Van Eyk, Dunn - Proteomic and Genomic Analysis of Cardiovascular Disease - 2003 (522919), страница 5
Текст из файла (страница 5)
1.2are chosen from different regions of the gene.Each perfect match (PM) oligonucleotide ispaired with an oligonucleotide that has a mismatch (MM) base in the central position,making up a probe pair. Oligonucleotides aresynthesized in-situ to a glass wafer by a lightdirected chemical synthesis process calledphotolithography. Multiple probe arrays aresynthesized simultaneously on a large glasswafer, which is then diced into individualprobe arrays. Prices for cDNA microarraysand Affymetrix GeneChipsTM are estimates asof 3/2002.781 Microarray Expression Profiling in Cardiovascular DiseaseIn contrast to high-density oligonucleotide arrays, cDNA are “home-made” byseveral academic laboratories. Detailed information about the manufacture ofcDNA microarrays is available from Pat Brown’s laboratory in Stanford (Tab.
1.1).The advantage of cDNA microarrays is the great degree of flexibility in the choiceof arrayed elements, allowing for the preparation of customized and tissue-specific arrays for specific investigations. In addition, spotting clones from unsequencedlibraries makes a great tool for new gene discovery. These advantages, togetherwith the comparatively low costs, have made cDNA microarrays the most frequently used platform in academic institutions (Fig. 1.2). However, cDNA microarrays require the time-consuming handling and amplification of cDNA libraries orclone collections (I.M.A.G.E., Riken), which carry a high risk of clone contamination.
Using oligonucleotides (50mers and up) instead of cDNAs might provide asolution to this problem, but on the expense of cost efficiency and detection sensitivity [10]. Spotted or printed cDNA microarrays and related technology are available from Agilent, Ambion, Clontech, Packard Bioscience/Perkin Elmer, andothers, as well as through Core Facilities that have been established at diverse academic institutions (Tab. 1.1).High density oligonucleotide microarrays consist of short 20–25mer oligonucleotides that are either printed onto glass slides or synthesized in situ by ink-jet technology (Agilent Technologies) or by photolithography onto silicon wafers (Affymetrix). Affymetrix high-density oligonucleotide GeneChipsTM contain up to1,000,000 unique oligonucleotide features covering more than 39,000 transcriptvariants (represented on two microarrays HG-U133A and HG-U133B).
Each geneis represented by at least one set of 11–20 different “probe pairs” (Fig. 1.2). Aprobe pair consists of a 25-base-pair (bp) “perfect-match” (PM) oligonucleotideprobe and a 25-bp “mismatch” probe (MM), in which the 13th position is designednot to match the target (cellular) sequence. The information across all 20 pairedPM and MM probes (the “probe set”) is integrated by proprietary Affymetrix Microarray Suite (MAS) software.For hybridization of eukaryotic samples to Affymetrix GeneChipsTM, doublestranded cDNA is synthesized from 5 lg total RNA (or a minimum of 0.2 lg purified poly(A)+ messenger RNA) isolated from tissue or cells.
The first strand synthesis is primed using a T7-(dT)24 primer. After second strand synthesis, an in vitro transcription (IVT) reaction is done to produce biotin-labeled cRNA from thecDNA template. The cRNA is fragmented before hybridization. In contrast tocDNA microarrays, each mRNA preparation is hybridized to a separate oligonucleotide array. Hybridization, washing, and staining of Affymetrix GeneChipsTMtakes place in a highly standardized setup, which greatly reduces the variability ofmicroarray processing.
After scanning of the arrays, various analytical methodsare applied to distinguish specific signals from noise, calculate background, andto scale and normalize the signals (see 2. Computational analysis of microarraydata) before expression levels can be compared across arrays.If the amount of starting material is not sufficient to generate the minimumamount of RNA needed for the standard protocols, for example if human biopsy1.1 DNA Microarray Technologiessamples or microdissected single cells are to be analyzed, alternative labeling protocols have to be used.
Starting amounts of less than 100 ng of total RNA require tworounds of amplification, which generally results in a loss of signals from low abundance RNAs. Optimized protocols have been developed that allow to faithfully amplify as little as 2 ng total RNA [11, 12]. However, it is also reported that the qualityof the amplified product, as measured by the specific signal intensity after hybridization, drops significantly when less than 10–20 ng of starting material is used [12].Advantages and disadvantages of both microarray technology platforms aresummarized in Fig. 1.2. However, the crucial difference between cDNA and oligonucleotide microarrays is that cDNA microarrays return the amount of each transcript relative to another sample, whereas oligonucleotide microarrays theoreticallyreturn an absolute amount of each transcript.
This implies a major difference inthe ability to group and universally compare microarrays, which is discussed below under analysis techniques. Regardless of the technology platform chosen, microarray experiments yield far more information than we are used to process.Each experiment typically results in hundreds of genes that are differentially expressed across the time points, conditions, or phenotypes that are being compared.
It is extremely important to design meaningful experiments in order toavoid “drowning” in a glut of uninterpretable data.1.1.2Designing Meaningful ExperimentsThe number of review articles on gene expression technologies probably exceedsthe number of primary research publications in this field [13]. This is not the result of any paucity of primary data since Microarray Core Facilities have been established in nearly every academic institution, each processing hundreds of microarrays each year, and numerous laboratories manufacture and run their owncDNA microarrays, resulting in millions of independent gene expression measurements.
One might argue that the limited availability of efficient publicly availabletools for data processing, functional annotation, and literature-searching make ithard to filter through the vast amount of gene expression data, find meaningfulresults, and integrate it with existing knowledge. The more likely reason is thatmost scientists have been trained to look for an individual gene or signaling pathway, the expression change of which is responsible for a biological phenomenonor disease state. Facing lists of hundreds of gene expression changes leaves a formidable challenge in identifying changes that are important in establishing thephenotype from changes that are secondary phenomena.
Browsing through thelists of genes in order to decide which gene represents a key regulatory moleculeor a potential new therapeutic target, we will most likely chose candidate genesthat are known to us and conform with pre-existing hypotheses rather than takingthe risk to potentially waste time on unknown genes. However, genome-wide expression profiling should lead to the generation of new hypotheses rather thanconfirm existing pathways. How can we enrich the gene lists for positive candidates and get more meaningful results?9101 Microarray Expression Profiling in Cardiovascular DiseaseFirstly, performing independent replicates is essential to identify the biologicaland experimental variability in an experiment and will substantially decrease thenumber of false positive gene expression changes.
In our experience, analyzing independent triplicate samples has been sufficient in most cases. However, highvariability or limited availability of starting material, as for human biopsy tissues,might make replications impossible.Secondly, although this might seem obvious, designing meaningful experimentsis critical. In general, cell lines have the advantage of cell type homogeneity andwill give relatively little biological variability.
When looking for causes of phenotypic changes, for example the induction of differentiation, it is important to lookfor early events. Comparing undifferentiated to differentiated cells may not giveyou clues on the initial trigger, e.g. the underlying key regulatory molecules orsignaling cascades, but reflect the consequences of the induction of differentiation. Similarly, comparing a transgenic mouse line with its transgene-negative littermate will give one long gene list but it is very difficult, if not impossible, to distinguish between primary and secondary changes in gene expression without additional effort placed into experimental design.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
