Topological analysis of metabolic networks integrating co-segregating transcriptomes and metabolomes in type 2 diabetic rat congenic series.

BACKGROUND: The genetic regulation of metabolic phenotypes (i.e., metabotypes) in type 2 diabetes mellitus occurs through complex organ-specific cellular mechanisms and networks contributing to impaired insulin secretion and insulin resistance. Genome-wide gene expression profiling systems can dissect the genetic contributions to metabolome and transcriptome regulations. The integrative analysis of multiple gene expression traits and metabolic phenotypes (i.e., metabotypes) together with their underlying genetic regulation remains a challenge. Here, we introduce a systems genetics approach based on the topological analysis of a combined molecular network made of genes and metabolites identified through expression and metabotype quantitative trait locus mapping (i.e., eQTL and mQTL) to prioritise biological characterisation of candidate genes and traits. METHODS: We used systematic metabotyping by 1H NMR spectroscopy and genome-wide gene expression in white adipose tissue to map molecular phenotypes to genomic blocks associated with obesity and insulin secretion in a series of rat congenic strains derived from spontaneously diabetic Goto-Kakizaki (GK) and normoglycemic Brown-Norway (BN) rats. We implemented a network biology strategy approach to visualize the shortest paths between metabolites and genes significantly associated with each genomic block. RESULTS: Despite strong genomic similarities (95-99 %) among congenics, each strain exhibited specific patterns of gene expression and metabotypes, reflecting the metabolic consequences of series of linked genetic polymorphisms in the congenic intervals. We subsequently used the congenic panel to map quantitative trait loci underlying specific mQTLs and genome-wide eQTLs. Variation in key metabolites like glucose, succinate, lactate, or 3-hydroxybutyrate and second messenger precursors like inositol was associated with several independent genomic intervals, indicating functional redundancy in these regions. To navigate through the complexity of these association networks we mapped candidate genes and metabolites onto metabolic pathways and implemented a shortest path strategy to highlight potential mechanistic links between metabolites and transcripts at colocalized mQTLs and eQTLs. Minimizing the shortest path length drove prioritization of biological validations by gene silencing. CONCLUSIONS: These results underline the importance of network-based integration of multilevel systems genetics datasets to improve understanding of the genetic architecture of metabotype and transcriptomic regulation and to characterize novel functional roles for genes determining tissue-specific metabolism.

Transcriptome Profiling in Rat Inbred Strains and Experimental Cross Reveals Discrepant Genetic Architecture of Genome-Wide Gene Expression.

To test the impact of genetic heterogeneity on cis- and trans-mediated mechanisms of gene expression regulation, we profiled the transcriptome of adipose tissue in 20 inbred congenic strains derived from diabetic Goto-Kakizaki (GK) rats and Brown-Norway (BN) controls, which contain well-defined blocks (1-183 Mb) of genetic polymorphisms, and in 123 genetically heterogeneous rats of an (GK × BN)F2 offspring. Within each congenic we identified 73-1351 differentially expressed genes (DEGs), only 7.7% of which mapped within the congenic blocks, and which may be regulated in cis The remainder localized outside the blocks, and therefore must be regulated in trans Most trans-regulated genes exhibited approximately twofold expression changes, consistent with monoallelic expression. Altered biological pathways were replicated between congenic strains sharing blocks of genetic polymorphisms, but polymorphisms at different loci also had redundant effects on transcription of common distant genes and pathways. We mapped 2735 expression quantitative trait loci (eQTL) in the F2 cross, including 26% predominantly cis-regulated genes, which validated DEGs in congenic strains. A hotspot of >300 eQTL in a 10 cM region of chromosome 1 was enriched in DEGs in a congenic strain. However, many DEGs among GK, BN and congenic strains did not replicate as eQTL in F2 hybrids, demonstrating distinct mechanisms of gene expression when alleles segregate in an outbred population or are fixed homozygous across the entire genome or in short genomic regions. Our analysis provides conceptual advances in our understanding of the complex architecture of genome expression and pathway regulation, and suggests a prominent impact of epistasis and monoallelic expression on gene transcription.

Assessing similarity to primary tissue and cortical layer identity in induced pluripotent stem cell-derived cortical neurons through single-cell transcriptomics.

Induced pluripotent stem cell (iPSC)-derived cortical neurons potentially present a powerful new model to understand corticogenesis and neurological disease. Previous work has established that differentiation protocols can produce cortical neurons, but little has been done to characterize these at cellular resolution. In particular, it is unclear to what extent in vitro two-dimensional, relatively disordered culture conditions recapitulate the development of in vivo cortical layer identity. Single-cell multiplex reverse transcriptase-quantitative polymerase chain reaction (RT-qPCR) was used to interrogate the expression of genes previously implicated in cortical layer or phenotypic identity in individual cells. Totally, 93.6% of single cells derived from iPSCs expressed genes indicative of neuronal identity. High proportions of single neurons derived from iPSCs expressed glutamatergic receptors and synaptic genes. And, 68.4% of iPSC-derived neurons expressing at least one layer marker could be assigned to a laminar identity using canonical cortical layer marker genes. We compared single-cell RNA-seq of our iPSC-derived neurons to available single-cell RNA-seq data from human fetal and adult brain and found that iPSC-derived cortical neurons closely resembled primary fetal brain cells. Unexpectedly, a subpopulation of iPSC-derived neurons co-expressed canonical fetal deep and upper cortical layer markers. However, this appeared to be concordant with data from primary cells. Our results therefore provide reassurance that iPSC-derived cortical neurons are highly similar to primary cortical neurons at the level of single cells but suggest that current layer markers, although effective, may not be able to disambiguate cortical layer identity in all cells.

Genetic control of differential acetylation in diabetic rats.

Post-translational protein modifications such as acetylation have significant regulatory roles in metabolic processes, but their relationship to both variation in gene expression and DNA sequence is unclear. We address this question in the Goto-Kakizaki (GK) rat inbred strain, a model of polygenic type 2 diabetes. Expression of the NAD-dependent deacetylase Sirtuin-3 is down-regulated in GK rats compared to normoglycemic Brown Norway (BN) rats. We show first that a promoter SNP causes down-regulation of Sirtuin-3 expression in GK rats. We then use mass-spectrometry to identify proteome-wide differential lysine acetylation of putative Sirtuin-3 protein targets in livers of GK and BN rats. These include many proteins in pathways connected to diabetes and metabolic syndrome. We finally sequence GK and BN liver transcriptomes and find that mRNA expression of these targets does not differ significantly between GK and BN rats, in contrast to other components of the same pathways. We conclude that physiological differences between GK and BN rats are mediated by a combination of differential protein acetylation and gene transcription and that genetic variation can modulate acetylation independently of expression.

A modified RNA-Seq approach for whole genome sequencing of RNA viruses from faecal and blood samples.

To date, very large scale sequencing of many clinically important RNA viruses has been complicated by their high population molecular variation, which creates challenges for polymerase chain reaction and sequencing primer design. Many RNA viruses are also difficult or currently not possible to culture, severely limiting the amount and purity of available starting material. Here, we describe a simple, novel, high-throughput approach to Norovirus and Hepatitis C virus whole genome sequence determination based on RNA shotgun sequencing (also known as RNA-Seq). We demonstrate the effectiveness of this method by sequencing three Norovirus samples from faeces and two Hepatitis C virus samples from blood, on an Illumina MiSeq benchtop sequencer. More than 97% of reference genomes were recovered. Compared with Sanger sequencing, our method had no nucleotide differences in 14,019 nucleotides (nt) for Noroviruses (from a total of 2 Norovirus genomes obtained with Sanger sequencing), and 8 variants in 9,542 nt for Hepatitis C virus (1 variant per 1,193 nt). The three Norovirus samples had 2, 3, and 2 distinct positions called as heterozygous, while the two Hepatitis C virus samples had 117 and 131 positions called as heterozygous. To confirm that our sample and library preparation could be scaled to true high-throughput, we prepared and sequenced an additional 77 Norovirus samples in a single batch on an Illumina HiSeq 2000 sequencer, recovering >90% of the reference genome in all but one sample. No discrepancies were observed across 118,757 nt compared between Sanger and our custom RNA-Seq method in 16 samples. By generating viral genomic sequences that are not biased by primer-specific amplification or enrichment, this method offers the prospect of large-scale, affordable studies of RNA viruses which could be adapted to routine diagnostic laboratory workflows in the near future, with the potential to directly characterize within-host viral diversity.

Untargeted metabolome quantitative trait locus mapping associates variation in urine glycerate to mutant glycerate kinase.

With successes of genome-wide association studies, molecular phenotyping systems are developed to identify genetically determined disease-associated biomarkers. Genetic studies of the human metabolome are emerging but exclusively apply targeted approaches, which restricts the analysis to a limited number of well-known metabolites. We have developed novel technical and statistical methods for systematic and automated quantification of untargeted NMR spectral data designed to perform robust and accurate quantitative trait locus (QTL) mapping of known and previously unreported molecular compounds of the metabolome. For each spectral peak, six summary statistics were calculated and independently tested for evidence of genetic linkage in a cohort of F2 (129S6xBALB/c) mice. The most significant evidence of linkages were obtained with NMR signals characterizing the glycerate (LOD10-42) at the mutant glycerate kinase locus, which demonstrate the power of metabolomics in quantitative genetics to identify the biological function of genetic variants. These results provide new insights into the resolution of the complex nature of metabolic regulations and novel analytical techniques that maximize the full utilization of metabolomic spectra in human genetics to discover mappable disease-associated biomarkers.

Chromosomal mapping of pancreatic islet morphological features and regulatory hormones in the spontaneously diabetic (Type 2) Goto-Kakizaki rat.

Insulin resistance and altered endocrine pancreas function are central pathophysiological features of type 2 diabetes mellitus (T2DM). The Goto-Kakizaki (GK) rat is a model of spontaneous T2DM characterised by reduced beta cell mass and genetically determined glucose intolerance and altered insulin secretion. To identify genetic determinants of endocrine pancreas histopathology, we carried out quantitative trait locus (QTL) mapping of histological phenotypes (beta cell mass -BCM and insulin-positive cell area -IPCA) and plasma concentration of hormones and growth factors in a F2 cohort derived from GK and normoglycemic Brown Norway rats. Although IPCA and BCM in the duodenal region of the pancreas were highly positively correlated (P < 10(-6)), and similarly in the splenic region, both measures were poorly correlated when comparing duodenal and splenic phenotypes. Strongest evidence of linkage to pancreas morphological traits was obtained between BCM and chromosome 10 (LOD 3.2). Evidence of significant linkage (LOD 4.2) to plasma corticosterone was detected in a region of chromosome 1 distal to other QTLs previously identified in the GK. Male-specific genetic effects were detected, including linkages (LOD > 4) to growth hormome (GH) on chromosome 6 and prolactin on chromosome 17. These data suggest independent genetic control of the structure and function of ontologically different regions of the endocrine pancreas. Novel QTLs for corticosterone, prolactin and GH may contribute to diabetes in the GK. The QTLs that we have identified in this, and previous genetic studies collectively underline the complex and multiple mechanisms involved in diabetes in the GK strain.

Functional annotations of diabetes nephropathy susceptibility loci through analysis of genome-wide renal gene expression in rat models of diabetes mellitus.

BACKGROUND: Hyperglycaemia in diabetes mellitus (DM) alters gene expression regulation in various organs and contributes to long term vascular and renal complications. We aimed to generate novel renal genome-wide gene transcription data in rat models of diabetes in order to test the responsiveness to hyperglycaemia and renal structural changes of positional candidate genes at selected diabetic nephropathy (DN) susceptibility loci. METHODS: Both Affymetrix and Illumina technologies were used to identify significant quantitative changes in the abundance of over 15,000 transcripts in kidney of models of spontaneous (genetically determined) mild hyperglycaemia and insulin resistance (Goto-Kakizaki-GK) and experimentally induced severe hyperglycaemia (Wistar-Kyoto-WKY rats injected with streptozotocin [STZ]). RESULTS: Different patterns of transcription regulation in the two rat models of diabetes likely underlie the roles of genetic variants and hyperglycaemia severity. The impact of prolonged hyperglycaemia on gene expression changes was more profound in STZ-WKY rats than in GK rats and involved largely different sets of genes. These included genes already tested in genetic studies of DN and a large number of protein coding sequences of unknown function which can be considered as functional and, when they map to DN loci, positional candidates for DN. Further expression analysis of rat orthologs of human DN positional candidate genes provided functional annotations of known and novel genes that are responsive to hyperglycaemia and may contribute to renal functional and/or structural alterations. CONCLUSION: Combining transcriptomics in animal models and comparative genomics provides important information to improve functional annotations of disease susceptibility loci in humans and experimental support for testing candidate genes in human genetics.

Comparative analysis of methods for gene transcription profiling data derived from different microarray technologies in rat and mouse models of diabetes.

BACKGROUND: Microarray technologies are widely used to quantify the abundance of transcripts corresponding to thousands of genes. To maximise the robustness of transcriptome results, we have tested the performance and reproducibility of rat and mouse gene expression data obtained with Affymetrix, Illumina and Operon platforms. RESULTS: We present a thorough analysis of the degree of reproducibility provided by analysing the transcriptomic profile of the same animals of several experimental groups under different popular microarray technologies in different tissues. Concordant results from inter- and intra-platform comparisons were maximised by testing many popular computational methods for generating fold changes and significances and by only considering oligonucleotides giving high expression levels. The choice of Affymetrix signal extraction technique was shown to have the greatest effect on the concordance across platforms. In both species, when choosing optimal methods, the agreement between data generated on the Affymetrix and Illumina was excellent; this was verified using qRT-PCR on a selection of genes present on all platforms. CONCLUSION: This study provides an extensive assessment of analytical methods best suited for processing data from different microarray technologies and can assist integration of technologically different gene expression datasets in biological systems.

Pathophysiological, genetic and gene expression features of a novel rodent model of the cardio-metabolic syndrome.

BACKGROUND: Complex etiology and pathogenesis of pathophysiological components of the cardio-metabolic syndrome have been demonstrated in humans and animal models. METHODOLOGY/PRINCIPAL FINDINGS: We have generated extensive physiological, genetic and genome-wide gene expression profiles in a congenic strain of the spontaneously diabetic Goto-Kakizaki (GK) rat containing a large region (110 cM, 170 Mb) of rat chromosome 1 (RNO1), which covers diabetes and obesity quantitative trait loci (QTL), introgressed onto the genetic background of the normoglycaemic Brown Norway (BN) strain. This novel disease model, which by the length of the congenic region closely mirrors the situation of a chromosome substitution strain, exhibits a wide range of abnormalities directly relevant to components of the cardio-metabolic syndrome and diabetes complications, including hyperglycaemia, hyperinsulinaemia, enhanced insulin secretion both in vivo and in vitro, insulin resistance, hypertriglyceridemia and altered pancreatic and renal histological structures. Gene transcription data in kidney, liver, skeletal muscle and white adipose tissue indicate that a disproportionately high number (43-83%) of genes differentially expressed between congenic and BN rats map to the GK genomic interval targeted in the congenic strain, which represents less than 5% of the total length of the rat genome. Genotype analysis of single nucleotide polymorphisms (SNPs) in strains genetically related to the GK highlights clusters of conserved and strain-specific variants in RNO1 that can assist the identification of naturally occurring variants isolated in diabetic and hypertensive strains when different phenotype selection procedures were applied. CONCLUSIONS: Our results emphasize the importance of rat congenic models for defining the impact of genetic variants in well-characterised QTL regions on in vivo pathophysiological features and cis-/trans- regulation of gene expression. The congenic strain reported here provides a novel and sustainable model for investigating the pathogenesis and genetic basis of risks factors for the cardio-metabolic syndrome.

Direct quantitative trait locus mapping of mammalian metabolic phenotypes in diabetic and normoglycemic rat models.

Characterizing the relationships between genomic and phenotypic variation is essential to understanding disease etiology. Information-dense data sets derived from pathophysiological, proteomic and transcriptomic profiling have been applied to map quantitative trait loci (QTLs). Metabolic traits, already used in QTL studies in plants, are essential phenotypes in mammalian genetics to define disease biomarkers. Using a complex mammalian system, here we show chromosomal mapping of untargeted plasma metabolic fingerprints derived from NMR spectroscopic analysis in a cross between diabetic and control rats. We propose candidate metabolites for the most significant QTLs. Metabolite profiling in congenic strains provided evidence of QTL replication. Linkage to a gut microbial metabolite (benzoate) can be explained by deletion of a uridine diphosphate glucuronosyltransferase. Mapping metabotypic QTLs provides a practical approach to understanding genome-phenotype relationships in mammals and may uncover deeper biological complexity, as extended genome (microbiome) perturbations that affect disease processes through transgenomic effects may influence QTL detection.

Mapping diabetes QTL in an intercross derived from a congenic strain of the Brown Norway and Goto-Kakizaki rats.

Genetic studies in experimental crosses derived from the inbred Goto-Kakizaki (GK) rat model of spontaneous diabetes mellitus have identified quantitative trait loci (QTL) for diabetes phenotypes in a large region of rat Chromosome (RNO) 1. To test the impact of GK variants on QTL statistical and biological features, we combined genetic and physiologic studies in a cohort of F(2) hybrids derived from a QTL substitution congenic strain (QTLSCS) carrying a 110-cM GK haplotype of RNO1 introgressed onto the genetic background of the Brown Norway (BN) strain. Glucose intolerance and altered insulin secretion in QTLSCS rats when compared with BN controls were consistent with original QTL features in a GK x BN F(2) cross. Segregating GK alleles in the QTLSCS F(2) cross account for most of these phenotypic differences between QTLSCS and BN rats. However, significant QTL for diabetes traits in both the QTLSCS and GK x BN F(2) cohorts account for a similar small proportion of their variance. Comparing results from these experimental systems provides indirect estimates of the contribution of genetic interactions and environmental factors to QTL architecture as well as locus and biological targets for future post-QTL mapping studies in congenic substrains.

Genomic organization of the rat Clock gene and sequence analysis in inbred rat strains.

While mutations in genes that function in the core molecular clock may disrupt circadian periodicity, their relevance to diurnal variation in metabolic, cardiovascular, and respiratory function is unknown. The circadian Clock gene product is an essential regulator of central and peripheral circadian rhythms in mammals. We have elucidated the complete exon-intron organization of the Clock gene in rat and have carried out an extensive search for single nucleotide polymorphisms (SNPs) in a panel of 12 inbred rat strains that exhibit diversity in studies of central and peripheral organ function and disease. The rat Clock gene consists of 23 exons spanning approximately 75 kb. Comparative sequence analysis identified 33 novel SNPs, including 32 that distinguish the Brown Norway (BN) rat from the other strains studied. Most notable were two novel mutations in the BN sequence at exon 8, Ile131Val and Ile132Val, occurring in a segment of the highly conserved PAS-A domain of the protein. These results afford the opportunity to assess the impact of genetic variation in Clock on central and peripheral functions subject to the core molecular clock and to test the importance of Clock variants in explaining diversity among rat strains in the expression of phenotypes, such as blood pressure, subject to circadian oscillation.

Chromosomal mapping of quantitative trait loci controlling elastin content in rat aorta.

Extracellular matrix molecules such as elastin and collagens provide mechanical support to the vessel wall. In addition to its structural role, elastin is a regulator that maintains homeostasis through biologic signaling. Genetically determined minor modifications in elastin and collagen in the aorta could influence the onset and evolution of arterial pathology, such as hypertension and its complications. We previously demonstrated that the inbred Brown Norway (BN) rat shows an aortic elastin deficit in both abdominal and thoracic segments, partly because of a decrease in tropoelastin synthesis when compared with the LOU rat, that elastin gene polymorphisms in these strains do not significantly account for. After a genome-wide search for quantitative trait loci (QTL) influencing the aortic elastin, collagen, and cell protein contents in an F2 population derived from BN and LOU rats, we identified on chromosomes 2 and 14, 3 QTL specifically controlling elastin levels, and a further highly significant QTL on chromosome 17 linked to the level of cell proteins. We also mapped 3 highly significant QTL linked to body weight (on chromosomes 1 and 3) and heart weight (on chromosome 1) in the cross. This study demonstrates the polygenic control of the content of key components of the arterial wall. Such information represents a first step in understanding possible mechanisms involved in dysregulation of these parameters in arterial pathology.

Quantitative trait locus dissection in congenic strains of the Goto-Kakizaki rat identifies a region conserved with diabetes loci in human chromosome 1q.

Genetic studies in human populations and rodent models have identified regions of human chromosome 1q21-25 and rat chromosome 2 showing evidence of significant and replicated linkage to diabetes-related phenotypes. To investigate the relationship between the human and rat diabetes loci, we fine mapped the rat locus Nidd/gk2 linked to hyperinsulinemia in an F2 cross derived from the diabetic (type 2) Goto-Kakizaki (GK) rat and the Brown Norway (BN) control rat, and carried out its genetic and pathophysiological characterization in BN.GK congenic strains. Evidence of glucose intolerance and enhanced insulin secretion in a congenic strain allowed us to localize the underlying diabetes gene(s) in a rat chromosomal interval of approximately 3-6 cM conserved with an 11-Mb region of human 1q21-23. Positional diabetes candidate genes were tested for transcriptional changes between congenics and controls and sequence variations in a panel of inbred rat strains. Congenic strains of the GK rats represent powerful novel models for accurately defining the pathophysiological impact of diabetes gene(s) at the locus Nidd/gk2 and improving functional annotations of diabetes candidates in human 1q21-23.

Integration of the rat recombination and EST maps in the rat genomic sequence and comparative mapping analysis with the mouse genome.

Inbred strains of the laboratory rat are widely used for identifying genetic regions involved in the control of complex quantitative phenotypes of biomedical importance. The draft genomic sequence of the rat now provides essential information for annotating rat quantitative trait locus (QTL) maps. Following the survey of unique rat microsatellite (11,585 including 1648 new markers) and EST (10,067) markers currently available, we have incorporated a selection of 7952 rat EST sequences in an improved version of the integrated linkage-radiation hybrid map of the rat containing 2058 microsatellite markers which provided over 10,000 potential anchor points between rat QTL and the genomic sequence of the rat. A total of 996 genetic positions were resolved (avg. spacing 1.77 cM) in a single large intercross and anchored in the rat genomic sequence (avg. spacing 1.62 Mb). Comparative genome maps between rat and mouse were constructed by successful computational alignment of 6108 mapped rat ESTs in the mouse genome. The integration of rat linkage maps in the draft genomic sequence of the rat and that of other species represents an essential step for translating rat QTL intervals into human chromosomal targets.