Van Eyk, Dunn - Proteomic and Genomic Analysis of Cardiovascular Disease - 2003 (522919), страница 27
Текст из файла (страница 27)
Depending on the questionor the hypothesis, two broad approaches can be used, the horizontal approach andthe vertical approach (Fig. 5.2). In the horizontal approach, the tools of functionalgenomics are used to determine how a change in the function of the organ results in a change of gene expression. This approach is horizontal, because it willanalyze a large number of genes at one level of investigation (usually the mRNAexpression) and lead to the determination of a genomic profile.
In the verticalapproach, the question asked to functional genomics is what is the function of aspecific gene (Fig. 5.2). In that approach, the investigator concentrates on onegene and conducts his experiments by the combination of various techniquesThe horizontal and vertical approaches of functional genomics. The horizontal approach shows how gene expressionFig.
5.2is affected by function. The vertical approachshows the function of a specific gene product.83845 Genomics by Subtractive Hybridization(mRNA expression, protein-protein interaction, cellular localization, transfection,genetically-modified animals . . .). This last approach is used for the characterization of unknown and novel genes, which still represent a substantial part of ourgenome [30, 31]. Both approaches are complementary. As we will see below, thesubtractive hybridization methodology is used to investigate a large array of genesin a specific condition (horizontal approach), but also brings some informationabout unknown genes, which in turn determines the vertical approach.
Whereasthe horizontal approach leads to the discovery of new concepts, the verticalapproach leads to the discovery of new functions. Combining both approachesprovides a link between description and mechanism.5.1.2The ModelA good model of investigation is equally important. Most of the experiments offunctional genomics are performed on samples from in vivo studies. This adds tothe complexity of the experiments because, in addition to the potential problem ofreproducibility and sample-to-sample variation, we must also consider the relevance of the model used and the possibility of inter-species differences. For obvious epidemiological reasons, most of us are interested in determining how thehuman heart develops and ages, and how it can be injured and repaired.
Wetherefore need to develop and use the animal models which are relevant to clinical problems in the human heart. So far, these questions are mainly addressed inrodent models, which points out to the issue of species differences. A classical example of genomic differences between species is the “fetal gene program”. Thepioneering experiments of gene expression in the heart pinpointed that a rodentheart submitted to increased workload switches the expression of contractile proteins from an “adult” isoform back to an isoform preferentially expressed in thefetal heart.
For instance, the cardiac-specific isoform of alpha-actin is partially replaced by the “fetal” skeletal isoform, whereas the alpha-isoform of myosin heavychain is partially replaced by the beta-isoform [32–34]. This protein switching results from a switch in the expression of the corresponding genes. It has beenshown since that this concept can be extended to other categories of genes (suchas genes involved in the regulation of calcium handling and metabolism) [11, 12,35, 36], and is also reproduced when the workload is decreased rather than increased [13]. However, such “fetal gene program” is hardly reproduced in largermammals [37]. For instance, the predominant isoform of myosin heavy chain inthe adult is already the beta-isoform [38] and other mechanisms of adaptation ofcontractile performance, such as the phosphorylation or dephosphorylation ofmyosin light chains [39, 40], is more developed in large mammals.
Other adaptivemechanisms are activated in larger mammals that do not necessarily exist in rodents. An obvious reason for the discrepancy between species is that a mouseheart beating at 500 beats per minute has different regulatory mechanisms than ahuman heart beating at 75 beats per minute, although the overall goal of cardiacfunction remains the same. Heart rate, cardiac chamber geometry, loading condi-5.2 Analyzing Gene Expression by Subtractive Hybridizationtions, mechanisms of calcium handling, coronary anatomy and collaterals are justa few examples of biological and anatomical parameters responsible for speciesdifferences.
It is therefore crucial to determine the relevance of the model for theproblem investigated. Rodents, especially mice, are particularly attractive as “testtubes” to study the effects of one specific gene, either overexpressed or knockedout, on global gene expression. As shown below, larger mammals can be usefulfor more complex models of ischemic heart disease or heart failure.5.1.3The Technological ApproachOne of the first techniques used for gene profiling was the differential display[41–43]. Although elegant and simple, this technique is usually extremely timeconsuming because it addresses one gene at a time. Currently, the technology ofmicroarrays is extremely popular.
One chip contains most of the expressed transcripts from one species. The sample preparation is relatively easy, provided thatsome RNA of good quality is extracted from the tissue or the cells. The procedurecan be started with as low as 5 lg total RNA. Processing the chips is a quick andautomated process. The downside is that the signal to noise ratio can be weak insome hybridizations, which necessitates to repeat the experiments or to increasethe number of samples in the series. Each chip represents a compromise betweensensitivity and specificity.
The choice between cDNA and oligonucleotide chips isan illustration of this compromise. Also, the analysis of the data is not necessarilyeasy, as thousands of genes have to be analyzed and clustered [44]. It results thatthe statistical tests needed to determine significant differences are far more complex than a simple Student’s t test [45]. Biological variations from sample to sample must be taken into account and several recent studies stressed the importanceof repeating the experiments in a representative series of samples [46, 47].
Evenmore importantly, the investigator must often rely on the use of commercialchips, which limits the possibility to control their quality or content. Cross-contaminations between different clones can not be excluded, and the chips aresometimes printed erroneously [48]. Therefore, even if the technique is powerfuland attractive, its use requires a validation of the results by alternative moleculartechniques.
In addition, these chips are so far available for a restricted number ofspecies only. For these reasons, the subtractive hybridization can be an attractivealternative in specific cases.5.2Analyzing Gene Expression by Subtractive HybridizationThe subtractive suppression hybridization represents a large-scale, unbiased method for detecting transcriptionally and post-transcriptionally regulated genes, bothknown and unknown, independently of the prevalence of these transcripts [49].
Afruitful utilization of this technique requires some RNA of excellent quality, a se-85865 Genomics by Subtractive Hybridizationquencing facility and alternative methods of validation. It can be coupled to theprinting and hybridization of microarrays [50].5.2.1MethodologyThe procedure is summarized in Fig. 5.3. The subtraction is performed betweentwo biological samples from which total RNA and, subsequently, poly-A messenger RNA are extracted.
The subtraction is followed by the subcloning of the different subtracted cDNAs. After purification, the clones can be used for sequencingor microarray printing. Querying the sequences in different databases identifiesthe subtracted genes. The microarrays offer a semi-quantitation of the variation ingene expression between the samples. The results need to be validated by alternative methods, such as Northern blot, quantitative PCR and in-situ hybridization.1. Experimental samples. Experiments are usually started from frozen tissue. Acomparison can be made between an experimental heart and a sham, or, forlarger species, between two areas of the same hearts. The last choice allows apaired comparison of the data.
About 200–400 mg of tissue is usually enoughto collect the amount of RNA needed for the experiment. Compared to other tissues, the mRNA content per gram of cardiac tissue is relatively low. It is recom-Overview of the different steps of thesubtractive hybridization. See text for details.Fig.
5.35.2 Analyzing Gene Expression by Subtractive Hybridizationmended to pool several samples from each group, to ensure that the subtraction is well representative of the condition tested.2. RNA isolation. This step is probably the most important of the whole procedure, as all the information gathered from the subtraction tightly depends onthe quality of the starting material, which is the poly-A RNA. Total RNA extraction is usually performed by the phenol/chloroform extraction [51] and does notrequire further purification. RNAse-free conditions are mandatory to preservethe structure and the full length of the RNA as well as possible. Purification ofthe poly-A mRNA requires more attention. Different techniques are available,some insisting more on the exclusion of the contaminating ribosomal RNA,some insisting more on the recovery of rare transcripts. The latter choice isprobably the best in this case, to ensure that the library to be subtracted contains as many different transcripts as possible.
A contamination by ribosomalRNA will usually not jeopardize the experiment. The quality of the RNA mustbe verified by the 260/280 absorbance ratio, by Northern blot or by automatedmethods.3. Subtractive hybridization. The subtraction between two libraries begins by thereverse transcription of the poly-A mRNA into double-stranded DNA (Fig. 5.4).The reverse transcription is usually performed with an oligo-dT primer, ratherthan random hexamers, to increase the length of the products.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
