MATERIALS AND METHODS

BAC array preparation

DNA from the 1 Mb resolution Sanger BAC set (http://www.ensembl.org/Homo_sapiens/cytoview), the OncoBac set (http://informa.bio.caltech.edu/Bac_onc.html), and in house clones of interest, supplemented with clones from the Children’s Hospital Oakland Research Institute (CHORI), amounting to a total of 4202 clones with known chromosomal positions, was isolated according to the CHORI protocol (http://bacpac.chori.org/dnaprep.htm). Amplification of the DNA was done by ligation mediated polymerase chain reaction (PCR) according to Snijders et al,⁷ and PCR clones were purified using Montage PCR^® μ96 filter plates (Millipore BV, Amsterdam, the Netherlands). Clones were spotted at a concentration of 10 μg/μl, in 150mM sodium phosphate (pH 8.5), on CodeLink™ slides (Amersham BioSciences, Roosendaal, the Netherlands), using a Spot Array 72 robot (Perkin Elmer Life Sciences, Zaventum, Belgium), and processed according to the manufacturer’s protocol.

Oligoarray preparation

A collection of 18 861 oligo sequences, representing 18 664 unique genes (Sigma-Genosys, Zwijndrecht, the Netherlands), was spotted at a concentration of 10μM, in 150mM sodium phosphate (pH 8.5), on CodeLink™ slides (Amersham BioSciences), using a Spot Array 72 robot (Perkin Elmer Life Sciences), and processed according to the manufacturer’s protocol.

Hybridisation procedure

Genomic DNA was isolated from normal kidney (female donor), blood (male donor), or liver (female donor) as a reference, and from the breast cancer cell line BT474, according to Snijders et al.⁷ The DNA from the GM00143 cell line (Coriell Institute for Medical Research, Camden, New Jersey, USA) was provided by the Albertson laboratory (University of California, San Francisco, USA). For the BAC arrays, DNA labelling and hybridisation were performed according to Snijders et al,⁷ using ProbeQuant™ G50 microcolumns (Amersham Biosciences). For the oligoarray hybridisations, the concentration of formamide was decreased to 34% and washes were performed at 36°C to compensate for the reduced target length. Before scanning slides (Scan Array Express; Perkin Elmer Life Sciences) both oligoarrays and BAC arrays were dried by centrifugation (141 ×g), omitting DAPI staining. Spots were quantified by Imagene 5.5.4 software (Biodiscovery Ltd, Marina del Rey, California, USA), using the default settings. Subtraction of local background was done for the signal mean intensities in both test and reference DNA channels. Normalisation in the calculated ratios was done against the average of all ratios. On the BAC arrays, because the clones were spotted in triplicate, average and standard deviations were calculated for each clone. On the oligoarrays, each experiment was performed three times and data were taken from one representative experiment. Average and standard deviations of log₂ ratios were calculated for each oligonucleotide across the three experiments. A moving average (window of eight by eight) was applied to plot genome wide graphs.

RESULTS AND DISCUSSION

We hybridised a 19 K human 60 mer oligoarray with breast tumour cell line (BT474) DNA, labelled with Cy3, and normal genomic kidney (female) DNA, labelled with Cy5. Ratios for the non-flagged oligos (35%) were ordered by their position on the chromosome (June 2002 freeze; http://genome.ucsc.edu/). We compared the oligo CGH profile with the BAC array CGH profile (fig 1). Both array profiles showed the same pattern—for example, on the short arm of chromosome 1 neither profile showed a change in DNA copy number, whereas on the q arm two amplified areas are present in both profiles. No aberrations can be seen on chromosome 2. The standard deviation of the log₂ ratio of the individual probes was 0.21 for the BAC array and 0.45 for the oligoarray. On chromosome 3, a loss on the short arm was evident on the oligoarray, and was also seen with the BAC array (fig 1). Figure 2 shows two regions of amplification on the q arm of chromosome 17: one narrow peak over the chromosomal region containing c-Erb-B2/neu (Her2)¹⁰ and a second amplicon distal to c-Erb-b2. The BAC array has three clones over c-Erb-B2, and the best possible judgment towards the start and end of the amplicon, according to the April 2003 freeze, is therefore 2.5 Mb. With the oligo approach, 38 non-flagged oligos represent amplified ratios in this region and the size of the amplicon is 2.4 Mb according to the April 2003 freeze. Thus, the actual resolution in the region is 63 kb on average. The log₂ ratios for the three replicate BACs containing the c-Erb-b2 gene are 2.91, 3.06, and 2.53, a similar order of magnitude as that obtained in three independent experiments with the oligoarray for the single oligo corresponding to the c-Erb-b2 gene: 3.4, 3.4, and 3.9. The same observation can be made for the second amplicon on chromosome 17q.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Microarray comparative genomic hybridisation (CGH) profile of the BT474 cell line. (A) Bacterial/phage artificial chromosome polymerase chain reaction representations were used as probe DNAs spotted on to the glass slide; (B) oligonucleotides were used as probe DNAs spotted on to the glass slide. Moving average applied to the log₂ ratio of the oligo CGH profile. The vertical bars indicate the spacing between the chromosomes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Microarray comparative genomic hybridisation (CGH) profile of the long arm of chromosome 17 of the BT474 cell line. Grey triangles, bacterial/phage artificial chromosome polymerase chain reaction representations were used as probe DNAs spotted on to the glass slide; black squares, oligonucleotides were used as probe DNAs spotted on to the glass slide (log₂ ratio without moving average). The horizontal bars indicate the two separate amplicons observed.

The numbers for the comparison of those specific arrays not withstanding, it is obvious that an almost infinite amount of oligos can be rapidly synthesised for a specific region of interest, whereas the BACs are limited, especially by size, but also because availability and processing of the BACs is slower than for oligos.

As a further test of whether the oligo CGH is sensitive enough to detect single copy number changes, we performed a hybridisation with DNA from cell line GM0143 (female donor), which has trisomy of chromosome 18, and normal liver DNA (female donor). The theoretical ratio of a trisomy should be 1.5 (log₂ ratio, 0.58) and the BAC array showed a log₂ ratio value of 0.49 (SD, 0.05).⁷ For the oligoarray the average of the log₂ ratio was 0.31 (SD, 0.49) (fig 3). Although this single gain can be clearly distinguished by using a moving average (eight by eight), the high standard deviation indicates the need to optimise the hybridisation conditions.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Microarray comparative genomic hybridisation profile of the GM00143 cell line with a trisomy of chromosome 18. Moving average applied to the log₂ ratio. The vertical bars indicate spacing between the chromosomes.

“An almost infinite amount of oligos can be rapidly synthesised for a specific region of interest, whereas the bacterial artificial chromosomes are limited, especially by size”

In this first set of experiments, the percentage of spots that were flagged was very high (65%). These non-hybridising oligos could be partially explained by the fact that some oligos span exon–exon boundaries. Therefore, a dedicated oligo library specifically designed for array CGH, analogous to the situation with expression microarrays, could improve the performance of oligo CGH even further.