Van Eyk, Dunn - Proteomic and Genomic Analysis of Cardiovascular Disease - 2003 (522919), страница 33
Текст из файла (страница 33)
A large number (~60) of gene changes were observed. However, a limitation of this type of study is that it is difficult to distinguish which changes in gene expression are causally important in cardioprotection. A number of inhibitors of delayed preconditioning have been administeredto animals. Therefore a promising strategy would be to determine which genesare not activated by preconditioning when inhibitors of protection are administered. This approach will essentially subtract out the effect of ischemia, and potentially may result in a small number of more relevant gene changes.6.5.2Transgenic Models of ProtectionA similar approach to that of adding agents that mimic or inhibit preconditioning, is to compare levels of gene expression between transgenic models that exhibit cardioprotection and control non-protected wild-type littermates.
Programs suchas, the NHLBI-Program for Genomic Applications which has as its goal “linkinggenes to structure, function, dysfunction and structural abnormalities of the cardiovascular system caused by clinically relevant genetic and environmental stimuli”,and the NIEHS National Center for Toxicogenomics, that has the goal of identifying mechanisms that underlie a variety of environmentally associated diseases,should help facilitate this type of study.
An increasing number of transgenicmouse models exhibit cardioprotection. By comparing differences in gene expression between these transgenic mice and their WT control littermates one can po-6.5 Approaches to Identify Genes Involved in Cardioprotectiontentially identify gene changes involved in the protection.
In addition, there are anumber of genetic mouse models that show enhanced susceptibility to injury andan analysis of differences in gene expression between these animals and their WTlittermates could also provide insight into gene changes involved in cardioprotection. One consideration in studies of transgenic animals is that although typicallya single gene has been changed, there can be other compensatory changes ingene expression. One of the challenges is to determine which gene changes (initial or compensatory) are causally involved in protection.
An additional strategythat may be useful to uncover cardioprotective gene changes in transgenic animals is to examine gene changes that occur when the animals are stressed or injured. It is possible that the gene modified in the transgenic animal leads to protection by altering the genes that are induced or repressed during the response toinjury. These gene changes would only be uncovered by stressing the animals(e.g. via pressure overload or an ischemic insult).6.5.3Cardioprotection in FemalesThere are several recent studies in animal models, as well as data in humans,suggesting that, particularly under conditions of increased stress, females showdecreased susceptibility to ischemic injury and heart failure [36–39, 49].
Onecould therefore examine differences in gene expression between males and females. This approach has the advantage that it can be applied to humans as wellas animal models. However, the disadvantage of this approach is that even in theheart there are likely to be a large number of male/female differences in gene expression that are unrelated to cardioprotection.6.5.4Human StudiesAs the goal of these studies is to identify genes which are cardioprotective in humans, it would be useful to perform studies on human myocardium. One potential approach is to determine what genes are altered in diseases, such as ischemiaor heart failure. The idea is that by identifying the gene changes that cause injury,we can better understand how to prevent the injury.In summary, microarray technology could potentially elucidate genes or set ofgenes that could be manipulated to enhance cardioprotection.
The challenge withthis type of study is to distinguish which changes in gene expression are causallyinvolved in protection. By comparing changes in gene expression across multiplemodels of cardioprotection it is hoped that epigenetic changes can be eliminatedand the important gene changes will emerge. This assumes that the mechanismof protection in the different models is due to similar changes in gene expression,and this may not be the case.
If there is not a common pathway for cardioprotection then comparison between these models will not show common genechanges.1051066 DNA Microarray Gene Profiling6.6Practical Considerations in Microarray TechnologyThere are a number of practical issues that are important in microarray studies.Many of these will be covered in other chapters. This section will highlight a fewimportant issues, including, sample selection, the necessity for replicates, crosshybridization, validation, the relationship of RNA to protein, and bioinformatics.Details of the cDNA microarray method are described elsewhere [50].6.6.1Sample SelectionTo identify cardioprotective genes by comparing differences in gene expression between cardioprotected and non-protected models, it is important to make comparisons between animals with the same genetic background. Comparisons in thesame strain or among littermates are desirable.
Also, as females appear to havesome inherent cardioprotection, it is important to compare males to males and females to females, unless one is specifically studying gender differences in cardioprotection. As in others studies, it is also important to compare animals of similarage and with similar treatments. It is also important to sample from the same location, for example left ventricle versus right ventricle.For studies with human samples there are a number of additional challenges.There have been a number of recent studies utilizing explanted human heartsfrom patients in heart failure undergoing transplantation.
The availability of control tissue is an important issue. Control heart tissue is typically obtained fromdonor hearts that could not be used for technical reasons, such as age of donor,death of the recipient and other logistical problems. The control donor hearts arefrequently on inotropic support such as dopamine or norepinephrine prior to tissue harvest. Thus, the “control” hearts are more appropriately termed non-failinghearts. Rapid autopsy samples are another potential source of control tissue, butit is difficult to obtain heart autopsy samples in less than two to three hours afterdeath, and there can be substantial RNA degradation during this time.
Thus, amajor limitation of studies of human heart tissue is the lack of suitable controltissue.Explanted failing hearts are often obtained from patients with ischemic heartdisease and thus contain fibrotic areas. In sampling failing hearts, it is importantto avoid areas of fibrosis. It is also important to perform histology on both failingand non-failing hearts in order to correlate gene expression changes against aphenotypic description. Another issue with human heart samples is that bothnon-failing and failing patients are commonly treated with numerous medicationsthat could influence gene expression.
As discussed, non-failing patients are typically on inotropic support. The heart failure patients are also typically on a varietyof medications. Thus changes in gene expression due to these medications will besuperimposed on any changes in gene expression related to heart failure.6.6 Practical Considerations in Microarray Technology6.6.2Variability in HybridizationThere can be variability within an array due to differences in labeling and hybridization. For example in cDNA array experiments using a direct labeling method,one of the fluorophores (Cy5) is more bulky and sometimes does not incorporateinto newly transcribed cDNA as well as the other commonly used fluorophore(Cy3).
This can lead to some differences in labeling and hybridization. One way tominimize this problem is to do multiple hybridizations, and label the control withthe Cy5 fluorophore for half of the hybridization and with Cy3 for the other half.This “dye reversal” cancels out variation due to differential labeling with Cy3 andCy5.6.6.3Cross-hybridization of SequencesCross hybridization may occur between sequences that are similar to each otherwithin the coding sequence. By using a large portion of the 3’ untranslated sequence for each clone one can reduce potential cross hybridization.
To determinewhich other genes might be hybridizing with the sequence on the chip one can“BLAST” the sequence against GenBank databases. In addition, one can follow upwith Northern analysis to determine if the chip probe recognizes more than onetranscript.6.6.4Importance of Replicate Hybridizations and Validation of Significantly Changed GenesThere have been recent papers stressing the importance of replicates [51].
Theoutput for a gene array experiment is a list of genes that are significantly inducedor repressed. There are multiple methods to determine these genes. Some groupsuse an arbitrary cut-off. We have implemented a confidence interval described byChen et al. [52]. However, application of a confidence interval when there arethousands of genes on a chip is accompanied by a certain probability of a chanceoccurrence of a gene appearing as induced or repressed. The probability of detecting a change in gene expression by random chance can be calculated by binomialprobability [53].
For example, by random chance one would expect 240 outlierswhen performing a single hybridization using a chip with 2,000 genes; triplicatehybridizations reduces the expected number of random outliers to 0 where a genemust be significantly changed in all 3 hybridization. To be sure that changes ingene expression observed on the microarray are correct, it is important to validatechanges in representative or important genes by northern blot or quantitative realtime-PCR. In addition, it is also important to resequence and validate the originalcDNA clone for verification of identification if possible.1071086 DNA Microarray Gene Profiling6.6.5Changes in Gene Expression versus Protein or ActivityOne issue to consider with gene profiling data is how well the changes in gene expression reflect changes in protein. For example, it is possible that increased degradation of a protein would lead to increased RNA expression.
In this case interpreting an increase in gene expression as an increase in protein would lead to thewrong conclusion. Even if gene expression and protein levels change in the samedirection, there could be differences in the magnitude of the change. Furthermore, even measurements of protein levels may not tell the whole story. Recentstudies have shown that protein targeting and post-transcriptional modificationare important for understanding the activity of proteins [16, 54].6.6.6Bioinformatics or How to Tell the Forest from the TreesAs mentioned above, there are multiple methods to determine significantly changed genes within an experiment. One should implement a strategy to balance sensitivity with conservatism to reduce error.
The most robust models require replication of both chips as well as biological samples.Once the significantly altered genes from a single array or biological samplehave been determined, higher order analyses may be used to visualize patterns ortrends across multiple samples One popular form of analysis is clustering [55].Clustering of genes or samples across a set of significantly changed genes oracross genes of a similar function, may allow clear visualization of similarity ofsamples or activation/inactivation of a biological process. For example we clustered data derived from analysis of human idiopathic cardiomyopathic heartsusing gene changes that were statistically significant.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
