Array CGH Platforms

Arrays Based on Clone Inserts cDNA Arrays

The first genome-wide application of array CGH was based on cDNA arrays, with each spot representing one reversely transcribed mRNA.5 Apart from the fact that these arrays were readily available, the main advantage of cDNA arrays was that they facilitated a direct comparison of DNA copy number changes with gene expression data derived from the same tumor.i7 However, cDNA arrays are extremely gene focused and thus only a small percentage of the genome is actually represented. DNA copy number changes concerning introns or intergenic sequences are missed. Another problem arises from paralogous genes or shared sequence motifs. Together with the low signal-to-noise ratio usually obtained with this kind of arrays, these disadvantages have limited the use of cDNAs for the detection of DNA copy number changes.

Large-Insert Clone Arrays

Array CGH platforms based on large-insert clones are typically made up of bacterial artificial chromosome (BAC) and P1 artificial chromosome (PAC) clones and, to a lesser extent, also cosmids (Fig. 10.2). The first large-insert clone sets for whole genome analysis provided a resolution of about 1 Mb.8,9 Subsequently, a more comprehensive clone set has been assembled,10,11 and in 2004, the first sub-megabase-resolution whole genome tiling path array CGH study has been published, which was based on an array comprising 32,433 large-insert clones covering the whole genome in an overlapping fashion i2 (http://bacpac.chori.org/). Because of the overlap of clones, tiling path arrays can provide a theoretical resolution that is below the average insert size of a BAC of 150 kb. Figure 10.3 shows the array CGH analysis of a squamous cell lung carcinoma using a sub-megabase-resolution whole genome tiling path with more than 36,000 clones.

Despite their robustness and widespread application, BAC arrays have some limitations. The low copy number of BACs and PACs within the propagating bacteria requires special isolation methods that preserve the integrity of the clone insert while at the same time eliminate as much bacterial genomic DNA as possible. Most laboratories are not spotting this isolated DNA directly but instead amplify the material to generate a renewable stock of amplicons that can be printed several times (see below for discussion on amplification

  1. 10.1. Principle of array comparative genomic hybridization (CGH). Differentially labeled test and reference DNA (green and red spheres, respectively) are cohybridized onto an array of DNA spots printed on a glass slide. In case of a deletion in the test DNA, fewer tests DNA will bind to the corresponding spots and the red
  2. 10.1. Principle of array comparative genomic hybridization (CGH). Differentially labeled test and reference DNA (green and red spheres, respectively) are cohybridized onto an array of DNA spots printed on a glass slide. In case of a deletion in the test DNA, fewer tests DNA will bind to the corresponding spots and the red can be identified by a dominance of the green label of the test DNA. Spots representing sequences with the same copy number in the test genome relative to the reference genome appear yellow. For BAC arrays, an excess of repetitive Cot DNA (blue spheres) has to be added to suppress otherwise unspecifically binding label of the reference DNA will prevail; gains in the test genome repetitive sequences.
Chromosome Array
Fig. 10.2. Bacterial artificial chromosome (BAC) array. Each spot on a BAC array represents the very specific part of the genome that is contained in the BAC. A whole genome tiling path BAC array comprises as many BAC clones as necessary to cover the whole

genome in an overlapping manner (~32,400 for the human genome). The upper part of the image is based on a screenshot of the UCSC Human Genome Browser. The red lines illustrate the selection of overlapping clones from a comprehensive BAC library.

  1. 10.3. ArrayCGHanalysis ofasquamouscelllung carcinoma using a sub-megabase-resolution whole genome tiling path BAC array comprising more than 36,000 spots. Cy3:Cy5 intensity ratios of each clone are plotted in a size-dependent manner along the chromosome ideograms. The red and green lines indicate the log 2 ratio thresholds -0.3 (loss) and 0.3 (gain), respectively. Note the very small high copy amplicons that would have been missed by low-resolution methods .
  2. 10.3. ArrayCGHanalysis ofasquamouscelllung carcinoma using a sub-megabase-resolution whole genome tiling path BAC array comprising more than 36,000 spots. Cy3:Cy5 intensity ratios of each clone are plotted in a size-dependent manner along the chromosome ideograms. The red and green lines indicate the log 2 ratio thresholds -0.3 (loss) and 0.3 (gain), respectively. Note the very small high copy amplicons that would have been missed by low-resolution methods .
  3. A protocol for the high-throughput isolation and purification of BAC/PAC clone inserts can be downloaded from our website (http://www.molgen.mpg.de/~abt_rop/ molecular_cytogenetics/Protocols.html). Whatever protocol is followed, the setup of a comprehensive BAC array platform remains time consuming and costly. Other shortcomings of BAC arrays are directly related to specific features of the respective genomic sequence. Otherwise unspecific binding repetitive sequences must be blocked using a considerable excess of (expensive) Cot-DNA. Low copy repeats, that is, stretches of DNA that are longer than 1 kb and have a sequence similarity of more than 90% to other locations in the genome, can lead to ambiguous results. This problem especially applies to tiling path arrays because low-resolution arrays usually avoid low copy repeats. 1 3 Finally, with the coming of tiling path arrays, BAC arrays have met the limits of resolution, which are simply given by clone insert size.

Repeat-Free and Nonredundant Sequence Arrays

In the light of the problems connected to the presence of repetitive sequences in the genome, researches have set out to generate genomic arrays that are depleted for repetitive and redundant sequences. This depletion, for example, has been accomplished by means of selective amplification using sequence-specific primers. Mantipragada et al have used this approach to create arrays focusing on the DiGeorge region (22q11 deletion syndrome).14 Another array, based on sequence-specific polymerase chain reaction (PCR) products of 162 exons of five genes, has been generated to test a spectrum of inherited human disorders.15 However, workload and high costs associated with this approach, as well as the upcoming of commercial oligonucleotide arrays, have hampered the widespread use of this approach.

Oligonucleotide Arrays Using Presynthesized Oligonucleotides

In contrast to the oligonucleotide platforms described below, these arrays are either generated by printing prefabricated, commercially available sets of oligonucleotides on glass slides or by coupling oligonucleotides to beads that are assembled on the slide afterward. One example for the use of printed arrays is reported by Carvalho et al, who have used a set of 18,861 oligos to identify DNA copy number changes in several tumor cell lines. 1 6 Presynthesized oligos coupled to beads are used for a platform developed by Illumina (http://www.illumina.com). The company provides several designs that are dedicated to either gene expression, linkage, or DNA copy number analysis. Some of their array designs enable the simultaneous detection of DNA copy number changes and loss of heterozygosity (LOH).17

Although prefabrication of oligonucleotides enables highly efficient synthesis, at the same time it also reduces flexibility in terms of sequences on the array. Custom design becomes considerably expensive and requires a minimal batch size to pay off.

Oligonucleotide Arrays Based on In Situ Synthesis

Meanwhile, there are numerous ways to synthesize an oli-gonucleotide directly on the slide. Despite this diversity, the common principle is shared and is already known from PCR primer synthesis: the growing oligonucleotide is alternatively exposed to As, Gs, Cs, and Ts, but only when the last oligonucleotide of the growing chain is activated by splitting off a protective group can a new nucleotide be attached. The main difference between the platforms comes from how this protective group is inactivated. Some companies use light, selectively distributed through fixed photolithographic masks (www.affymetrix.com) or micro-mirrors (http://www. nimblegen.com/http://www.febit.de). Others de-protect by means of a current-induced change of pH value (http:// www.combimatrix.com/) or control synthesis by specifically addressing each spot separately with high-resolution printers (http://www.home.agilent.com). Usually, oligonucle-otides on such arrays are designed to be both isothermal (i.e., they share the same melting temperature, Tm) and single copy sequences; this sometimes results in an uneven distribution of oligonucleotides, leading to considerable variability in terms of resolution across the genome. Nevertheless, given the current developments, it can be expected that oligonucleotide arrays will replace all other platforms in the near future. Shortcomings with respect to hybridization kinetics (see below) and coverage will be compensated by the incredible increase of features on the array. Oligonucleotide arrays with more than 500,000 features are readily available. Those platforms that are not dependent on fixed photolithographic masks especially can offer extreme flexibility, which is limited only by the setup fees charged by some companies. Single nucleotide polymorphism (SNP) arrays, consisting of short oligonucle-otides in the range of 16-20 mers and originally dedicated to linkage analysis, have been successfully used for the simultaneous detection of LOH (loss of heterozygosity) and DNA copy number changes.18

General Platform Considerations

Before setting up an array CGH facility, several decisions have to be made. The first one refers to the expected number of array CGH experiments. In many cases, it will be much cheaper to cooperate with other groups that already have established the technique or send the samples to a company which is offering a hybridization service. The next decision concerns the type of array that should be used. Certainly, this decision depends on the scientific problem that should be addressed with the analysis, but often the consequences of this decision are far ranging. Frequently, the choice implies the purchase of expensive machines only useful for arrays sold by the same company; this is especially true for the most expensive devices necessary for array CGH analysis, namely, the hybridization machine and the scanner. A hybridization machine is designed to provide controlled temperature and even circulation of the hybridization mix to promote hybridization efficiency (see below). Some of these machines also accomplish the posthybridization washing of slides. Important criteria when selecting a particular machine could be flexibility in terms of slide formats, handling, maintenance/ follow-up costs, and, most important, performance in one's own laboratory. High-quality scanners are essential for the errorless readout of hybridization results. Reliability, flexibility, and resolution, the latter especially in the light of the continuing minimization of feature sizes on the array, are important issues. Other arguments can be the availability of autoloaders to support high-throughput analysis or the need for more than two color channels. Note that there are devices that are scanning from the back of the slide and need transparent substrates.

+1 0

Post a comment