose an approach to cluster cells that reduces biases and enable further analyses.

Results

We identified distinct types of neurons, both excitatory and inhibitory, and glial cells, including astrocytes, oligodendrocytes, and microglia, among others. In particular, we identified a reduced proportion of excitatory neurons in the Mendelian mutation carrier, but a similar distribution of inhibitory neurons. Furthermore, we investigated whether single-nuclei RNA-seq from the human brains recapitulate the expression profile of disease-associated microglia (DAM) discovered in mouse models. We also determined that when analyzing human single-nuclei data, it is critical to control for biases introduced by donor-specific expression profiles.

Conclusion

We propose a collection of best practices to generate a highly detailed molecular cell atlas of highly informative frozen tissue stored in brain banks. Importantly, we have developed a new web application to make this unique single-nuclei molecular atlas publicly available.

Electronic supplementary material

The online version of this article (10.1186/s13195-019-0524-x) contains supplementary material, which is available to authorized users.

The transcriptional activity of a temperature-sensitive transcription factor module is associated with pollen shedding time in pine

It has long been known that the pollen shedding time in pine trees is correlated with temperature, but the molecular basis for this has remained largely unknown. To better understand the mechanisms driving temperature response and to identify the hub regulators of pollen shedding time regulation in Pinus tabuliformis Carr., we identified a set of temperature-sensitive genes by carrying out a comparative transcriptome analysis using six early pollen shedding trees (EPs) and six late pollen shedding trees (LPs) during mid-winter and at three consecutive time points in early spring. We carried out a weighted gene co-expression network analysis and constructed a transcription factor (TF) collaborative network, merging the common but differentially expressed TFs of the EPs and LPs into a joint network. We found five hub genes in the core TF module whose expression was rapidly induced by low temperatures. The transcriptional activity of this TF module was strongly associated with pollen shedding time, and likely to produce the fine balance between cold hardiness and growth activity in early spring. We confirmed the key role of temperature in regulating flowering time and identified a transcription factor module associated with pollen shedding time in P. tabuliformis. This suggests that repression of growth activity by repressors is the main mechanism balancing growth and cold hardiness in pine trees in early spring. Our results provide new insights into the molecular mechanisms regulating seasonal flowering time in pines.

Analysis of networks of host proteins in the early time points following HIV transduction

Background

Utilization of quantitative proteomics data on the network level is still a challenge in proteomics data analysis. Currently existing models use sophisticated, sometimes hard to implement analysis techniques. Our aim was to generate a relatively simple strategy for quantitative proteomics data analysis in order to utilize as much of the data generated in a proteomics experiment as possible.

Results

In this study, we applied label-free proteomics, and generated a network model utilizing both qualitative, and quantitative data, in order to examine the early host response to Human Immunodeficiency Virus type 1 (HIV-1). A weighted network model was generated based on the amount of proteins measured by mass spectrometry, and analysis of weighted networks and functional sub-networks revealed upregulation of proteins involved in translation, transcription, and DNA condensation in the early phase of the viral life-cycle.

Conclusion

A relatively simple strategy for network analysis was created and applied to examine the effect of HIV-1 on host cellular proteome. We believe that our model may prove beneficial in creating algorithms, allowing for both quantitative and qualitative studies of proteome change in various biological and pathological processes by quantitative mass spectrometry.

Electronic supplementary material

The online version of this article (10.1186/s12859-019-2990-3) contains supplementary material, which is available to authorized users.

Identifying novel biomarkers in hepatocellular carcinoma by weighted gene co-expression network analysis

Hepatocellular carcinoma (HCC) is a highly malignant tumor found in the bile duct epithelial cells, and the second most common tumor of the liver. However, the pivotal roles of most molecules of tumorigenesis in HCC are still unclear. Hence, it is essential to detect the tumorigenic mechanism and develop novel prognostic biomarkers for clinical application. The data of HCC mRNA-seq and clinical information from The Cancer Genome Atlas (TCGA) database were analyzed by weighted gene co-expression network analysis (WGCNA). Co-expression modules and clinical traits were constructed by the Pearson correlation analysis, interesting modules were selected and gene ontology and pathway enrichment analysis were performed. Intramodule analysis and protein-protein interaction construction of selected modules were conducted to screen hub genes. In addition, upstream transcription factors and microRNAs of hub genes were predicted by miRecords and NetworkAnalyst database. Afterward, a high connectivity degree of hub genes from two networks was picked out to perform the differential expression validation in the Gene Expression Profiling Interactive Analysis database and Human Protein Atlas database and survival analysis in Kaplan-Meier plotter online tool. By utilizing WGCNA, several hub genes that regulate the mechanism of tumorigenesis in HCC were identified, which was associated with clinical traits including the pathological stage, histological grade, and liver function. Surprisingly, ZWINT, CENPA, RACGAP1, PLK1, NCAPG, OIP5, CDCA8, PRC1, and CDK1 were identified statistically as hub genes in the blue module, which were closely implicated in pathological T stage and histologic grade of HCC. Moreover, these genes also were strongly associated with the HCC cell growth and division. Network and survival analyses found that nine hub genes may be considered theoretically as indicators to predict the prognosis of patients with HCC or clinical treatment target, it will be necessary for basic experiments and large-scale cohort studies to validate further.

Prefrontal Coexpression of Schizophrenia Risk Genes Is Associated With Treatment Response in Patients

BACKGROUND

Gene coexpression networks are relevant to functional and clinical translation of schizophrenia risk genes. We hypothesized that schizophrenia risk genes converge into coexpression pathways that may be associated with gene regulation mechanisms and with response to treatment in patients with schizophrenia.

METHODS

We identified gene coexpression networks in two prefrontal cortex postmortem RNA sequencing datasets (n = 688) and replicated them in four more datasets (n = 1295). We identified and replicated (p values < .001) a single module enriched for schizophrenia risk loci (13 risk genes in 10 loci). In silico screening of potential regulators of the schizophrenia risk module via bioinformatic analyses identified two transcription factors and three microRNAs associated with the risk module. To translate postmortem information into clinical phenotypes, we identified polymorphisms predicting coexpression and combined them to obtain an index approximating module coexpression (Polygenic Coexpression Index [PCI]).

RESULTS

The PCI-coexpression association was successfully replicated in two independent brain transcriptome datasets (n = 131; p values < .05). Finally, we tested the association between the PCI and short-term treatment response in two independent samples of patients with schizophrenia treated with olanzapine (n = 167). The PCI was associated with treatment response in the positive symptom domain in both clinical cohorts (p values < .05).

CONCLUSIONS

In summary, our findings in 1983 samples of human postmortem prefrontal cortex show that coexpression of a set of genes enriched for schizophrenia risk genes is relevant to treatment response. This coexpression pathway may be coregulated by transcription factors and microRNA associated with it.

A core transcriptional signature of human microglia: Derivation and utility in describing region-dependent alterations associated with Alzheimer's disease

Growing recognition of the pivotal role microglia play in neurodegenerative and neuroinflammatory disorders has accentuated the need to characterize their function in health and disease. Studies in mouse have applied transcriptome-wide profiling of microglia to reveal key features of microglial ontogeny, functional profile, and phenotypic diversity. While similar, human microglia exhibit clear differences to their mouse counterparts, underlining the need to develop a better understanding of the human microglial profile. On examining published microglia gene signatures, limited consistency was observed between studies. Hence, we sought to derive a core microglia signature of the human central nervous system (CNS), through a comprehensive analysis of existing transcriptomic datasets. Nine datasets derived from cells and tissues, isolated from various regions of the CNS across numerous donors, were subjected independently to an unbiased correlation network analysis. From each dataset, a list of coexpressing genes corresponding to microglia was identified, with 249 genes highly conserved between them. This core signature included known microglial markers, and compared with other signatures provides a gene set specific to microglia in the context of the CNS. The utility of this signature was demonstrated by its use in detecting qualitative and quantitative region-specific alterations in aging and Alzheimer's disease. These analyses highlighted the reactive response of microglia in vulnerable brain regions such as the entorhinal cortex and hippocampus, additionally implicating pathways associated with disease progression. We believe this resource and the analyses described here, will support further investigations to the contribution of human microglia in CNS health and disease.

Exploration of the hub genes and miRNAs in lung adenocarcinoma

In order to investigate the oncogenic mechanisms of lung adenocarcinoma (LUAD), hub genes can be identified by constructing co-expression networks, and the potential linkages between hub genes, transcription factors (TFs) and microRNAs (miRNAs/miRs) can be visualized and identified. In the present study, a total of 12 co-expressed modules were constructed, and 9 of these were significantly correlated with clinical traits in LUAD. The differentially expressed genes and differentially expressed miRNAs were determined, and the targets of differentially expressed miRNA were identified from the hub genes or TFs. The results of the present study demonstrated that 10 hub genes and 12 TFs are the predicted targets for the 5 and 8 differentially expressed miRNAs, respectively. Genes in pink and red modules, which have a high correlation with the clinical trait of days to death, are significantly enriched in 'nucleosome assembly' and 'microtubule-based process', respectively. These results indicated that miR-206, miR-137, miR-153, hub genes and enriched TFs in the pink and red modules exert a potentially pivotal function in the development of LUAD.

Gene selection of rat hepatocyte proliferation using adaptive sparse group lasso with weighted gene co-expression network analysis

Grouped gene selection is the most important task for analyzing the microarray data of rat liver regeneration. Many existing gene selection methods cannot outstand the interactions among the selected genes. In the process of rat liver regeneration, one of the most important events involved in many biological processes is the proliferation of rat hepatocytes, so it can be used as a measure of the effectiveness of the method. Here we proposed an adaptive sparse group lasso to select genes in groups for rat hepatocyte proliferation. The weighted gene co-expression networks analysis was used to identify modules corresponding to gene pathways, based on which a strategy of dividing genes into groups was proposed. A strategy of adaptive gene selection was also presented by assessing the gene significance and introducing the adaptive lasso penalty. Moreover, an improved blockwise descent algorithm was proposed. Experimental results demonstrated that the proposed method can improve the classification accuracy, and select less number of significant genes which act jointly in groups and have direct or indirect effects on rat hepatocyte proliferation. The effectiveness of the method was verified by the method of rat hepatocyte proliferation.

Molecular profiling of mucinous epithelial ovarian cancer by weighted gene co-expression network analysis

PURPOSE

In order to identify the molecular characteristics and improve the efficacy of early diagnosis of mucinous epithelial ovarian cancer (mEOC), here, the transcriptome profiling by weighted gene co-expression network analysis (WGCNA) has been proposed as an effective method.

METHODS

The gene expression dataset GSE26193 was reanalyzed with a systematical approach, WGCNA. mEOC-related gene co-expression modules were detected and the functional enrichments of these modules were performed at GO and KEGG terms. Ten hub genes in the mEOC-related modules were validated using two independent datasets GSE44104 and GSE30274.

RESULTS

11 co-expressed gene modules were identified by WGCNA based on 4917 genes and 99 epithelial ovarian cancer samples. The turquoise module was found to be significantly associated with the subtype of mEOC. KEGG pathway enrichment analysis showed genes in the turquoise module significantly enriched in metabolism of xenobiotics by cytochrome P450 and steroid hormone biosynthesis. Ten hub genes (LIPH, BCAS1, FUT3, ZG16B, PTPRH, SLC4A4, MUC13, TFF1, HNF4G and TFF2) in the turquoise module were validated to be highly expressed in mEOC using two independent gene expression datasets GSE44104 and GSE30274.

CONCLUSION

Our work proposed an applicable framework of molecular characteristics for patients with mEOC, which may help us to obtain a precise and comprehensive understanding on the molecular complexities of mEOC. The hub genes identified in our study, as potential specific biomarkers of mEOC, may be applied in the early diagnosis of mEOC in the future.

Gene co-expression network analysis identifies trait-related modules in Arabidopsis thaliana

A comprehensive network of the Arabidopsis transcriptome was analyzed and may serve as a valuable resource for candidate gene function investigations. A web tool to explore module information was also provided. Arabidopsis thaliana is a widely studied model plant whose transcriptome has been substantially profiled in various tissues, development stages and other conditions. These data can be reused for research on gene function through a systematic analysis of gene co-expression relationships. We collected microarray data from National Center for Biotechnology Information Gene Expression Omnibus, identified modules of co-expressed genes and annotated module functions. These modules were associated with experiments/traits, which provided potential signature modules for phenotypes. Novel heat shock proteins were implicated according to guilt by association. A higher-order module networks analysis suggested that the Arabidopsis network can be further organized into 15 meta-modules and that a chloroplast meta-module has a distinct gene expression pattern from the other 14 meta-modules. A comparison with the rice transcriptome revealed preserved modules and KEGG pathways. All the module gene information was available from an online tool at http://bioinformatics.fafu.edu.cn/arabi/ . Our findings provide a new source for future gene discovery in Arabidopsis.

Molecular windows into the human brain for psychiatric disorders

Delineating the pathophysiology of psychiatric disorders has been extremely challenging but technological advances in recent decades have facilitated a deeper interrogation of molecular processes in the human brain. Initial candidate gene expression studies of the postmortem brain have evolved into genome wide profiling of the transcriptome and the epigenome, a critical regulator of gene expression. Here, we review the potential and challenges of direct molecular characterization of the postmortem human brain, and provide a brief overview of recent transcriptional and epigenetic studies with respect to neuropsychiatric disorders. Such information can now be leveraged and integrated with the growing number of genome-wide association databases to provide a functional context of trait-associated genetic variants linked to psychiatric illnesses and related phenotypes. While it is clear that the field is still developing and challenges remain to be surmounted, these recent advances nevertheless hold tremendous promise for delineating the neurobiological underpinnings of mental diseases and accelerating the development of novel medication strategies.

Integrative Analysis of DiseaseLand Omics Database for Disease Signatures and Treatments: A Bipolar Case Study

Transcriptomics technologies such as next-generation sequencing and microarray platforms provide exciting opportunities for improving diagnosis and treatment of complex diseases. Transcriptomics studies often share similar hypotheses, but are carried out on different platforms, in different conditions, and with different analysis approaches. These factors, in addition to small sample sizes, can result in a lack of reproducibility. A clear understanding and unified picture of many complex diseases are still elusive, highlighting an urgent need to effectively integrate multiple transcriptomic studies for disease signatures. We have integrated more than 3,000 high-quality transcriptomic datasets in oncology, immunology, neuroscience, cardiovascular and metabolic disease, and from both public and internal sources (DiseaseLand database). We established a systematic data integration and meta-analysis approach, which can be applied in multiple disease areas to create a unified picture of the disease signature and prioritize drug targets, pathways, and compounds. In this bipolar case study, we provided an illustrative example using our approach to combine a total of 30 genome-wide gene expression studies using postmortem human brain samples. First, the studies were integrated by extracting raw FASTQ or CEL files, then undergoing the same procedures for preprocessing, normalization, and statistical inference. Second, both p-value and effect size based meta-analysis algorithms were used to identify a total of 204 differentially expressed (DE) genes (FDR < 0.05) genes in the prefrontal cortex. Among these were BDNF, VGF, WFS1, DUSP6, CRHBP, MAOA, and RELN, which have previously been implicated in bipolar disorder. Finally, pathway enrichment analysis revealed a role for GPCR, MAPK, immune, and Reelin pathways. Compound profiling analysis revealed MAPK and other inhibitors may modulate the DE genes. The ability to robustly combine and synthesize the information from multiple studies enables a more powerful understanding of this complex disease.

Single-cell transcriptome-based multilayer network biomarker for predicting prognosis and therapeutic response of gliomas

Occurrence and development of cancers are governed by complex networks of interacting intercellular and intracellular signals. The technology of single-cell RNA sequencing (scRNA-seq) provides an unprecedented opportunity for dissecting the interplay between the cancer cells and the associated microenvironment. Here we combined scRNA-seq data with clinical bulk gene expression data to develop a computational pipeline for identifying the prognostic and predictive signature that connects cancer cells and microenvironmental cells. The pipeline was applied to glioma scRNA-seq data and revealed a tumor-associated microglia/macrophage-mediated EGFR/ERBB2 feedback-crosstalk signaling module, which was defined as a multilayer network biomarker (MNB) to predict survival outcome and therapeutic response of glioma patients. We used publicly available clinical data sets from large cohorts of glioma patients to examine the prognostic significance and predictive accuracy of the MNB, which outperformed conventional gene biomarkers and other methods. Additionally, the MNB was found to be predictive of the sensitivity or resistance of glioma patients to molecularly targeted therapeutics. Moreover, the MNB was an independent and the strongest prognostic factor when adjusted for clinicopathologic risk factors and other existing gene signatures. The robustness of the MNB was further tested on additional data sets. Our study presents a promising scRNA-seq transcriptome-based multilayer network approach to elucidate the interactions between tumor cell and tumor-associated microenvironment and to identify prognostic and predictive signatures of cancer patients. The proposed MNB method may facilitate the design of more effective biomarkers for predicting prognosis and therapeutic resistance of cancer patients.

Moving beyond neurons: the role of cell type-specific gene regulation in Parkinson's disease heritability

Parkinson’s disease (PD), with its characteristic loss of nigrostriatal dopaminergic neurons and deposition of α-synuclein in neurons, is often considered a neuronal disorder. However, in recent years substantial evidence has emerged to implicate glial cell types, such as astrocytes and microglia. In this study, we used stratified LD score regression and expression-weighted cell-type enrichment together with several brain-related and cell-type-specific genomic annotations to connect human genomic PD findings to specific brain cell types. We found that PD heritability attributable to common variation does not enrich in global and regional brain annotations or brain-related cell-type-specific annotations. Likewise, we found no enrichment of PD susceptibility genes in brain-related cell types. In contrast, we demonstrated a significant enrichment of PD heritability in a curated lysosomal gene set highly expressed in astrocytic, microglial, and oligodendrocyte subtypes, and in LoF-intolerant genes, which were found highly expressed in almost all tested cellular subtypes. Our results suggest that PD risk loci do not lie in specific cell types or individual brain regions, but rather in global cellular processes detectable across several cell types.

Defining developmental diversification of diencephalon neurons through single cell gene expression profiling

The embryonic diencephalon forms integration centers and relay stations in the forebrain. Anecdotal expression studies suggest that the diencephalon contains multiple developmental compartments and subdivisions. Here, we utilized single cell RNA sequencing to profile transcriptomes of dissociated cells from the diencephalon of E12.5 mouse embryos. We identified the divergence of different progenitors, intermediate progenitors, and emerging neurons. By mapping the identified cell groups to their spatial origins, we characterized the molecular features of cell types and cell states arising from various diencephalic domains. Furthermore, we reconstructed the developmental trajectory of distinct cell lineages, and thereby identified the genetic cascades and gene regulatory networks underlying the progression of the cell cycle, neurogenesis and cellular diversification. The analysis provides new insights into the molecular mechanisms underlying the amplification of intermediate progenitor cells in the thalamus. The single cell-resolved trajectories not only confirm a close relationship between the rostral thalamus and prethalamus, but also uncover an unexpected close relationship between the caudal thalamus, epithalamus and rostral pretectum. Our data provide a useful resource for systematic studies of cell heterogeneity and differentiation kinetics within the diencephalon.

Understanding Molecular Mechanisms of the Brain Through Transcriptomics

The brain is the most complicated organ in the human body with more than ten thousand genes expressed in each region. The molecular activity of the brain is divergent in various brain regions, both spatially and temporally. The function of each brain region lies in the fact that each region has different gene expression profiles, the possibility of differential RNA splicing, as well as various post-transcriptional and translational modification processes. Understanding the overall activity of the brain at the molecular level is essential for a comprehensive understanding of how the brain works. Fortunately, the development of next generation sequencing technology has made it possible to measure the molecular activity of a specific tissue as a daily routine approach of research. Therefore, at the molecular level, the application of sequencing technology to investigate the molecular organization of the brain has become a novel field, and significant progress has been made recently in this field. In this paper, we reviewed the major computational methods used in the analysis of brain transcriptome, including the application of these methods to the research of human and non-human mammal brains. Finally, we discussed the utilization of transcriptome methods in neurological diseases.

Gene expression shifts in yellow-bellied marmots prior to natal dispersal

The causes and consequences of vertebrate natal dispersal have been studied extensively, yet little is known about the molecular mechanisms involved. We used RNA-seq to quantify transcriptomic gene expression in blood of wild yellow-bellied marmots (Marmota flaviventer) prior to dispersing from or remaining philopatric to their natal colony. We tested 3 predictions. First, we hypothesized dispersers and residents will differentially express genes and gene networks since dispersal is physiologically demanding. Second, we expected differentially expressed genes to be involved in metabolism, circadian processes, and immune function. Finally, in dispersing individuals, we predicted differentially expressed genes would change as a function of sampling date relative to dispersal date. We detected 150 differentially expressed genes, including genes that have critical roles in lipid metabolism and antigen defense. Gene network analysis revealed a module of 126 coexpressed genes associated with dispersal that was enriched for extracellular immune function. Of the dispersal-associated genes, 22 altered expression as a function of days until dispersal, suggesting that dispersal-associated genes do not initiate transcription on the same time scale. Our results provide novel insights into the fundamental molecular changes required for dispersal and suggest evolutionary conservation of functional pathways during this behavioral process.

SinNLRR: a robust subspace clustering method for cell type detection by non-negative and low-rank representation

Motivation

The development of single-cell RNA-sequencing (scRNA-seq) provides a new perspective to study biological problems at the single-cell level. One of the key issues in scRNA-seq analysis is to resolve the heterogeneity and diversity of cells, which is to cluster the cells into several groups. However, many existing clustering methods are designed to analyze bulk RNA-seq data, it is urgent to develop the new scRNA-seq clustering methods. Moreover, the high noise in scRNA-seq data also brings a lot of challenges to computational methods.

Results

In this study, we propose a novel scRNA-seq cell type detection method based on similarity learning, called SinNLRR. The method is motivated by the self-expression of the cells with the same group. Specifically, we impose the non-negative and low rank structure on the similarity matrix. We apply alternating direction method of multipliers to solve the optimization problem and propose an adaptive penalty selection method to avoid the sensitivity to the parameters. The learned similarity matrix could be incorporated with spectral clustering, t-distributed stochastic neighbor embedding for visualization and Laplace score for prioritizing gene markers. In contrast to other scRNA-seq clustering methods, our method achieves more robust and accurate results on different datasets.

Availability and implementation

Our MATLAB implementation of SinNLRR is available at, https://github.com/zrq0123/SinNLRR.

Supplementary information

Supplementary data are available at Bioinformatics online.

Analysis of whole genome-transcriptomic organization in brain to identify genes associated with alcoholism

Alcohol exposure triggers changes in gene expression and biological pathways in human brain. We explored alterations in gene expression in the Pre-Frontal Cortex (PFC) of 65 alcoholics and 73 controls of European descent, and identified 129 genes that showed altered expression (FDR < 0.05) in subjects with alcohol dependence. Differentially expressed genes were enriched for pathways related to interferon signaling and Growth Arrest and DNA Damage-inducible 45 (GADD45) signaling. A coexpression module (thistle2) identified by weighted gene co-expression network analysis (WGCNA) was significantly correlated with alcohol dependence, alcohol consumption, and AUDIT scores. Genes in the thistle2 module were enriched with genes related to calcium signaling pathways and showed significant downregulation of these pathways, as well as enrichment for biological processes related to nicotine response and opioid signaling. A second module (brown4) showed significant upregulation of pathways related to immune signaling. Expression quantitative trait loci (eQTLs) for genes in the brown4 module were also enriched for genetic associations with alcohol dependence and alcohol consumption in large genome-wide studies included in the Psychiatric Genetic Consortium and the UK Biobank's alcohol consumption dataset. By leveraging multi-omics data, this transcriptome analysis has identified genes and biological pathways that could provide insight for identifying therapeutic targets for alcohol dependence.

Differentially expressed gene networks, biomarkers, long noncoding RNAs, and shared responses with cocaine identified in the midbrains of human opioid abusers

Opioid abuse is now the most common cause of accidental death in the US. Although opioids and most other drugs of abuse acutely increase signaling mediated by midbrain dopamine (DA)-synthesizing neurons, little is known about long-lasting changes in DA cells that may contribute to continued opioid abuse, craving, and relapse. A better understanding of the molecular and cellular bases of opioid abuse could lead to advancements in therapeutics. This study comprises, to our knowledge, the first unbiased examination of genome-wide changes in midbrain gene expression associated with human opioid abuse. Our analyses identified differentially expressed genes and distinct gene networks associated with opioid abuse, specific genes with predictive capability for subject assignment to the opioid abuse cohort, and genes most similarly affected in chronic opioid and cocaine abusers. We also identified differentially expressed long noncoding RNAs capable of regulating known drug-responsive protein-coding genes. Opioid-regulated genes identified in this study warrant further investigation as potential biomarkers and/or therapeutic targets for human substance abuse.

Establishing Cerebral Organoids as Models of Human-Specific Brain Evolution

Direct comparisons of human and non-human primate brains can reveal molecular pathways underlying remarkable specializations of the human brain. However, chimpanzee tissue is inaccessible during neocortical neurogenesis when differences in brain size first appear. To identify human-specific features of cortical development, we leveraged recent innovations that permit generating pluripotent stem cell-derived cerebral organoids from chimpanzee. Despite metabolic differences, organoid models preserve gene regulatory networks related to primary cell types and developmental processes. We further identified 261 differentially expressed genes in human compared to both chimpanzee organoids and macaque cortex, enriched for recent gene duplications, and including multiple regulators of PI3K-AKT-mTOR signaling. We observed increased activation of this pathway in human radial glia, dependent on two receptors upregulated specifically in human: INSR and ITGB8. Our findings establish a platform for systematic analysis of molecular changes contributing to human brain development and evolution.

Shared co-expression networks in frontal cortex of the normal aged brain and schizophrenia

Previous studies on the brain of people with schizophrenia have identified structural changes and gene expression changes, suggesting that brain aging maybe accelerated in people with schizophrenia. To better characterize gene expression profiles in schizophrenia and in the aged population we constructed co-expression networks using RNA-Seq data from frontal cortex. The first data set analysed was from 62 subjects with schizophrenia and 51 unaffected controls ranging in age from 19 to 63 years. The second separate data set was from normal control individuals ranging in age from 29 to 106 years. In the first data set, we found two co-expression modules significantly associated with schizophrenia. One was a downregulated co-expression module enriched for neuron function related genes and the other was an upregulated immune/inflammation-related module. In the second data set of normal individuals, we found seven co-expression modules significantly correlated with age. A comparison of the co-expression modules from the two data sets revealed a significant consensus in nodes associated with schizophrenia and those associated with normal aging. The results indicate that a co-expression module related to neuronal function is downregulated and an immune/inflammation related co-expression module is upregulated, and associated with cells of the blood vessels, in both schizophrenia and in normal aging. This finding adds further support to the hypothesis that there may be accelerated brain aging in schizophrenia.

A proteomic network approach across the ALS-FTD disease spectrum resolves clinical phenotypes and genetic vulnerability in human brain

Amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) are neurodegenerative diseases with overlap in clinical presentation, neuropathology, and genetic underpinnings. The molecular basis for the overlap of these disorders is not well established. We performed a comparative unbiased mass spectrometry‐based proteomic analysis of frontal cortical tissues from postmortem cases clinically defined as ALS, FTD, ALS and FTD (ALS/FTD), and controls. We also included a subset of patients with the C9orf72 expansion mutation, the most common genetic cause of both ALS and FTD. Our systems‐level analysis of the brain proteome integrated both differential expression and co‐expression approaches to assess the relationship of these differences to clinical and pathological phenotypes. Weighted co‐expression network analysis revealed 15 modules of co‐expressed proteins, eight of which were significantly different across the ALS‐FTD disease spectrum. These included modules associated with RNA binding proteins, synaptic transmission, and inflammation with cell‐type specificity that showed correlation with TDP‐43 pathology and cognitive dysfunction. Modules were also examined for their overlap with TDP‐43 protein–protein interactions, revealing one module enriched with RNA‐binding proteins and other causal ALS genes that increased in FTD/ALS and FTD cases. A module enriched with astrocyte and microglia proteins was significantly increased in ALS cases carrying the C9orf72 mutation compared to sporadic ALS cases, suggesting that the genetic expansion is associated with inflammation in the brain even without clinical evidence of dementia. Together, these findings highlight the utility of integrative systems‐level proteomic approaches to resolve clinical phenotypes and genetic mechanisms underlying the ALS‐FTD disease spectrum in human brain.

Biological Pathways Leading From ANGPTL8 to Diabetes Mellitus-A Co-expression Network Based Analysis

Angiopoietin like protein 8 (ANGPTL8) is a newly identified hormone with unique nature due to its ability to regulate both glucose and lipid metabolic pathways. It is characterized as an important molecular player of insulin induced nutrient storage and utilization pathway during fasting to re-feeding metabolic transition. Several studies have contributed to increase our knowledge regarding its function and mechanism of action. Moreover, its altered expression levels have been observed in Insulin Resistance, Diabetes Mellitus (Types I & II) and Non Alcohlic Fatty Liver Disease emphasizing its assessment as a drug target. However, there is still a great deal of information that remains to be investigated including its associated biological processes, partner proteins in these processes, its regulators and its association with metabolic pathogenesis. In the current study, the analysis of a transcriptomic data set was performed for functional assessment of ANGPTL8 in liver. Weighted Gene Co-expression Network Analysis coupled with pathway analysis tools was performed to identify genes that are significantly co-expressed with ANGPTL8 in liver and investigate their presence in biological pathways. Gene ontology term enrichment analysis was performed to select the gene ontology classes that over-represent the hepatic ANGPTL8-co-expressed genes. Moreover, the presence of diabetes linked SNPs within the genes set co-expressed with ANGPTL8 was investigated. The co-expressed genes of ANGPTL8 identified in this study (n = 460) provides narrowed down list of molecular targets which are either co-regulated with it and/or might be regulation partners at different levels of interaction. These results are coherent with previously demonstrated roles and regulators of ANGPTL8. Specifically, thirteen co-expressed genes (MAPK8, CYP3A4, PIK3R2, PIK3R4,PRKAB2, G6PC, MAP3K11, FLOT1, PIK3C2G, SHC1, SLC16A2, and RAPGEF1) are also present in the literature curated pathway of ANGPTL8 (WP3915¹). Moreover, the gene-SNP analysis of highly associated biological processes with ANGPTL8 revealed significant genetic signals associated to Diabetes Mellitus and similar phenotypic traits. It provides meaningful insights on the influencing genes involved and co-expressed in these pathways. Findings of this study have implications in functional characterization of ANGPTL8 with emphasis on the identified genes and pathways and their possible involvement in the pathogenesis of Diabetes Mellitus and Insulin Resistance.

A measure of agreement across numerous conditions: assessing when changes in network structures are tissue-specific

BACKGROUND

There is great interest to study how gene pathways change their structure across different tissues. The assessment of inter-study reliability of pathway changes across tissues can inform on the fraction of tissues with specific functional changes in network structure. However, there is a lack of agreement measures among studies that independently observe how a group of observations change across conditions. We, therefore, propose λ, a new inter-study reliability measure that determines the consistency to distinguish observations by condition.

RESULTS

We derived λ's distributional characteristics, determine its reliability properties and compared it with Cohen's κ. We studied the co-expression structure of 287 gene pathways across four brain regions in two transcriptomic studies and applied λ to assess the inter-study reliability of the pathways' brain-regional changes. Brain-related pathways showed highest λ; the top value was for the nicotine addiction pathway whose structure was reliably distinguishable among regions with dopaminergic projections.

CONCLUSION

Our results offer novel substantial evidence that changes in network structure across tissues can be inferred independently of samples, algorithms and experiments (RNA-sequencing or microarrays). Reliability measures, such as λ, can inform on the tissues where changes in a network's structure are likely functional. An R package is available at https://github.com/isglobal-brge/lambda .

Analysis of ambient temperature-responsive transcriptome in shoot apical meristem of heat-tolerant and heat-sensitive broccoli inbred lines during floral head formation

BACKGROUND

Head formation of broccoli (Brassica oleracea var. italica) is greatly reduced under high temperature (22 °C and 27 °C). Broccoli inbred lines that are capable of producing heads at high temperatures in summer are varieties that are unique to Taiwan. However, knowledge of the early-activated pathways of broccoli head formation under high temperature is limited.

RESULTS

We compared heat-tolerant (HT) and heat-sensitive (HS) transcriptome of broccoli under different temperatures. Weighted gene correlation network analysis (WGCNA) revealed that genes involved in calcium signaling pathways, mitogen-activated protein kinase (MAPK) cascades, leucine-rich repeat receptor-like kinases (LRR-RLKs), and genes coding for heat-shock proteins and reactive oxygen species homeostasis shared a similar expression pattern to BoFLC1, which was highly expressed at high temperature (27 °C). Of note, these genes were less expressed in HT than HS broccoli at 22 °C. Co-expression analysis identified a model for LRR-RLKs in survival-reproduction tradeoffs by modulating MAPK- versus phytohormones-signaling during head formation. The difference in head-forming ability in response to heat stress between HT and HS broccoli may result from their differential transcriptome profiles of LRR-RLK genes. High temperature induced JA- as well as suppressed auxin- and cytokinin-related pathways may facilitate a balancing act to ensure fitness at 27 °C. BoFLC1 was less expressed in HT than HS at 22 °C, whereas other FLC homologues were not. Promoter analysis of BoFLC1 showed fewer AT dinucleotide repeats in HT broccoli. These results provide insight into the early activation of stress- or development-related pathways during head formation in broccoli. The identification of the BoFLC1 DNA biomarker may facilitate breeding of HT broccoli.

CONCLUSIONS

In this study, HT and HS broccoli genotypes were used to determine the effect of temperature on head formation by transcriptome profiling. On the basis of the expression pattern of high temperature-associated signaling genes, the HS transcriptome may be involved in stress defense instead of transition to the reproductive phase in response to heat stress. Transcriptome profiling of HT and HS broccoli helps in understanding the molecular mechanisms underlying head-forming capacity and in promoting functional marker-assisted breeding.

Effects of APOE Genotype on Brain Proteomic Network and Cell Type Changes in Alzheimer's Disease

Polymorphic alleles in the apolipoprotein E (APOE) gene are the main genetic determinants of late-onset Alzheimer's disease (AD) risk. Individuals carrying the APOE E4 allele are at increased risk to develop AD compared to those carrying the more common E3 allele, whereas those carrying the E2 allele are at decreased risk for developing AD. How ApoE isoforms influence risk for AD remains unclear. To help fill this gap in knowledge, we performed a comparative unbiased mass spectrometry-based proteomic analysis of post-mortem brain cortical tissues from pathologically-defined AD or control cases of different APOE genotypes. Control cases (n = 10) were homozygous for the common E3 allele, whereas AD cases (n = 24) were equally distributed among E2/3, E3/3, and E4/4 genotypes. We used differential protein expression and co-expression analytical approaches to assess how changes in the brain proteome are related to APOE genotype. We observed similar levels of amyloid-β, but reduced levels of neurofibrillary tau, in E2/3 brains compared to E3/3 and E4/4 AD brains. Weighted co-expression network analysis revealed 33 modules of co-expressed proteins, 12 of which were significantly different by APOE genotype in AD. The modules that were significantly different by APOE genotype were associated with synaptic transmission and inflammation, among other biological processes. Deconvolution and analysis of brain cell type changes revealed that the E2 allele suppressed homeostatic and disease-associated cell type changes in astrocytes, microglia, oligodendroglia, and endothelia. The E2 allele-specific effect on brain cell type changes was validated in a separate cohort of 130 brains. Our systems-level proteomic analyses of AD brain reveal alterations in the brain proteome and brain cell types associated with allelic variants in APOE, and suggest further areas for investigation into the upstream mechanisms that drive ApoE-associated risk for AD.

Comprehensive functional genomic resource and integrative model for the human brain

Despite progress in defining genetic risk for psychiatric disorders, their molecular mechanisms remain elusive. Addressing this, the PsychENCODE Consortium has generated a comprehensive online resource for the adult brain across 1866 individuals. The PsychENCODE resource contains ~79,000 brain-active enhancers, sets of Hi-C linkages, and topologically associating domains; single-cell expression profiles for many cell types; expression quantitative-trait loci (QTLs); and further QTLs associated with chromatin, splicing, and cell-type proportions. Integration shows that varying cell-type proportions largely account for the cross-population variation in expression (with >88% reconstruction accuracy). It also allows building of a gene regulatory network, linking genome-wide association study variants to genes (e.g., 321 for schizophrenia). We embed this network into an interpretable deep-learning model, which improves disease prediction by ~6-fold versus polygenic risk scores and identifies key genes and pathways in psychiatric disorders.

Transcriptome-wide isoform-level dysregulation in ASD, schizophrenia, and bipolar disorder

Most genetic risk for psychiatric disease lies in regulatory regions, implicating pathogenic dysregulation of gene expression and splicing. However, comprehensive assessments of transcriptomic organization in diseased brains are limited. In this work, we integrated genotypes and RNA sequencing in brain samples from 1695 individuals with autism spectrum disorder (ASD), schizophrenia, and bipolar disorder, as well as controls. More than 25% of the transcriptome exhibits differential splicing or expression, with isoform-level changes capturing the largest disease effects and genetic enrichments. Coexpression networks isolate disease-specific neuronal alterations, as well as microglial, astrocyte, and interferon-response modules defining previously unidentified neural-immune mechanisms. We integrated genetic and genomic data to perform a transcriptome-wide association study, prioritizing disease loci likely mediated by cis effects on brain expression. This transcriptome-wide characterization of the molecular pathology across three major psychiatric disorders provides a comprehensive resource for mechanistic insight and therapeutic development.

Dopamine perturbation of gene co-expression networks reveals differential response in schizophrenia for translational machinery

The dopaminergic hypothesis of schizophrenia (SZ) postulates that positive symptoms of SZ, in particular psychosis, are due to disturbed neurotransmission via the dopamine (DA) receptor D2 (DRD2). However, DA is a reactive molecule that yields various oxidative species, and thus has important non-receptor-mediated effects, with empirical evidence of cellular toxicity and neurodegeneration. Here we examine non-receptor-mediated effects of DA on gene co-expression networks and its potential role in SZ pathology. Transcriptomic profiles were measured by RNA-seq in B-cell transformed lymphoblastoid cell lines from 514 SZ cases and 690 controls, both before and after exposure to DA ex vivo (100 μM). Gene co-expression modules were identified using Weighted Gene Co-expression Network Analysis for both baseline and DA-stimulated conditions, with each module characterized for biological function and tested for association with SZ status and SNPs from a genome-wide panel. We identified seven co-expression modules under baseline, of which six were preserved in DA-stimulated data. One module shows significantly increased association with SZ after DA perturbation (baseline: P = 0.023; DA-stimulated: P = 7.8 × 10^-5; ΔAIC = -10.5) and is highly enriched for genes related to ribosomal proteins and translation (FDR = 4 × 10^-141), mitochondrial oxidative phosphorylation, and neurodegeneration. SNP association testing revealed tentative QTLs underlying module co-expression, notably at FASTKD2 (top P = 2.8 × 10^-6), a gene involved in mitochondrial translation. These results substantiate the role of translational machinery in SZ pathogenesis, providing insights into a possible dopaminergic mechanism disrupting mitochondrial function, and demonstrates the utility of disease-relevant functional perturbation in the study of complex genetic etiologies.

Transcriptomic Evidence for Alterations in Astrocytes and Parvalbumin Interneurons in Subjects With Bipolar Disorder and Schizophrenia

BACKGROUND

High-throughput expression analyses of postmortem brain tissue have been widely used to study bipolar disorder and schizophrenia. However, despite the extensive efforts, no consensus has emerged as to the functional interpretation of the findings. We hypothesized that incorporating information on cell type-specific expression would provide new insights.

METHODS

We reanalyzed 15 publicly available bulk tissue expression datasets on schizophrenia and bipolar disorder, representing various brain regions from eight different cohorts of subjects (unique subjects: 332 control, 129 bipolar disorder, 341 schizophrenia). We studied changes in the expression profiles of cell type marker genes and evaluated whether these expression profiles could serve as surrogates for relative abundance of their corresponding cells.

RESULTS

In both bipolar disorder and schizophrenia, we consistently observed an increase in the expression profiles of cortical astrocytes and a decrease in the expression profiles of fast-spiking parvalbumin interneurons. No changes in astrocyte expression profiles were observed in subcortical regions. Furthermore, we found that many of the genes previously identified as differentially expressed in schizophrenia are highly correlated with the expression profiles of astrocytes or fast-spiking parvalbumin interneurons.

CONCLUSIONS

Our results indicate convergence of transcriptome studies of schizophrenia and bipolar disorder on changes in cortical astrocytes and fast-spiking parvalbumin interneurons, providing a unified interpretation of numerous studies. We suggest that these changes can be attributed to alterations in the relative abundance of the cells and are important for understanding the pathophysiology of the disorders.

Postmitotic cell longevity-associated genes: a transcriptional signature of postmitotic maintenance in neural tissues

Different cell types have different postmitotic maintenance requirements. Nerve cells, however, are unique in this respect as they need to survive and preserve their functional complexity for the entire lifetime of the organism, and failure at any level of their supporting mechanisms leads to a wide range of neurodegenerative conditions. Whether these differences across tissues arise from the activation of distinct cell type-specific maintenance mechanisms or the differential activation of a common molecular repertoire is not known. To identify the transcriptional signature of postmitotic cellular longevity (PMCL), we compared whole-genome transcriptome data from human tissues ranging in longevity from 120 days to over 70 years and found a set of 81 genes whose expression levels are closely associated with increased cell longevity. Using expression data from 10 independent sources, we found that these genes are more highly coexpressed in longer-living tissues and are enriched in specific biological processes and transcription factor targets compared with randomly selected gene samples. Crucially, we found that PMCL-associated genes are downregulated in the cerebral cortex and substantia nigra of patients with Alzheimer's and Parkinson's disease, respectively, as well as Hutchinson-Gilford progeria-derived fibroblasts, and that this downregulation is specifically linked to their underlying association with cellular longevity. Moreover, we found that sexually dimorphic brain expression of PMCL-associated genes reflects sexual differences in lifespan in humans and macaques. Taken together, our results suggest that PMCL-associated genes are part of a generalized machinery of postmitotic maintenance and functional stability in both neural and non-neural cells and support the notion of a common molecular repertoire differentially engaged in different cell types with different survival requirements.

Emerging Biomarkers in Bladder Cancer Identified by Network Analysis of Transcriptomic Data

Bladder cancer is a very common malignancy. Although new treatment strategies have been developed, the identification of new therapeutic targets and reliable diagnostic/prognostic biomarkers for bladder cancer remains a priority. Generally, they are found among differentially expressed genes between patients and healthy subjects or among patients with different tumor stages. However, the classical approach includes processing these data taking into consideration only the expression of each single gene regardless of the expression of other genes. These complex gene interaction networks can be revealed by a recently developed systems biology approach called Weighted Gene Co-expression Network Analysis (WGCNA). It takes into account the expression of all genes assessed in an experiment in order to reveal the clusters of co-expressed genes (modules) that, very probably, are also co-regulated. If some genes are co-expressed in controls but not in pathological samples, it can be hypothesized that a regulatory mechanism was altered and that it could be the cause or the effect of the disease. Therefore, genes within these modules could play a role in cancer and thus be considered as potential therapeutic targets or diagnostic/prognostic biomarkers. Here, we have reviewed all the studies where WGCNA has been applied to gene expression data from bladder cancer patients. We have shown the importance of this new approach in identifying candidate biomarkers and therapeutic targets. They include both genes and miRNAs and some of them have already been identified in the literature to have a role in bladder cancer initiation, progression, metastasis, and patient survival.

Deep proteomic network analysis of Alzheimer's disease brain reveals alterations in RNA binding proteins and RNA splicing associated with disease

BACKGROUND

The complicated cellular and biochemical changes that occur in brain during Alzheimer's disease are poorly understood. In a previous study we used an unbiased label-free quantitative mass spectrometry-based proteomic approach to analyze these changes at a systems level in post-mortem cortical tissue from patients with Alzheimer's disease (AD), asymptomatic Alzheimer's disease (AsymAD), and controls. We found modules of co-expressed proteins that correlated with AD phenotypes, some of which were enriched in proteins identified as risk factors for AD by genetic studies.

METHODS

The amount of information that can be obtained from such systems-level proteomic analyses is critically dependent upon the number of proteins that can be quantified across a cohort. We report here a new proteomic systems-level analysis of AD brain based on 6,533 proteins measured across AD, AsymAD, and controls using an analysis pipeline consisting of isobaric tandem mass tag (TMT) mass spectrometry and offline prefractionation.

RESULTS

Our new TMT pipeline allowed us to more than double the depth of brain proteome coverage. This increased depth of coverage greatly expanded the brain protein network to reveal new protein modules that correlated with disease and were unrelated to those identified in our previous network. Differential protein abundance analysis identified 350 proteins that had altered levels between AsymAD and AD not caused by changes in specific cell type abundance, potentially reflecting biochemical changes that are associated with cognitive decline in AD. RNA binding proteins emerged as a class of proteins altered between AsymAD and AD, and were enriched in network modules that correlated with AD pathology. We developed a proteogenomic approach to investigate RNA splicing events that may be altered by RNA binding protein changes in AD. The increased proteome depth afforded by our TMT pipeline allowed us to identify and quantify a large number of alternatively spliced protein isoforms in brain, including AD risk factors such as BIN1, PICALM, PTK2B, and FERMT2. Many of the new AD protein network modules were enriched in alternatively spliced proteins and correlated with molecular markers of AD pathology and cognition.

CONCLUSIONS

Further analysis of the AD brain proteome will continue to yield new insights into the biological basis of AD.

Peroxisome Proliferator Activated Receptor Agonists Modulate Transposable Element Expression in Brain and Liver

Peroxisome proliferator activated receptors (PPARs) are nuclear hormone receptors that act as transcription factors in response to endogenous lipid messengers. The fibrates and thiazolidinediones are synthetic PPAR agonists used clinically to treat dyslipidemia and Type 2 Diabetes Mellitus, respectively, but also improve symptoms of several other diseases. Transposable elements (TEs), repetitive sequences in mammalian genomes, are implicated in many of the same conditions for which PPAR agonists are therapeutic, including neurodegeneration, schizophrenia, and drug addiction. We tested the hypothesis that there is a link between actions of PPAR agonists and TE expression. We developed an innovative application of microarray data by mapping Illumina mouse WG-6 microarray probes to areas of the mouse genome that contain TEs. Using this information, we assessed the effects of systemic administration of three PPAR agonists with different PPAR subtype selectivity: fenofibrate, tesaglitazar, and bezafibrate, on TE probe expression in mouse brain [prefrontal cortex (PFC) and amygdala] and liver. We found that fenofibrate, and bezafibrate to a lesser extent, up-regulated probes mapped to retrotransposons: Short-Interspersed Elements (SINEs) and Long-Interspersed Elements (LINEs), in the PFC. Conversely, all PPAR agonists down-regulated LINEs and tesaglitazar and bezafibrate also down-regulated SINEs in liver. We built gene coexpression networks that partitioned the diverse transcriptional response to PPAR agonists into groups of probes with highly correlated expression patterns (modules). Most of the differentially expressed retrotransposons were within the same module, suggesting coordinated regulation of their expression, possibly by PPAR signaling. One TE module was conserved across tissues and was enriched with genes whose products participate in epigenetic regulation, suggesting that PPAR agonists affect TE expression via epigenetic mechanisms. Other enriched functional categories included phenotypes related to embryonic development and learning and memory, suggesting functional links between these biological processes and TE expression. In summary, these findings suggest mechanistic relationships between retrotransposons and PPAR agonists and provide a basis for future exploration of their functional roles in brain and liver.

A Comprehensive Survey of Immune Cytolytic Activity-Associated Gene Co-Expression Networks across 17 Tumor and Normal Tissue Types

Cytolytic immune activity in solid tissue can be quantified by transcript levels of two genes, GZMA and PRF1, which is named the CYT score. A previous study has investigated the molecular and genetic properties of tumors associated CYT, but a systematic exploration of how co-expression networks across different tumors are shaped by anti-tumor immunity is lacking. Here, we examined the connectivity and biological themes of CYT-associated modules in gene co-expression networks of 14 tumor and 3 matched normal tissues constructed from the RNA-Seq data of the "The Cancer Genome Atlas" project. We first found that tumors networks have more diverse CYT-correlated modules than normal networks. We next identified and investigated tissue-specific CYT-associated modules across 14 tumor types. Finally, a common CYT-associated network across 14 tumor types was constructed. Two common modules have mixed signs of correlation with CYT in different tumors. Given the tumors and normal tissues surveyed, our study presents a systematic view of the regulation of cytolytic immune activity across multiple tumor tissues.

Variation among intact tissue samples reveals the core transcriptional features of human CNS cell classes

It is widely assumed that cells must be physically isolated to study their molecular profiles. However, intact tissue samples naturally exhibit variation in cellular composition, which drives covariation of cell-class-specific molecular features. By analyzing transcriptional covariation in 7,221 intact CNS samples from 840 neurotypical individuals, representing billions of cells, we reveal the core transcriptional identities of major CNS cell classes in humans. By modeling intact CNS transcriptomes as a function of variation in cellular composition, we identify cell-class-specific transcriptional differences in Alzheimer's disease, among brain regions, and between species. Among these, we show that PMP2 is expressed by human but not mouse astrocytes and significantly increases mouse astrocyte size upon ectopic expression in vivo, causing them to more closely resemble their human counterparts. Our work is available as an online resource ( http://oldhamlab.ctec.ucsf.edu/ ) and provides a generalizable strategy for determining the core molecular features of cellular identity in intact biological systems.

Gene co-expression network analysis reveals coordinated regulation of three characteristic secondary biosynthetic pathways in tea plant (Camellia sinensis)

BACKGROUND

The leaves of tea plants (Camellia sinensis) are used to produce tea, which is one of the most popular beverages consumed worldwide. The nutritional value and health benefits of tea are mainly related to three abundant characteristic metabolites; catechins, theanine and caffeine. Weighted gene co-expression network analysis (WGCNA) is a powerful system for investigating correlations between genes, identifying modules among highly correlated genes, and relating modules to phenotypic traits based on gene expression profiling. Currently, relatively little is known about the regulatory mechanisms and correlations between these three secondary metabolic pathways at the omics level in tea.

RESULTS

In this study, levels of the three secondary metabolites in ten different tissues of tea plants were determined, 87,319 high-quality unigenes were assembled, and 55,607 differentially expressed genes (DEGs) were identified by pairwise comparison. The resultant co-expression network included 35 co-expression modules, of which 20 modules were significantly associated with the biosynthesis of catechins, theanine and caffeine. Furthermore, we identified several hub genes related to these three metabolic pathways, and analysed their regulatory relationships using RNA-Seq data. The results showed that these hub genes are regulated by genes involved in all three metabolic pathways, and they regulate the biosynthesis of all three metabolites. It is notable that light was identified as an important regulator for the biosynthesis of catechins.

CONCLUSION

Our integrated omics-level WGCNA analysis provides novel insights into the potential regulatory mechanisms of catechins, theanine and caffeine metabolism, and the identified hub genes provide an important reference for further research on the molecular biology of tea plants.

Sequencing Diversity One Cell at a Time

Single-cell RNA sequencing provides a new approach to an old problem: how to study cellular diversity in complex biological systems. Three studies-Saunders et al., Zeisel et al., and Davie et al.-deploy this technique on an unprecedented scale to reveal transcriptional patterns that distinguish cells in the nervous systems of mice and flies.

The age-regulated zinc finger factor ZNF367 is a new modulator of neuroblast proliferation during embryonic neurogenesis

Global population aging is one of the major social and economic challenges of contemporary society. During aging the progressive decline in physiological functions has serious consequences for all organs including brain. The age-related incidence of neurodegenerative diseases coincides with the sharp decline of the amount and functionality of adult neural stem cells. Recently, we identified a short list of brain age-regulated genes by means of next-generation sequencing. Among them znf367 codes for a transcription factor that represents a central node in gene co-regulation networks during aging, but whose function in the central nervous system (CNS), is completely unknown. As proof of concept, we analysed the role of znf367 during Xenopus laevis neurogenesis. By means of a gene loss of function approach limited to the CNS, we suggested that znf367 might act as a key controller of the neuroblast cell cycle, particularly in the progression of mitosis and spindle checkpoint. A candidate gene approach based on a weighted-gene co-expression network analysis, revealed fancd2 and ska3 as possible targets of znf367. The age-related decline of znf367 correlated well with its role during embryonic neurogenesis, opening new lines of investigation also in adult neurogenesis to improved maintenance and even repair of neuronal function.

Functional Annotation of Caenorhabditis elegans Genes by Analysis of Gene Co-Expression Networks

Caenorhabditis elegans (C. elegans) is a well-characterized metazoan, whose transcriptome has been profiled in different tissues, development stages, or other conditions. Large-scale transcriptomes can be reused for gene function annotation through systematic analysis of gene co-expression relationships. We collected 2101 microarray data from National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO), and identified 48 modules of co-expressed genes that correspond to tissues, development stages, and other experimental conditions. These modules provide an overview of the transcriptional organizations that may work under different conditions. By analyzing higher-order module networks, we found that nucleus and plasma membrane modules are more connected than other intracellular modules. Module-based gene function annotation may help to extend the candidate cuticle gene list. A comparison with other published data validates the credibility of our result. Our findings provide a new source for future gene discovery in C. elegans.

Coexpression network revealing the plasticity and robustness of population transcriptome during the initial stage of domesticating energy crop Miscanthus lutarioriparius

Coexpression network revealing genes with Co-variation Expression pattern (CE) and those with Top rank of Expression fold change (TE) played different roles in responding to new environment of Miscanthus lutarioriparius. Variation in gene expression level, the product of genetic and/or environmental perturbation, determines the robustness-to-plasticity spectrum of a phenotype in plants. Understanding how expression variation of plant population response to a new field is crucial to domesticate energy crops. Weighted Gene Coexpression Network Analysis (WGCNA) was used to explore the patterns of expression variation based on 72 Miscanthus lutarioriparius transcriptomes from two contrasting environments, one near the native habitat and the other in one harsh domesticating region. The 932 genes with Co-variation Expression pattern (CE) and other 932 genes with Top rank of Expression fold change (TE) were identified and the former were strongly associated with the water use efficiency (r ≥ 0.55, P ≤ 10^-7). Functional enrichment of CE genes were related to three organelles, which well matched the annotation of twelve motifs identified from their conserved noncoding sequence; while TE genes were mostly related to biotic and/or abiotic stress. The expression robustness of CE genes with high genetic diversity kept relatively stable between environments while the harsh environment reduced the expression robustness of TE genes with low genetic diversity. The expression plasticity of CE genes was increased less than that of TE genes. These results suggested that expression variation of CE genes and TE genes could account for the robustness and plasticity of acclimation ability of Miscanthus, respectively. The patterns of expression variation revealed by transcriptomic network would shed new light on breeding and domestication of energy crops.