European rational approach for the genetics of diabetic complications (EURAGEDIC)

The genotyping of 538 haplotype-tagging SNPs or non-synonymous polymorphisms selected in 155 genes is completed in the DNA collection for case/control association studies of diabetic nephropathy including 2944 patients with type 1 diabetes from three different European populations, 91 of these polymorphisms have been further genotyped in an additional 1472 DNAs from 598 trios (patients with or without nephropathy and both parents) for intra-familial association. In addition 78 SNPs are typed in 736 DNAs from non diabetic controls from each population. Lack of population stratification was shown by genotyping of 94 markers for genomic control. A total of 2010832 genotypes have been generated. Results of association analysis in case/control samples have been published in Diabetes (Diabetes. 2008 Jul 15. [Epub ahead of print]; Diabetes. 2006 Nov;55(11):3166-74.), in Nephrol Dial Transplant. 2008 Jan;23(1):161-8. Epub 2007 Aug 17. and in J Am Soc Nephrol. 2007 Apr;18(4):1284-91. Epub 2007 Mar 21. Genotype data for the 150 genes as well as clinical data are available at http://genecanvas.idf.inserm.fr/news.php.

THESIAS is a popular software for performing haplotype association analysis in unrelated individuals. THESIAS has recently been upgraded (Tregouet & Garelle Bioinformatics 2007 23; 1038-9) to incorporate categorical phenotype analysis. This extension has been proposed to apply haplotype association analysis to phenotype characterizing the renal disease severity (microalbuminuria, macroalbuminuria, renal end stage failure). The new JAVA interface implementation of THESIAS: testing haplotype effects in association studies is available at http://genecanvas.idf.inserm.fr/news.php).

We derived an inventory of positional and functional candidate genes for diabetic nephropathy using gene transcription profiling carried out with rat models of diabetes mellitus, including WKY rats made hyperglycaemic by injection of streptozotocin (STZ) and diabetic rats of the Goto Kakizaki (GK) and BN.GK congenic strains. The ultimate aim was to provide human geneticists with an inventory of genes differentially expressed that could be directly applied to association studies in EURAGEDIC cohorts of diabetic patients characterised for evidence of nephropathy. Renal gene expression profiling from WKY, STZ treated and GK rats was carried out with Affymetrix arrays allowing the interrogation of the expression level of over 14,000 rat genes. Data interpretation was based on results from three pairwise comparisons designed to investigate the general effects of hyperglycaemia on renal gene expression (GKvsWKY and STZvsWKY), as well as the effects of diabetes severity (GKvsSTZ) and diabetes susceptibility genetic factors (GKvsWKY) on gene transcription. Overall results show that increased severity of hyperglycaemia is associated with an increased number of gene differentially expressed. There were no significant differences in the number of genes down- or up-regulated (Table 1). All gene lists and statistical data (significance and fold changes of transcription changes) have been made available to consortium partners for association studies in human. Table 1. Overview of differential gene transcription regulation in kidney of diabetic and control rats. Total number of probe sets showing significant changes (P<0.05) in transcript levels in each of the three comparisons and available annotation in Unigene for the corresponding genes. Comparisons Total Up-regulated Down-regulated Transcripts in Unigene GK vs WKY 252 136 116 219 STZ-WKY vs WKY 682 276 406 597 STZ-WKY vs GK 834 305 529 737 A specific focus of our research was to participate in the functional annotation of human susceptibility loci associated with diabetes nephropathy. Using comparative genomics and existing annotations of the rat genome sequence, we investigated in details rat chromosomal regions homologous to four human loci (3q23-29; 7q32.3-q33, 18q12-23, 20p12.3-p13) which show the most convincing evidence of linkage with diabetic nephropathy. We were able to show that several positional candidate genes at these loci previously investigated in association studies in human diabetic nephropathy, including for example gene that encode carnosinases 1 and 2 (Cndp1; Cndp2) and solute carrier family 2 member 2 (Glut2; Slc2a2), are differentially expressed between diabetic rats and WKY controls. More interestingly, we show that the transcription of other genes, including in particular predicted or functionally unannotated transcripts, are actually affected by hyperglycaemia and can contribute to renal histopathology as seen in GK and STZ-treated rats. The full list of rat genes differentially expressed between diabetic and control rats was made available to consortium partners for further genetic studies in human. Overall our results participate in the definition of new functional candidate genes for diabetic nephropathy genetics. A manuscript reporting our results has been submitted to BMC Genomics.

The inventory of common genetic variation in more than 150 candidate genes for cardio-vascular complications by sequencing of 32-96 individuals led to the identification of more than 2000 polymorphisms of which 60% are newly identified SNPs; allelic and haplotype frequencies have been estimated from the data on pooled DNAs; those data are available in public databases (NCBI at http://www.ncbi.nlm.nih.gov/) and at the INSTITUTION URL (http://genecanvas.idf.inserm.fr/news.php). The data have been published in Diabetes (Diabetes. 2008 Jul 15. [Epub ahead of print]; Diabetes. 2006 Nov;55(11):3166-74.) and are available to the scientific community. Those data are valuable for the study of these candidates genes with various phenotypes.

Genetic-association studies are unraveling the genetic basis of complex human diseases. The population-based case-control study, a commonly used approach for association studies, is subject to the problem of population admixture. Consequently, evidence of disease-marker associations obtained from such studies is ideally confirmed by alternative methods. The Transmission/Disequilibrium Test (TDT) is suitable to assess evidence of association obtained from case-control studies. Since data are increasingly available from both case-control and TDT studies of the same disease-marker association, it can be useful to obtain a combined estimate of disease-marker association. The odds ratio is a commonly used measure of the magnitude of a disease-marker association that can be easily obtained in case-control studies. We have developed a statistical method to show how an odds ratio estimate and its' associated standard error can be obtained from TDT results. Furthermore, we have developed a method for integrating results from case-control studies and the TDT to provide a combined estimate of disease-marker association. Such combined estimates can be used to contrast the results of the two studies and provides an overall picture of the effect size attributable to such polymorphism. The statistical method and an application to real-world data is described in a peer-reviewed paper: Authors: Kazeem GR, Farrall M. Title: Integrating case-control and TDT studies. Journal: Annals of Human Genetics. Publication year: 2005 Volume: 69(Pt 3) Pages:329-35.

We studied ~150 candidate genes for diabetic nephropathy. Polymorphisms were identified through sequencing of promoter, exon and flanking intron gene regions and database search. More than 400 haplotype-tagging SNPs and non-synonymous variants were tested for association with diabetic nephropathy (persistent albuminuria >300 mg/24h) in a large type 1 diabetes case/control (1176/1323) study from 3 European populations. Only one SNP, rs2281999 located in UNC13B gene, was significantly associated with DN after correction for multiple testing. Analyses of 21 additional markers fully characterizing the haplotypic variability of the UNC13B gene showed consistent association with SNP rs13293564 (G/T) located in intron 1 of the gene with DN in the 3 populations. The odds ratio (OR) for DN associated with the TT genotype was 1.68 [95%CI, 1.29 - 2.19], p = 1.0 10-4). This association was replicated in an independent population of 412 cases and 614 controls, (combined OR of 1.63 [95%CI, 1.30 - 2.05], p = 2.3 10-5). We identified a polymorphism in UNC13B gene associated with DN. UNC13B mediates apopotosis in glomerular cells in the presence of hyperglycemia, an event occurring early in the development of DN. We propose that this polymorphism could be a marker for the initiation of DN. However, further studies are needed to clarify the role of UNC13B in DN. These results have been published in Diabetes (Diabetes. 2008 Jul 15. [Epub ahead of print]; Diabetes. 2006 Nov;55(11):3166-74.).

Phenotypic Characterisation: We have generated phenotype data on nine ENU induced lines (GENA/ 263, GENA/348, GENA/389, IGT/1, IGT/3, IGT/4, IGT6, GENA394 and IGT/10) and in addition the Lepr db/db mice for diabetic complications in the kidney. Phenotypic characterisation included histological analysis of kidney sections measuring extracellular matrix area in the mesangium of glomeruli, plasma glucose and insulin levels and biochemical analysis of urine for glucose, albumin and creatinine. Morphological measurements of kidney and bodyweight have been recorded. Of the ENU induced lines characterised only GENA/348 (Gck mutant) and IGT/10 exhibited potentially interesting kidney phenotypes. GENA/348 animals showed impaired glucose tolerance which has been shown to be caused by a point mutation in the MODY2 gene, glucokinase. Histological analysis of kidney sections from GENA/348 progeny showed a 20% increase in glomerular mesangial matrix in comparision to age-matched control animals. IGT/10 animals exhibited hyperglycaemia, glycosuria and proteinuria. Preliminary histological analysis of IGT/10 kidney sections showed a 10% increase in glomerular mesangial matrix in comparision to age-matched control animals. The GENA 348 model is available on request. The Lepr db/db mouse was further characterized using metabolomic analysis of urine samples (Salek et al. "A metabolomic comparison of urinary changes in type 2 diabetes in mouse, rat, and man" (2007) Physiological Genomics 29: 99-108). This study demonstrated metabolic similarities between all 3 species with responses associated with general metabolic stress and changes in the TCA cycle. Expression Profiling: We selected an established severe diabetes mutant to be used for expression profiling experiments. Mice carrying a mutation in the leptin receptor (Lepr db/db) develop diabetes at 4-6 wks of age and become obese. Phenotype data was collected at 4 week intervals and included plasma glucose and insulin measurements, histological analysis of extracellular matrix expansion in the glomerulii of kidneys and glucose and protein levels in the urine. In addition, we collected 24hr urine samples in metabolic cages for both db/db and control animals for more detailed urine analysis (including an estimation of glomerular filtration rate) to correspond with microarray datapoints. An initial expression profiling experiment was performed using the mouse Compugene 7.5k "known" gene oligo arrays which were hybridised with RNA from the kidneys of 8wk, 12wk and 16wk old db/db and control animals. After analysis of this micro array data we were able to generate a list of 50 genes that were differentially expressed in at least two time-points. The identity of each gene is known from the sequence of each oligonucleotide on the array (in the original plan we would have used spotted cDNA arrays however we were able to gain access to larger and more specific oligonucleotide gene probe arrays) and the map position and sequence of all of these genes is known through analysis of the mouse genomic reference sequence and EST databases. This list was further refined by comparative analysis of genes differentially expressed in the GK rat expression profiling experiment (WP5) and published results in the mouse to produce a TOP50 list of candidate genes which would then have 10 to 15 genes chosen from it which would be screened in the 47 EURAGEDIC pools. During the final year an opportunity arose to use the ~32k Operon oligo array set. Instead of using a second line we decided to increase the time-span of our longitudinal study using the db/db mice so that a more comprehensive analysis of gene expression changes during the critical phase of development of diabetic kidney disease could be achieved (this was more than equivalent to carrying out an experiment on a second line). For this second expression profiling experiment we extended the time-course to include 20 and 24wk animals; the time-points analysed are 4wk, 8wk, 12wk, 16wk, 20wk and 24wks of age. Preliminary analysis of this microarray data has generated lists of differentially expressed genes that are statistically significantly at each time-point. Further analysis of longitudinal expression patterns is underway using this high quality dataset. Data has been deposited in ArrayStager and is being further characterised by the EUREGENE consortium. A manuscript is being planned and prepared.

Rezultaty

Udostępnij tę stronę

Pobierz