|Home | About | Journals | Submit | Contact Us | Français|
Array-CGH involves the comparison of a test to a reference genome using a microarray composed of target sequences with known chromosomal coordinates. The test and reference DNA samples are used as templates to generate two probe DNAs labeled with distinct fluorescent dyes. The two probe DNAs are co-hybridized on a microarray in the presence of Cot-1 DNA to suppress unspecific hybridization of repeat sequences. After slide washes and drying, microarray images are acquired on a laser scanner and fluorescent intensities from every target sequence spot on the array are extracted using dedicated computer programs. Intensity ratios are calculated and normalized to enable data interpretation. Although the protocols explained in this chapter correspond primarily to the use of large-insert clone microarrays in either manual or automated fashion, necessary adaptations for hybridization on microarrays composed of shorter target DNA sequences are also briefly described.
Array-CGH was developed in the late nineties (1,2) to detect DNA copy number changes at high resolution along the genome or locus of interest (see Chapter 3 for a general introduction on the method). This chapter details the methods employed to label, hybridize and detect genomic DNA probes for array-CGH.
As a first step, two genomic DNA samples - one test and one reference - are labeled separately using two different fluorescent dyes: generally Cyanine 3 (Cy3; excitation/emission wavelength maxima: 550/568nm) and Cyanine 5 (Cy5; 650/668nm). The probe DNA is commonly generated by enzymatic incorporation of Cy3 or Cy5 labeled nucleotides (usually dCTP or dUTP). Several strategies are possible for genomic DNA labeling, such as nick translation (1,3), direct labeling PCR (4) or random priming (5). Nick translation requires relatively high quantities of input DNA (usually 2μg), which can be problematic when working with small and/or precious DNA samples. PCR labeling overcomes this problem, as DNA is amplified and labeled in the same reaction. However, the exponential amplification of genomic DNA can bias the representation of the target by the amplified probe, thus leading to artifactual ratio variations after hybridization (4). Random primed labeling represents a good compromise between the two first techniques: (1) it requires moderate quantities of input DNA (usually 50 to 500 ng); (2) the representation of target sequences is not biased in the resulting probe, as the reaction consists of a moderate linear amplification of the template DNA by the Klenow fragment, using a mix of random hexanucleotides as primers. The probe DNA can also be labeled by chemical techniques such as cis-platinum labeling (6), which is based on the capability of monoreactive cisplatin derivatives to react at the N7 position of guanine moieties in DNA (7). However, cis-platinum labeling requires more than one microgram of template DNA as the chemical reaction does not create new DNA molecules.
After DNA labeling, both probes are co-precipitated and dissolved in hybridization buffer. The composition of this buffer, combined with the temperature of the hybridization, is critical to enable efficient and specific hybridization of the probes to the target DNA printed on the microarray. The hybridization buffer used in these protocols contains 50% of formamide, 0.1% of Tween20 and 5 to 10% of dextran sulfate. Formamide acts as a denaturing agent, which increases the hybridization stringency and prevents unspecific hybridization at lower temperatures such as 37°C. Tween20 is a nonionic detergent which minimizes non-specific fluorescence background on the surface of the slide. Dextran sulfate is a neutral component which consists in polymers of anhydroglucose in aqueous solutions: in homogeneous solution, it excludes DNA from the volume occupied by the polymer. In consequence, DNA concentration is “artificially” increased, which improves hybridization kinetics (8).
Hybridization to the array is performed at 37°C in the presence of Cot-1 DNA. Cot-1 DNA is the fraction of genomic DNA consisting largely of highly repetitive sequences. It is obtained from total genomic DNA by selecting for the most rapidly re-associating DNA fragments after denaturing. A large excess of Cot-1 DNA suppresses the hybridization of high-copy repeat sequences that are present in labeled DNA probes and thus prevents their hybridization to the corresponding repeat sequences that are also present in the target DNA (9,10).
After hybridization, slides are washed several times in PBS/0.05% Tween20 to remove excess hybridization buffer and reduce nonspecific background signal at the surface of the slide. The most critical wash is performed at 42°C in 50% formamide/2x SCC (or at 54°C in 0.1x SSC in the automated protocol): this stringent wash is essential for the selective and efficient elimination of probe fragments that are not hybridized specifically to the target arrayed DNA.
After washes and slide drying, microarray images are acquired for data analysis. Array-CGH was originally developed from CGH on chromosomes, a method in molecular cytogenetics using fluorescence microscopy (11,12). In consequence, the first acquisition systems used a CCD camera coupled to 0.5x or 1x magnification optical system (1) or a confocal laser scanning microscope (2). Today, with the huge expansion of microarray technologies, many scanners specific for DNA microarrays have become available that enable quick and easy image acquisition. The resolution of these scanners is typically 5 to 10 microns per pixel. Images are usually saved under Tagged Image File Format (TIFF), which is compatible with many distributed image analysis programs.
The methods below describe in detail how to label probes by random priming and how to hybridize and detect them on CGH microarrays (in particular on large-insert clone microarrays constructed using methods in Chapter 16) by either manual or automated procedures. Array-CGH profiles can then be produced from array images and interpreted using strategies described in Chapter 3.
Quantities and volumes described here correspond to the use of 2×3cm microarrays or Tecan Hs. PRO 51×20mm hybridization chambers: they should be rescaled when using arrays or chambers with different dimensions. The protocols were developed for the use of large-insert clone microarrays, constructed following the protocols described in Chapter 16. However, this protocol has been applied successfully to microarrays composed of smaller target DNA sequences, such as small-insert clones (size ranging from 1 to 4kb) or PCR products (150bp to 1kb), by following the automated procedure with slight modifications as described in the footnote of Table 1.
The manual protocol described in paragraph 2.2 enables the processing of up to 8 slides per person per day. For high-throughput array-CGH analysis, we have adapted the protocol for the use of automated hybridization stations (Hs400PRO/Hs4800PRO; Tecan, Inc). The Hs4800PRO station enables one person to process 12 slides per unit per day, and can be configured with up to 4 independent 12-position units.
There are two main changes in the automated procedure: (1) The hybridization buffer A contains only 7.5% of dextran sulfate: this reduces the viscosity of the solution and makes it compatible with injection and mixing in the hybridization chambers; (2) the second wash - with 50% formamide/2x SSC at 42°C for 30 minutes - has been replaced by washes with 0.1x SSC at 54°C, to preserve the components of the hybridization station. The Hs. PRO station should be programmed for array-CGH as described in Table 1, before starting Step 4 of the protocol below (see Note 5).
The authors would like to thank Heike Fiegler who developed some of the methods described in this chapter. This work was supported by the Wellcome Trust.
1Cy3 and Cy5 are the most commonly used fluorochromes for DNA labeling. However, both molecules are very sensitive to environmental conditions such as light, high ozone and humidity levels (13). To avoid any problem of dye degradation or fading, all experiments from DNA labeling to slide scanning should be performed in a laboratory with controlled temperature (20-25°C), relative humidity (25-35%) and ozone level (<0.02ppm). Alternatively, to limit premature degradation of Cy3 and Cy5, an anti-oxidant, such as cysteamine, can be added to hybridization buffers A and M (cysteamine 10mM) as well as PBS / 0.01% Tween 20 (cysteamine 2mM).
2The quality and purity of DNA samples used as templates for DNA labeling should be carefully monitored. Genomic DNA should show no degradation (by electrophoresis on agarose gel; see Fig. 1A) and no protein contamination (on spectrophotometer, 280/260 ratio should be greater than 1.8; see Fig. 1B, 1C). The use of DNA samples which do not fulfill these quality criteria may result in failure of the labeling reaction or low quality of the array-CGH results (i.e. higher technical variability of the array-CGH profile impairing the detection of copy number changes).
3DNA probe quantity and quality should be controlled after removal of unincorporated labeled nucleotides. First, 3μl of each sample collected in step 8 should be run on a 2.5% agarose gel to check for the presence of a DNA smear with most fragments below 500bp (Fig. 1A). In addition, probe DNA concentration as well as Cy3 or Cy5 incorporation can be measured by using only 1μl of each collected sample using a NanoDrop Spectrophotometer (Fig. 1B, 1C).
4One critical factor for the success of array-CGH is the efficient suppression of repeated sequences by the Cot-1 DNA during hybridization (see Introduction). We have noticed that commercially available Cot-1 DNA tends to show batch to batch variations in terms of suppression efficiency. Suppressive hybridization with high quality Cot-1 DNA results in log2ratio values close to 0.6 for single copy gains and −1 for single copy losses. The use of lower quality Cot-1 DNA may result to incomplete repeat suppression, compressed abnormal ratios and increased background ratio variability.
5The hybridization stations Hs.400PRO/4800PRO replace two other stations, which are still available: Hs400 and Hs4800. These two non-PRO stations can also be used for array-CGH. In this case, as the mixing system is different in the two non-PRO stations, the hybridization buffer A should contain only 5% of dextran sulfate instead of 7.5% in order to reduce the viscosity of the solution.
6Array images are generally acquired using a commercial DNA microarray scanner. We routinely use the scanner from Agilent technologies: this fully-automated system (with a 48-slide loading carousel) uses dynamic auto-focus to keep features in focus while scanning. Scanners can also be purchased from companies such as Perkin Elmer, Tecan or Axon Instruments. There are several software solutions available for image quantification (see example of microarray image in Fig.2A). Although spot intensity extraction programs are usually supplied with commercially available scanners, they can be purchased separately. The usual commercial programs include GenePix (Axon Instruments), ScanArray (Perkin Elmer), Agilent Feature extraction software (Agilent Technologies) and Bluefuse for microarrays (BlueGnome Ltd). Other programs are freely available, such as TIGR Spotfinder (14) (http://www.tm4.org/spotfinder.html) or UCSF Spot (15) (http://www.jainlab.org/downloads.html). For further data analysis (described in Chapter 3), the ratio values between the fluorescent intensities in both channels on every spot on the array are calculated (usually after the subtraction of local background fluorescence). Intensity ratios are then normalized, for example by dividing each individual ratio by the median ratio value of all clones. The normalized ratio for each clone is then plotted against its position along the genome (see example of normalized array-CGH profile in Fig. 2B).