Detecting single DNA copy number variations in complex genomes using o
http://www.100md.com
《核酸研究医学期刊》
Génomique Cellulaire des Cancers, CNRS UMR 8125 and 1 Service de Génétique, Institut Gustave Roussy, 39 rue Camille Desmoulins, 94805 Villejuif Cedex, France, 2 Service de Génomique Fonctionnelle, CEA and 3 Genoscope, 2 rue Gaston Crémieux, 91057 Evry Cedex, France
* To whom correspondence should be addressed. Tel: +33 1 42 11 54 21; Fax: +33 1 41 11 52 60; Email: gdanglot@igr.fr
ABSTRACT
Comparative genomic hybridization to bacterial artificial chromosome (BAC)-arrays (array-CGH) is a highly efficient technique, allowing the simultaneous measurement of genomic DNA copy number at hundreds or thousands of loci, and the reliable detection of local one-copy-level variations. We report a genome-wide amplification method allowing the same measurement sensitivity, using 1 ng of starting genomic DNA, instead of the classical 1 μg usually necessary. Using a discrete series of DNA fragments, we defined the parameters adapted to the most faithful ligation-mediated PCR amplification and the limits of the technique. The optimized protocol allows a 3000-fold DNA amplification, retaining the quantitative characteristics of the initial genome. Validation of the amplification procedure, using DNA from 10 tumour cell lines hybridized to BAC-arrays of 1500 spots, showed almost perfectly superimposed ratios for the non-amplified and amplified DNAs. Correlation coefficients of 0.96 and 0.99 were observed for regions of low-copy-level variations and all regions, respectively (including in vivo amplified oncogenes). Finally, labelling DNA using two nucleotides bearing the same fluorophore led to a significant increase in reproducibility and to the correct detection of one-copy gain or loss in >90% of the analysed data, even for pseudotriploid tumour genomes.
INTRODUCTION
Bacterial artificial chromosome (BAC)-array comparative genomic hybridization (array-CGH) is a very powerful technique, allowing the simultaneous quantitative analysis of all regions of large genomes. This approach was initially developed to identify the genomic regions gained and lost in human cancer cells (1–3), but also gave fruitful results in the analysis of constitutional DNA of patients affected by genetic disorders (4–6), in tumour animal models (7,8), and more recently, in interspecies comparisons between human and great ape genomes (9), as well as in characterization of very large size polymorphisms among human populations (10). A 32 000 BAC-array with complete coverage of the human genome is now available for all these applications (11). The array-CGH technique, also called matrix-CGH, involves the co-hybridization of differentially labelled specimen and reference genomic DNAs to an array of large insert genomic clones. The size of these targets, spotted on glass slides, allows the fully reproducible detection of single-copy gains and losses in the analysed DNA. However, the minimal amount of specimen DNA to be hybridized is 500 ng to 3 μg, obtained from a homogeneous cell population. Such amounts are often difficult to obtain from a restricted number of cells, especially when working with samples from infants, needle aspirates from tumours or laser capture-microdissected samples from heterogeneous tissues. To overcome this problem, we developed a procedure of whole genome amplification, especially adapted to high-sensitivity BAC-array CGH.
Several genome-wide amplification protocols have already been described, such as degenerated oligonucleotide-primed–PCR , ligation-mediated PCR (14), interspersed-repetitive-sequence–PCR , primer extension pre-amplification , strand displacement amplification (17) and rolling circle amplification (18). These techniques gave valuable results when used to sequence specific genes or to characterize various alleles. However, examination of quantitative analyses revealed occurrence of bias during amplification with most of them (19–21). According to these authors, the most reliable technique would be multiple strand displacement amplification with -29 DNA polymerase or related enzymes. However, these authors still indicate that the analysed loci are represented in the amplified DNA between 0.5 and 3.0 times the copy number in the starting genomic DNA template. This variation is still much higher than what can be tolerated in BAC-array CGH, which is specifically devoted to the detection of single-copy gains and losses in a diploid background. Even after amplification, one-copy gain or loss in tumour DNA (giving ratios of 1.5 and 0.5, respectively) has to be indisputably differentiated from normal genomic status (giving a ratio of 1), since these alterations have dramatic effects on tumour evolution, and finally on patient prognosis and survival (22–24).
We wanted to use the simplest method that would permit true amplification of the whole genome without any use of random primers, these being a source of inescapable affinity diversity for their respective targets. Accordingly, ligation-mediated PCR was favoured. This approach was previously proved efficient in whole genome amplification from a single cell (25). However, stoichiometric amplification of any complex DNA sample containing multiple fragments of various sizes and sequence compositions still remains a problematic goal (26). Here, we report the diverse parameters of amplification tested, and the optimized protocol successfully established. This approach efficiently amplifies as little as 1 ng of genomic DNA, and faithfully preserves the initial ratios observed with BAC-array CGH, allowing the reliable detection of one-copy-level variations among the amplified material.
MATERIALS AND METHODS
Cell lines
Human cell lines, all derived from neuroblastoma tumours (NJB, LAN1, LAN5, IGRN91, IMR32, SKNSH, SKNAS, SKNBE-2, CLB-GE and CLB-RE) were obtained from different laboratories, and cultured in RPMI-1640 medium supplemented with 10% foetal bovine serum, 2 mM L-glutamine, 100 U/ml penicillin and 100 mg/ml streptomycin. Genomic DNA was extracted using the DNeasy tissue kit (Qiagen) according to the manufacturer's instructions.
Construction of BAC/PAC microarray
BAC and P1-derived artificial chromosome (PAC) clones were selected from Pieter De Jong libraries (http://www.chori.org/bacpac/) according to their position on UCSC human genome . Clones were grown in 96-well blocks containing 1.2 ml of 2YT medium (Sigma) supplemented with 12.5 μg/ml chloramphenicol or 25 μg/ml kanamycin, for BACs or PACs, respectively, and DNA extracted according to the protocol used at Genoscope (Evry, France) (28). The authenticity of the clones was verified by fluoresence in situ hybridization (FISH) to normal human metaphase preparations. Valid BAC/P1 DNAs were amplified by ligation-mediated PCR as published by Snijders et al. (3) with the use of Taq DNA polymerase (MBI Fermentas). PCR products were ethanol-precipitated and re-dissolved in H2O–50% dimethyl sulfoxide. DNA solutions (final concentrations 800 ng/μl) were arrayed onto amino-silane-coated glass slides (Corning GAPS II) in six replicates using a Microgrid Pro spotter (BioRobotics Ltd).
Genome-wide amplification of specimen DNA
The amplification procedure used a ligation-mediated PCR technique. Multiple parameters were varied in order to identify the conditions most suited to fully quantitative amplification of all genomic regions. The details of these tests are presented in the results. The optimal protocol finally selected was the following. Human large-size genomic DNA was digested by Mse1 (2 U; New England Biolabs) in One-Phor-All Buffer PLUS (Amersham) and, after thermal inactivation of enzyme, directly processed to the following step. DNA concentration was determined on a Nanodrop spectrophotometer (Nanodrop Technologies, Inc.) and the digested DNA was adjusted to the concentration of 1 ng/μl. Ligation was performed by mixing 1 μl of the diluted DNA with 3.5 μl H2O, 0.5 μl of 10x One-Phor-All Buffer PLUS, 0.5 μl of 100 μM primer L6 (5'-CACTAGGAGCTGGCAGATCGTACATTGA-3'; Invitrogen) and 0.5 μl of 100 μM primer C8 (5'-TATCAATGTACGA-3'). After heating for 1 min at 65°C, the temperature was lowered to 15°C, ramping at 1°C/min. A mixture containing 2 μl H2O, 1 μl of T4 DNA ligase (5 U/μl; MBI Fermentas) and 1 μl of 10 mM rATP (Promega) was then added, and the ligation was performed overnight at 15°C. The ligation products were incubated for 4 min at 65°C and stored at 37°C. For PCR amplification, a 40 μl mixture, consisting of 31.2 μl H2O, 4.5 μl Taq PCR Buffer10x (MBI Fermentas), 2 μl of dNTPs (10 mM each; Amersham), 0.5 μl of 100 μM primer L6, 0.8 μl of 100 mM MgCl2 and 1 μl of Taq DNA polymerase (5 U/μl; MBI Fermentas), was similarly pre-warmed to 37°C and added to the ligation mixture. The 50 μl reaction mixture was incubated for 3 min at 68°C before starting the PCR programme: for 14 cycles: 94°C for 50 s, 65°C for 1 min and 68°C for 3 min. Amplification products were purified with Qiaquick gel extraction kit (Qiagen) and eluted in 40 μl. DNA concentration was measured and the size of amplified material determined by agarose gel electrophoresis.
DNA labelling for array-CGH
Test and reference DNAs were labelled by random priming in 25 μl reaction volumes. Briefly, a 17.5 μl pre-mixture was prepared containing between 0.4 and 0.6 μg of genomic Mse1 digested DNA (possibly amplified), and a final concentration of 0.2 mM random decamers solution (Invitrogen) in Klenow fragment exo(–) buffer. After denaturing DNA for 5 min at 95°C, two different procedures were tested. In the first one, only one fluorescent dNTP was used. A 7.5 μl mixture was added, leading to a final labelling reaction containing 0.2 mM dATP, 0.2 mM dGTP, 0.2 mM TTP, 0.1 mM dCTP, 0.1 mM Cy3 or Cy5-dCTP (Amersham) and 12.5 U Klenow fragment (5 U/μl; MBI Fermentas). In the second one, both fluorescent dCTP and dUTP similarly labelled were added to the same reaction. The labelling mixture finally contained 0.2 mM dATP and dGTP, 0.075 mM dCTP and TTP, 0.05 mM Cy3- or Cy5-dCTP, 0.05 mM Cy3- or Cy5-dUTP and 12.5 U Klenow fragment. After labelling, unincorporated nucleotides were removed by use of the Nucleospin extract kit (Macherey-Nagel).
Array hybridization, imaging and data analysis
Labelled test and reference DNAs were mixed with 80 μg of Cot-1 DNA (Roche), 80 μg of salmon sperm DNA (Gibo-BRL), 80 μg of yeast tRNA (Sigma) and ethanol-precipitated. The pellet was dissolved in 9 μl H2O and 21 μl of 1.43x hybridization buffer, leading to a final composition of 50% formamide, 2x SSC, pH 7, 10% dextran sulphate and 4% SDS. Probes were denatured for 5 min at 73°C, and incubated at 37°C for 30 min to allow blocking of repetitive sequences. Microarray slides were briefly hydrated with steam and irradiated at 150 mJ (GS Gene Linker, Bio-Rad). Arrays were pre-hybridized for 30 min at 50°C in 30 ml of 3.5x SSC, 4% SDS and 10 mg/ml BSA (Sigma). After two rinses of 1 min in H2O and isopropanol, respectively, the slides were dried by centrifugation. Hybridization was performed under a 24 x 50 mm coverslip, in hybridization chambers (Corning) incubated for 72 h at 37°C. Slides were washed at room temperature for 2 min in 0.5x SSC 0.3% Igepal (Sigma), at 72–75°C for 2 min 30 s in the same solution, at room temperature for 10 min in 0.1x SSC and 0.1% SDS, and finally dried by centrifugation.
Arrays were scanned with an Agilent scanner. The UCSF ‘SPOT’ software (29) was used to locate spots automatically on the Cy3 and Cy5 image acquisitions and to calculate fluorescence ratios. A dedicated array-CGH Excel Macro was developed, which automatically analyses and summarizes the results as follows: (i) averages the ratios of the replicates and calculates the standard deviation; (ii) rejects individual spot data based on several criteria (including too weak fluorescent signals); (iii) adjusts Cy5/Cy3 ratios such that ratios of normal genomic regions are always equal to 1, despite variations in dye labelling efficiency; and (iv) plots data relative to the position of the clones on human genome (according to July 2003 UCSC cartography).
Conventional CGH on metaphase chromosome preparations and FISH
CGH was performed as described previously (30) with slight modifications. Briefly, 400 ng of tumour DNA and 200 ng of reference DNA (sexed normal human genomic DNA; Promega) were labelled by random priming in the presence of Alexa 488-dUTP or Alexa 594-dUTP (Molecular Probes), respectively. Probes were ethanol-precipitated with 20 μg of Cot-1 DNA (Roche), and hybridized to metaphase preparations for 3 days at 37°C. The slides were washed for 2.5 min in 0.5x SSC and 0.3% Igepal (Sigma) at 72°C, followed by 5 min in 1x PBS at room temperature. Images were acquired with a Vysis station and analysed with Quips XL imaging software.
FISH experiments were performed as described previously (31), except that probes were labelled by random priming in the presence of Alexa 488-dUTP or Alexa 594-dUTP.
RESULTS
Validation of BAC-array and array-CGH procedure
As an initial step, the BAC-array was tested for its ability to measure directly the genome copy number in normal cells and known cancer cell lines. Control co-hybridizations (n = 4) of normal human genomic male and female DNAs, alternatively labelled with Cy3 and Cy5, respectively, regularly displayed a signal-to-background ratio around 7. Only microarray elements with reference fluorescence intensity at least 3-fold over the background were taken into account. Of the 1578 array features printed on the slides, 98% passed this selection criterion. Hybridization to heterologous DNA (tomato BAC clone DNA spotted as control) gave a fluorescent signal twice that of slide background. Very homogeneous fluorescence ratios centring on 1 were observed on human array elements corresponding to autosomal regions, with all of the normalized ratios falling within the range 0.83–1.20. The standard deviation was 0.038, ranging from 0.021 to 0.052 according to the experiments. The single-copy difference tested on spots corresponding to X chromosome DNA, led to a mean male/female ratio of 0.65 instead of the theoretical value 0.5.
Tumour cell (n = 10) versus normal DNA co-hybridizations were also performed on the BAC-array, and at the same time tumour genomes were characterized using conventional CGH and FISH experiments. An example of such comparative analysis is presented in Figure 1. The results obtained from all 10 cell lines exhibited a very good correlation between the various techniques, with only 5–6% of spots giving discordant results in genomic status evaluation. Symmetrical ‘dye swap’ experiments, with tumour DNA being alternatively labelled with Cy5 or Cy3, as well as repetition of analogous experiments on array batches spotted from fully independent amplification series of BAC DNAs, similarly led to a reproducibility of 94–95% in the detection of single-copy gains and losses. However, while the reproducibility was even better with pseudodiploid tumour cell lines, the percentage of discrepancies increased when working with pseudotriploid cells. Such pathological genomes are common in tumour specimens. Careful examination of the results indicated that a small number of BAC clones, generally containing a biased base composition, were responsible for this decreased reliability. To limit the consequences of these GC% variations in human DNA regions, the labelling of test and reference DNAs was modified. Although it was initially performed in the presence of the single Cy-dCTP nucleotide, subsequent labelling reactions contained both cyanine-labelled dCTP and dUTP nucleotides (see Materials and Methods). This modification allowed a more homogeneous labelling of the various genomic regions, leading to an overall increase in ratio reliability (Table 1). All the results obtained in these control experiments indicate that the array and the procedure allow the reproducible detection of one-copy-level variations in specimen DNA, in a pseudodiploid or pseudotriploid genomic background.
Figure 1. Agreement between results from array-CGH, conventional CGH and FISH techniques, in genomic status evaluation of NJB tumour cell line. (A) Conventional CGH profile showing genomic regions gained (green) and lost (red) in the tumour DNA. (B) FISH experiment with two different 17q BAC clones hybridized to a tumour metaphase preparation. The first probe (RP11-51F16), labelled in red, contains 17q23.3 material, and indicates one-copy gain in the tumour genome. The second probe (RP11-115C3), labelled in green, is specific for the 17q24.2 region, and shows two additional copies of the region in tumour DNA. (C) Detail of an array-CGH image. Tumour DNA is labelled in red and normal DNA in green. The red spots indicate the amplification of 2p24 region coding MYCN oncogene. (D) The evaluation of DNA copy number in genomic regions of NJB cell line, according to the ratio obtained from the array-CGH experiment. Array-CGH results are fully consistent with the observations made with other techniques.
Table 1. Concordance between array-CGH and conventional CGH: effect of labelling conditions
Parameter optimization for whole-genome amplification
Human genomic DNA was restricted with Mse1, giving fragments in the range of 70–2000 bp, with the bulk of the material between 200 and 1000 bp as expected (25). As BAC DNAs spotted on the array had been amplified by a ligation-PCR procedure using the previously described adapter (3,25), a new adapter, devoid of any potential cross-hybridization was designed to amplify DNA specimens. A 28 nt primer (called L6) with a base composition of 50% GC and a complementary 13mer oligonucleotide called C8 were selected. After ligation of this L6/C8 adapter, whole genomic fragments were submitted to various conditions of amplification.
Several publications (32,33) suggest that ThermoSequenase could favourably replace Taq polymerase in DOP–PCR whole-genome amplification. These polymerases were then compared in their ability to synthesize DNA products of the same composition as the starting material, after an adapter ligation reaction. A test plasmid DNA was digested by Mse1 to obtain clear-cut class sizes of DNA fragments ranging from 250 to 800 bp, and diluted to 1 ng/μl. After the adapter ligation reaction, aliquots were amplified using either Taq polymerase or ThermoSequenase (Figure 2A). The results reproducibly showed a low efficiency of ThermoSequenase amplification in comparison with Taq polymerase, with amounts of synthesized material almost 3–4-fold lower. In addition, ThermoSequenase appeared unable to amplify DNA fragments smaller than 300 bp, even in the presence of primer concentrations reaching 16 μM. Taq polymerase, in contrast, amplified fragments of all size. The comparison of the two polymerases was repeated on genomic DNA with analogous results.
Figure 2. Optimization of several parameters of the ligation-mediated PCR amplification. The protocol detailed in Materials and Methods was applied in every case with the exception of the parameter analysed. Amplified DNAs are characterized by agarose gel electrophoresis. (A) Amplifications of a PUC18 plasmid using two different DNA polymerases and increasing concentrations of primer. Lanes 1–3: Amplifications using ThermoSequenase; lanes 4–6: amplifications using Taq polymerase; lanes 1 and 4: 4 μM primer L6; lanes 2 and 5: 8 μM primer L6; lanes 3 and 6: 16 μM primer L6; lane 7: starting material (pUC18 plasmid digested with Mse1); and lane 8: molecular weight markers (mixture of Lambda DNA/HindIII fragments, and PhiX174/HaeIII fragments). (B) Amplifications of a BAC clone (RP11-111M5) with Taq polymerase, to determine the most suitable ligation buffer and the optimal number of PCR cycles. Lanes 1–3: ligation performed in the presence of 0.5x One-Phor-All buffer; lanes 5–7: ligation performed in the presence of 1x ligation buffer; lanes 1 and 5: 14 PCR cycles; lanes 2 and 6: 17 PCR cycles; lanes 3 and 7: 20 PCR cycles; and lane 4: molecular weight markers. (C) Test of increasing concentrations of primers during the PCR step. Lane 1: 1 μM primer; lane 2: 2 μM primer; lane 3: 4 μM primer; and lane 4: 8 μM primer. (D) Size distribution of the DNA before and after amplification by the optimized protocol. Lane 2: 1 μg of BAC digested DNA; lane 3: 1 μg of PCR-amplified BAC DNA; lane 4: 400 ng of digested LAN5 cell line DNA; lane 5: 400 ng of LAN5 amplified DNA; and lanes 1 and 6: molecular weight markers.
To avoid template loss, restriction enzyme digest, adapter ligation and PCR amplification reactions were linked, without intermediate purification steps. In order to check for potential negative interference between ligation buffer and optimal PCR conditions, two ligation conditions were tested on a BAC-digested DNA (RP11-111M5), one in classical ligation buffer and the second in 0.5x One-Phor-All Buffer. Subsequent amplifications gave very analogous qualitative results, with a quantitative advantage with the latter buffer (Figure 2B, lanes 1 and 5). One-Phor-All Buffer was then used for ligation.
Comparative amplifications were performed in the presence of various primer L6 concentrations, ranging between 1 and 16 μM (Figure 2A and C). The best amplification of the various class sizes of DNA fragments was observed with primer concentrations of 1 and 2 μM. The latter concentration was finally used because it allows the synthesis of a larger quantity of DNA. Primer concentrations of 4 μM and above led to preferential amplification of short DNA fragments, to the detriment of fragments longer than 600 bp.
Genomic specimen DNA has to be amplified as much as possible. However, great care was taken to avoid potential hybridization of DNA fragments originating from different chromosomes (by means of the adapter sequence), especially during the last cycles of amplification. To analyse this particular point, the digested and ligated BAC DNA was amplified for 14, 17 and 20 cycles. As can be seen in Figure 2B, after 14 cycling amplifications, a banding profile analogous to that observed with the original material was still present. However, after 17 cycles of PCR, a smear of increasingly higher molecular weight DNA progressively replaced the characteristic DNA banding of the starting DNA, suggesting physical association with irrelevant fragments.
Characteristics of amplified DNA using the optimized protocol
Starting with 1 ng of digested genomic DNA in every case, 12 independent amplifications were performed according to the optimized protocol detailed in Materials and Methods. After purification of the amplified DNA on small columns, 2.6–3.2 μg of material was consistently obtained, with a mean of 2.9 μg for all 12 experiments. Analysis of the size of amplified material showed a very good correlation with starting material (see Figure 2D, lanes 4 and 5) with apparently similar distribution. Only very small fragments ranging from 70 to 150 bp were absent in the amplified material, but this was partly related to the fact that during the ligation step primers of 28 bp are added at both ends of the DNA fragments. Accordingly, no DNA fragments <125 bp were expected in the amplified material. However, in order to define more quantitatively the efficiency of amplification of DNA fragments of various class sizes, the optimized protocol was applied to digested BAC DNA (Figure 2D, lanes 2 and 3). The amount of each of the DNA fragments before and after amplification was assessed and the results are shown in Table 2. It can be seen that the amplification is not fully homogeneous, since the amplification ratio ranged from 0.4 to 1.3 according to the size of DNA fragments. The most efficient amplification was observed for DNA fragments between 155 and 1100 bp. Amplification efficiency declined for smaller as well as larger fragments. However, 70–75% of initial material was amplified very homogeneously, with the amplification ratio ranging from 0.9 to 1.3.
Table 2. Size influence on the efficiency of amplification of DNA fragments in a complex DNA mixture
Validation of the amplification protocol using array-CGH hybridization
An aliquot of 1 ng of the genomic DNA from the 10 cell lines used previously was amplified 3000-fold according to the optimized protocol, and the 10 non-amplified as well as the 10 corresponding amplified DNAs were hybridized to the BAC-array. Examination of the absolute fluorescence values retained on tomato BAC clones showed non-specific background levels completely identical between the two series (means of background values differed by 3% between amplified and non-amplified DNAs), demonstrating an absence in any non-specific hybridization related to primer L6.
Several authors characterizing amplified genomic DNA with conventional CGH experiments (33,34) as well as the cDNA-array CGH technique (20) reported a distortion of ratios when the amplification procedure was only applied to the test sample. We wondered whether the same phenomenon would occur with our amplification procedure. In order to verify this point, conventional CGH and array-CGH were performed, with non-amplified reference DNA co-hybridized with amplified test DNA. Conventional CGH results showed occasional deviations of the ratios at several chromosome loci in comparison with the profiles obtained from non-amplified test DNA. Moreover, array-CGH experiments on five cell lines confirmed this bias. Variable discordance rates were observed, reaching 20% in some cases, while only 5% of incorrect ratios were obtained without amplification. When test and reference DNAs were similarly amplified, the distortions disappeared and the ratios complied with those observed with non-amplified DNA, both with conventional CGH and array-CGH techniques.
The 10 cell lines were then analysed by array-CGH, using both test and reference DNAs amplified 3000-fold. The reliability of the amplification method was assessed by comparing each pair of ratios of each BAC clone from the amplified and non-amplified datasets. A linear regression performed on 2473 pairs of ratio values corresponding to genomic regions bearing low-copy-number variations led to a correlation coefficient of 0.959, which was quite similar to the correlation coefficient of 0.957 obtained from direct replicate experiments (Figure 3A). When ratios from all genomic regions were considered, including those coding for in vivo amplified oncogenes, the correlation coefficients increased further, reaching 0.990 and 0.986, respectively. The scatter plot of the ratios obtained from in vitro amplified versus non-amplified DNA (focusing on low-copy-variation regions) is presented in Figure 3B. The low scattering of the dots and the slope of 0.936, very close to 1, attest to the fidelity of the amplification method. Two examples of the accuracy of the amplification are presented in Figure 4. These diagrams show that ratios obtained with and without amplification of DNAs are almost perfectly superimposed, indicating that the amplification has retained the copy number variations present in the initial DNA specimens. To further verify this point, the genomic copy number determined by array-CGH with amplified DNA samples was exhaustively compared with the results from conventional CGH. The results, summarized in Table 3, confirm the accuracy of the optimized amplification protocol, and the reliability of the results, even in the most difficult case of pseudotriploid genomic status.
Figure 3. Comparison of array-CGH data obtained from a series of 10 neuroblastoma cell lines. (A) Correlation coefficients determined from (i) two series of non-amplified DNAs (Direct) using alternative labelling of tumour DNAs with Cy3 or Cy5 nucleotides and (ii) series of non-amplified versus amplified tumour DNAs. A total of 2473 pairs of ratios were considered, all corresponding to genomic regions of low-level variations. (B) Scatter plot of ratios obtained from array-CGH analysis of non-amplified versus amplified DNAs, compiling the results of regions of low-level variations in all neuroblastoma cell lines. The tight distribution of points along the fitted line illustrates the high fidelity of the amplification.
Figure 4. Focus on individual ratio values obtained from array-CGH analyses, with or without amplification of starting genomic DNA. (A) Fluorescence ratios (log 2) observed for the SKNAS cell line as test DNA, with BAC clones ordered from chromosome1p to chromosome Y. (B) Analogous results for IMR32 cell line. The superimposed values underline the fidelity of amplification.
Table 3. Concordance between array-CGH and conventional CGH: effect of DNA amplification
DISCUSSION
The aim of the experiments reported here was to amplify several 1000-fold, and in a nearly stoichiometric way, all regions of a large genome, in order to preserve the one-copy-level differences present in the original DNA specimen. Various amplification methods have already been proposed. However, the results of Daigo et al. (35) and Veltman et al. (5) showed a poor fidelity of DOP–PCR amplification for array-CGH measurements of genomic DNA copy numbers. Multiple displacement amplifications demonstrated their reproducibility for gene dosage alterations of 3-fold or more, but not for one-copy-level variations (19–20). Similarly, Ren et al. (36) and Liu et al. (37), as well as Tanabe et al. (38) proposed amplification methods starting from randomly shared genomic DNA, within the framework of Chromatin Immunoprecipitation (ChIP) studies. The use of ligation-mediated PCR, or of T7-based linear amplifications, appeared to give reliable results for ratios in the range of 1–25, but few if any analyses were performed for ratios close to 1. The ligation PCR-based amplification method proved to give valuable results in measurements of one-copy-level variations using conventional CGH (25,39). However, BAC-arrays are much more demanding, since conventional CGH averages the ratios obtained from 10 to 20 Mb regions on chromosome arms, whereas BAC-arrays explore regions 150 kb in size. Here we propose a protocol of whole-genome amplification based on ligation-mediated PCR, with optimized conditions specifically adapted to the sensitivity of the array-CGH technique. The principle of ligation-mediated PCR amplification offers several advantages over other techniques. First, it uses the same primer for all fragments to be amplified. This circumvents the bias amplification problems related to the variable affinity of random or degenerated primers towards their respective targets. Second, complementary sequences, present at both ends of each DNA strand, favour an internal pairing of the smallest molecules, which counterbalances their otherwise preferential duplication during the PCR process (26). Finally, cycling induces a synchronized duplication of the various genomic DNA fragments present in starting material. This is not the case with multiple displacement amplifications, where the first duplicated regions may be, in turn, rapidly used as template. The three elements favour quantitative and homogeneous whole-genome amplification, provided no limiting elements interfere with the procedure. Our results clearly confirm this observation, as the whole starting DNA fragments appeared to be amplified with similar efficiencies, the bulk of which were between 0.9 and 1.3.
Critical aspects of the protocol described herein, in relation to its use before array-CGH experiments, are the following. The adapter ligation reaction was performed at low-DNA template concentrations, using rather high adapter concentrations, to avoid any potential ligation events between genomic fragments from different chromosomes. Here, the use of 1 ng of genomic DNA template and 5 μM of the adapter led to concentration ratios in the range of 5000/1 for adapter ends and genomic fragment ends, respectively. The second crucial step of this protocol was the number of PCR cycles, in relation to the primer concentration. In order to obtain amplified DNA of the same size as starting DNA, a maximum concentration of 2 μM of primers should be used in the PCR. In these conditions, a reproducible synthesis of 3 μg of DNA was achieved from 1 ng of starting material, after a 14-cycle Taq PCR amplification. This DNA amount corresponds to the upper limit for safe amplification, because in these conditions 25% of the initial primer was incorporated in the neosynthesized DNA. When additional cycling steps were scheduled, a rapidly high proportion of these molecules hybridized between themselves, instead of hybridizing free primers. Thus, if the starting DNA has not been damaged during storage, and if no stray impurities are present in the initial sample, a better security margin would be obtained if only 13 amplification cycles were scheduled, in order to synthesize between 1 and 2 μg of amplified genomic DNA. The present results also confirm the high processivity of Taq polymerase when used in its optimal conditions, since the observed duplication factor was 0.89, in comparison with the theoretical factor of 1, if all molecules were duplicated at each cycling. Taking this doubling factor into account, the present protocol may be adapted. Cycling number might be increased when starting with <1 ng of genomic DNA, which approximately corresponds to 100 cells. Performing the ligation reaction with >1 ng could rapidly lead to an increasing number of false results, especially if the cycling number is not reduced in proportion. Ligation reactions using too much starting genomic DNA (in comparison with the adapter concentration) or too many PCR cycles would lead to unwanted physical association with genomic fragments originally far away in the genome. As a consequence of this potential problem, ratios of altered genomic regions progressively evolve towards normality.
To check this particular point, we used a series of 10 altered genomes to test the amplification procedure proposed. All the validation experiments, performed on arrays of 6 replicates of 250 BAC clones, demonstrated the faithful character of the results obtained from amplified genomic DNAs. No false hybridization occurred between the two primers used to amplify arrayed BAC DNAs and specimen DNAs, respectively. Pairs of ratios from direct and amplified genomic DNAs were very close or superimposed on diagrams corresponding to each cell line. The correlation coefficient of 0.96, determined from regions bearing small copy number variations, as well as the narrow distribution of each pair in the scatter plot of the data, strengthened the validity of the procedure. Finally, the comparison of the results obtained from array-CGH and conventional CGH, showing >90% concordance for individual BAC loci, even in a pseudotriploid context, firmly attests to the sound nature of the procedure. The observed reproducibility is even better than expected, in relation with the slight inhomogeneity of amplification of the DNA according to its size. Our results showed that DNA fragments of 70–150 bp, as well as those >1100 bp, were amplified with an efficiency about one-half of that observed for intermediate fragments. Together, these fragments represent 25–30% of the starting genomic DNA. Their decreased amplification might have generated bias in the ratios obtained from amplified DNA, but interestingly, this was not the case. Three explanations, either alone or in combination, may be proposed. The first is that an analogous amplification process is applied to the test, reference and BAC-arrayed DNAs. Thus, the same slight amplification bias on all these DNAs should neutralize each other, with no consequence for the fluorescence ratios. The second parameter deals with the fact that array-CGH is performed in the presence of high amounts of Cot1-DNA, to block repetitive sequences. The very small and too large fragments, present after Mse1 digestion of genomic DNA, could be specially enriched in repetitive sequences. If it was the case, their decreased amplification should be neutral for the accurate genomic copy number evaluation. Repetitive sequences represent 44.8% of the human genome (40). The third factor of reliability might reside in the fact that the amplified DNA comprises a mixture of fragments independently amplified. Several hundreds of them hybridize to each of the BAC clones spotted on the array. The fluorescence recorded on each spot represents the sum of these individual elements, and the small differences in the efficiency of amplification between the DNA fragments might counterbalance each other.
In conclusion, array-CGH on BAC clones is currently the best procedure for measuring precisely the genome copy number at a wide range of loci. In many circumstances, there is insufficient DNA to be characterized with this technique. Several procedures of whole-genome amplification have been developed in the last decade but none has demonstrated its reliability for subsequent one-copy-level variation analyses on BAC-arrays. Recent publications suggest that every whole-genome amplification procedure has specific advantages and pitfalls. At present, the multiple displacement amplification seems the most accurate procedure to be used before genotyping and gene sequencing. However, the optimized protocol of ligation-mediated PCR amplification presented here should be the most reliable procedure in measuring one-copy-level variations in minute amounts of genomic DNA with BAC-array CGH technology.
ACKNOWLEDGEMENTS
We are grateful to D. Fauvet, A. Pitaval, Z. Skalli, P. Coullin, B. Clausse, M.A. Dillies-Pelletier, V. Lazar, H. Ripoche and P. Dessen for helpful advice, and to V. Combaret for some tumour cell lines. Warmest thanks are due to L. Saint Ange for assistance with editing and to Prof. G. Lenoir for his support. This work was funded by grants from the CNRS, Institut Gustave Roussy, the Association pour la Recherche sur le Cancer (ARC) and Paris 11 University. M.G.-B. is supported by a doctoral fellowship from the CNRS.
REFERENCES
Solinas-Toldo,S., Lampel,S., Stilgenbauer,S., Nickolenko,J., Benner,A., Dohner,H., Cremer,T. and Lichter,P. ( (1997) ) Matrix-based comparative genomic hybridization: biochips to screen for genomic imbalances. Genes Chromosomes Cancer, , 20, , 399–407.
Pinkel,D., Segraves,R., Sudar,D., Clark,S., Poole,I., Kowbel,D., Collins,C., Kuo,W.L., Chen,C., Zhai,Y. et al. ( (1998) ) High resolution analysis of DNA copy number variation using comparative genomic hybridization to microarrays. Nature Genet., , 20, , 207–211.
Snijders,A.M., Nowak,N., Segraves,R., Blackwood,S., Brown,N., Conroy,J., Hamilton,G., Hindle,A.K., Huey,B., Kimura,K. et al. ( (2001) ) Assembly of microarrays for genome-wide measurement of DNA copy number. Nature Genet., , 29, , 263–264.
Bruder,C.E., Hirvela,C., Tapia-Paez,I., Fransson,I., Segraves,R., Hamilton,G., Zhang,X.X., Evans,D.G., Wallace,A.J., Baser,M.E. et al. ( (2001) ) High resolution deletion analysis of constitutional DNA from neurofibromatosis type 2 (NF2) patients using microarray-CGH. Hum. Mol. Genet., , 10, , 271–282.
Veltman,J.A., Schoenmakers,E.F., Eussen,B.H., Janssen,I., Merkx,G., van Cleef,B., van Ravenswaaij,C.M., Brunner,H.G., Smeets,D. and van Kessel,A.G. ( (2002) ) High-throughput analysis of subtelomeric chromosome rearrangements by use of array-based comparative genomic hybridization. Am. J. Hum. Genet., , 70, , 1269–1276.
Vissers,L.E., de Vries,B.B., Osoegawa,K., Janssen,I.M., Feuth,T., Choy,C.O., Straatman,H., van der Vliet,W., Huys,E.H., van Rijk,A. et al. ( (2003) ) Array-based comparative genomic hybridization for the genomewide detection of submicroscopic chromosomal abnormalities. Am. J. Hum. Genet., , 73, , 1261–1270.
Hackett,C.S., Hodgson,J.G., Law,M.E., Fridlyand,J., Osoegawa,K., de Jong,P.J., Nowak,N.J., Pinkel,D., Albertson,D.G., Jain,A. et al. ( (2003) ) Genome-wide array CGH analysis of murine neuroblastoma reveals distinct genomic aberrations which parallel those in human tumors. Cancer Res., , 63, , 5266–5273.
O'Hagan,R.C., Brennan,C.W., Strahs,A., Zhang,X., Kannan,K., Donovan,M., Cauwels,C., Sharpless,N.E., Wong,W.H. and Chin,L. ( (2003) ) Array comparative genome hybridization for tumor classification and gene discovery in mouse models of malignant melanoma. Cancer Res., , 63, , 5352–5356.
Locke,D.P., Segraves,R., Carbone,L., Archidiacono,N., Albertson,D.G., Pinkel,D. and Eichler,E.E. ( (2003) ) Large-scale variation among human and great ape genomes determined by array comparative genomic hybridization. Genome Res., , 13, , 347–357.
Locke,D.P., Segraves,R., Nicholls,R.D., Schwartz,S., Pinkel,D., Albertson,D.G. and Eichler,E.E. ( (2004) ) BAC microarray analysis of 15q11-q13 rearrangements and the impact of segmental duplications. J. Med. Genet., , 41, , 175–182.
Ishkanian,A.S., Malloff,C.A., Watson,S.K., DeLeeuw,R.J., Chi,B., Coe,B.P., Snijders,A., Albertson,D.G., Pinkel,D., Marra,M.A. et al. ( (2004) ) A tiling resolution DNA microarray with complete coverage of the human genome. Nature Genet., , 36, , 299–303.
Ludecke,H.J., Senger,G., Claussen,U. and Horsthemke,B. ( (1989) ) Cloning defined regions of the human genome by microdissection of banded chromosomes and enzymatic amplification. Nature, , 338, , 348–350.
Telenius,H., Carter,N.P., Bebb,C.E., Nordenskjold,M., Ponder,B.A. and Tunnacliffe,A. ( (1992) ) Degenerate oligonucleotide-primed PCR: general amplification of target DNA by a single degenerate primer. Genomics, , 13, , 718–725.
Saunders,R.D., Glover,D.M., Ashburner,M., Siden-Kiamos,I., Louis,C., Monastirioti,M., Savakis,C. and Kafatos,F. ( (1989) ) PCR amplification of DNA microdissected from a single polytene chromosome band: a comparison with conventional microcloning. Nucleic Acids Res., , 17, , 9027–9037.
Lichter,P., Ledbetter,S.A., Ledbetter,D.H. and Ward,D.C. ( (1990) ) Fluorescence in situ hybridization with Alu and L1 polymerase chain reaction probes for rapid characterization of human chromosomes in hybrid cell lines. Proc. Natl Acad. Sci. USA, , 87, , 6634–6638.
Zhang,L., Cui,X., Schmitt,K., Hubert,R., Navidi,W. and Arnheim,N. ( (1992) ) Whole genome amplification from a single cell: implications for genetic analysis. Proc. Natl Acad. Sci. USA, , 89, , 5847–5851.
Walker,G.T., Fraiser,M.S., Schram,J.L., Little,M.C., Nadeau,J.G. and Malinowski,D.P. ( (1992) ) Strand displacement amplification—an isothermal, in vitro DNA amplification technique. Nucleic Acids Res., , 20, , 1691–1696.
Lizardi,P.M., Huang,X., Zhu,Z., Bray-Ward,P., Thomas,D.C. and Ward,D.C. ( (1998) ) Mutation detection and single-molecule counting using isothermal rolling-circle amplification. Nature Genet., , 19, , 225–232.
Dean,F.B., Hosono,S., Fang,L., Wu,X., Faruqi,A.F., Bray-Ward,P., Sun,Z., Zong,Q., Du,Y., Du,J. et al. ( (2002) ) Comprehensive human genome amplification using multiple displacement amplification. Proc. Natl Acad. Sci. USA, , 99, , 5261–5266.
Lage,J.M., Leamon,J.H., Pejovic,T., Hamann,S., Lacey,M., Dillon,D., Segraves,R., Vossbrinck,B., Gonzalez,A., Pinkel,D. et al. ( (2003) ) Whole genome analysis of genetic alterations in small DNA samples using hyperbranched strand displacement amplification and array-CGH. Genome Res., , 13, , 294–307.
Hosono,S., Faruqi,A.F., Dean,F.B., Du,Y., Sun,Z., Wu,X., Du,J., Kingsmore,S.F., Egholm,M. and Lasken,R.S. ( (2003) ) Unbiased whole-genome amplification directly from clinical samples. Genome Res., , 13, , 954–964.
Bown,N., Cotterill,S., Lastowska,M., O'Neill,S., Pearson,A.D., Plantaz,D., Meddeb,M., Danglot,G., Brinkschmidt,C., Christiansen,H. et al. ( (1999) ) Gain of chromosome arm 17q and adverse outcome in patients with neuroblastoma. N Engl. J. Med., , 340, , 1954–1961.
Pollack,J.R., Sorlie,T., Perou,C.M., Rees,C.A., Jeffrey,S.S., Lonning,P.E., Tibshirani,R., Botstein,D., Borresen-Dale,A.L. and Brown,P.O. ( (2002) ) Microarray analysis reveals a major direct role of DNA copy number alteration in the transcriptional program of human breast tumors. Proc. Natl Acad. Sci. USA, , 99, , 12963–12968.
Trotman,L.C., Niki,M., Dotan,Z.A., Koutcher,J.A., Cristofano,A.D., Xiao,A., Khoo,A.S., Roy-Burman,P., Greenberg,N.M., Dyke,T.V. et al. ( (2003) ) Pten dose dictates cancer progression in the prostate. PLoS Biol., , 1, , E59.
Klein,C.A., Schmidt-Kittler,O., Schardt,J.A., Pantel,K., Speicher,M.R. and Riethmuller,G. ( (1999) ) Comparative genomic hybridization, loss of heterozygosity, and DNA sequence analysis of single cells. Proc. Natl Acad. Sci. USA, , 96, , 4494–4499.
Shagin,D.A., Lukyanov,K.A., Vagner,L.L. and Matz,M.V. ( (1999) ) Regulation of average length of complex PCR product. Nucleic Acids Res., , 27, , e23.
Kent,W.J., Sugnet,C.W., Furey,T.S., Roskin,K.M., Pringle,T.H., Zahler,A.M. and Haussler,D. ( (2002) ) The human genome browser at UCSC. Genome Res., , 12, , 996–1006.
Roest Crollius,H., Jaillon,O., Dasilva,C., Ozouf-Costaz,C., Fizames,C., Fischer,C., Bouneau,L., Billault,A., Quetier,F., Saurin,W. et al. ( (2000) ) Characterization and repeat analysis of the compact genome of the freshwater pufferfish Tetraodon nigroviridis. Genome Res., , 10, , 939–949.
Jain,A.N., Tokuyasu,T.A., Snijders,A.M., Segraves,R., Albertson,D.G. and Pinkel,D. ( (2002) ) Fully automatic quantification of microarray image data. Genome Res., , 12, , 325–332.
Marchio,A., Meddeb,M., Pineau,P., Danglot,G., Tiollais,P., Bernheim,A. and Dejean,A. ( (1997) ) Recurrent chromosomal abnormalities in hepatocellular carcinoma detected by comparative genomic hybridization. Genes Chromosomes Cancer, , 18, , 59–65.
Regnier,V., Meddeb,M., Lecointre,G., Richard,F., Duverger,A., Nguyen,V.C., Dutrillaux,B., Bernheim,A. and Danglot,G. ( (1997) ) Emergence and scattering of multiple neurofibromatosis (NF1)-related sequences during hominoid evolution suggest a process of pericentromeric interchromosomal transposition. Hum. Mol. Genet., , 6, , 9–16.
Kuukasjarvi,T., Tanner,M., Pennanen,S., Karhu,R., Visakorpi,T. and Isola,J. ( (1997) ) Optimizing DOP-PCR for universal amplification of small DNA samples in comparative genomic hybridization. Genes Chromosomes Cancer, , 18, , 94–101.
Huang,Q., Schantz,S.P., Rao,P.H., Mo,J., McCormick,S.A. and Chaganti,R.S. ( (2000) ) Improving degenerate oligonucleotide primed PCR-comparative genomic hybridization for analysis of DNA copy number changes in tumors. Genes Chromosomes Cancer, , 28, , 395–403.
Voullaire,L., Wilton,L., Slater,H. and Williamson,R. ( (1999) ) Detection of aneuploidy in single cells using comparative genomic hybridization. Prenat. Diagn., , 19, , 846–851.
Daigo,Y., Chin,S.F., Gorringe,K.L., Bobrow,L.G., Ponder,B.A., Pharoah,P.D. and Caldas,C. ( (2001) ) Detection of aneuploidy in single cells using comparative genomic hybridization. Degenerate oligonucleotide primed-polymerase chain reaction-based array comparative genomic hybridization for extensive amplicon profiling of breast cancers: a new approach for the molecular analysis of paraffin-embedded cancer tissue. Am. J. Pathol., , 158, , 1623–1631.
Ren,B., Robert,F., Wyrick,J.J., Aparicio,O., Jennings,E.G., Simon,I., Zeitlinger,J., Schreiber,J., Hannett,N., Kanin,E. et al. ( (2000) ) Genome-wide location and function of DNA binding proteins. Science, , 290, , 2306–2309.
Liu,C.L., Schreiber,S.L. and Bernstein,B.E. ( (2003) ) Development and validation of a T7 based linear amplification for genomic DNA. BMC Genomics, , 4, , 19.
Tanabe,C., Aoyagi,K., Sakiyama,T., Kohno,T., Yanagitani,N., Akimoto,S., Sakamoto,M., Sakamoto,H., Yokota,J., Ohki,M. et al. ( (2003) ) Evaluation of a whole-genome amplification method based on adaptor-ligation PCR of randomly sheared genomic DNA. Genes Chromosomes Cancer, , 38, , 168–176.
Stoecklein,N.H., Erbersdobler,A., Schmidt-Kittler,O., Diebold,J., Schardt,J.A., Izbicki,J.R. and Klein,C.A. ( (2002) ) SCOMP is superior to degenerated oligonucleotide primed-polymerase chain reaction for global amplification of minute amounts of DNA from microdissected archival tissue samples. Am. J. Pathol., , 161, , 43–51.
Lander,E.S., Linton,L.M., Birren,B., Nusbaum,C., Zody,M.C., Baldwin,J., Devon,K., Dewar,K., Doyle,M., FitzHugh,W. et al. ( (2001) ) Initial sequencing and analysis of the human genome. Nature, , 409, , 860–921.(Marine Guillaud-Bataille, Alexander Vale)
* To whom correspondence should be addressed. Tel: +33 1 42 11 54 21; Fax: +33 1 41 11 52 60; Email: gdanglot@igr.fr
ABSTRACT
Comparative genomic hybridization to bacterial artificial chromosome (BAC)-arrays (array-CGH) is a highly efficient technique, allowing the simultaneous measurement of genomic DNA copy number at hundreds or thousands of loci, and the reliable detection of local one-copy-level variations. We report a genome-wide amplification method allowing the same measurement sensitivity, using 1 ng of starting genomic DNA, instead of the classical 1 μg usually necessary. Using a discrete series of DNA fragments, we defined the parameters adapted to the most faithful ligation-mediated PCR amplification and the limits of the technique. The optimized protocol allows a 3000-fold DNA amplification, retaining the quantitative characteristics of the initial genome. Validation of the amplification procedure, using DNA from 10 tumour cell lines hybridized to BAC-arrays of 1500 spots, showed almost perfectly superimposed ratios for the non-amplified and amplified DNAs. Correlation coefficients of 0.96 and 0.99 were observed for regions of low-copy-level variations and all regions, respectively (including in vivo amplified oncogenes). Finally, labelling DNA using two nucleotides bearing the same fluorophore led to a significant increase in reproducibility and to the correct detection of one-copy gain or loss in >90% of the analysed data, even for pseudotriploid tumour genomes.
INTRODUCTION
Bacterial artificial chromosome (BAC)-array comparative genomic hybridization (array-CGH) is a very powerful technique, allowing the simultaneous quantitative analysis of all regions of large genomes. This approach was initially developed to identify the genomic regions gained and lost in human cancer cells (1–3), but also gave fruitful results in the analysis of constitutional DNA of patients affected by genetic disorders (4–6), in tumour animal models (7,8), and more recently, in interspecies comparisons between human and great ape genomes (9), as well as in characterization of very large size polymorphisms among human populations (10). A 32 000 BAC-array with complete coverage of the human genome is now available for all these applications (11). The array-CGH technique, also called matrix-CGH, involves the co-hybridization of differentially labelled specimen and reference genomic DNAs to an array of large insert genomic clones. The size of these targets, spotted on glass slides, allows the fully reproducible detection of single-copy gains and losses in the analysed DNA. However, the minimal amount of specimen DNA to be hybridized is 500 ng to 3 μg, obtained from a homogeneous cell population. Such amounts are often difficult to obtain from a restricted number of cells, especially when working with samples from infants, needle aspirates from tumours or laser capture-microdissected samples from heterogeneous tissues. To overcome this problem, we developed a procedure of whole genome amplification, especially adapted to high-sensitivity BAC-array CGH.
Several genome-wide amplification protocols have already been described, such as degenerated oligonucleotide-primed–PCR , ligation-mediated PCR (14), interspersed-repetitive-sequence–PCR , primer extension pre-amplification , strand displacement amplification (17) and rolling circle amplification (18). These techniques gave valuable results when used to sequence specific genes or to characterize various alleles. However, examination of quantitative analyses revealed occurrence of bias during amplification with most of them (19–21). According to these authors, the most reliable technique would be multiple strand displacement amplification with -29 DNA polymerase or related enzymes. However, these authors still indicate that the analysed loci are represented in the amplified DNA between 0.5 and 3.0 times the copy number in the starting genomic DNA template. This variation is still much higher than what can be tolerated in BAC-array CGH, which is specifically devoted to the detection of single-copy gains and losses in a diploid background. Even after amplification, one-copy gain or loss in tumour DNA (giving ratios of 1.5 and 0.5, respectively) has to be indisputably differentiated from normal genomic status (giving a ratio of 1), since these alterations have dramatic effects on tumour evolution, and finally on patient prognosis and survival (22–24).
We wanted to use the simplest method that would permit true amplification of the whole genome without any use of random primers, these being a source of inescapable affinity diversity for their respective targets. Accordingly, ligation-mediated PCR was favoured. This approach was previously proved efficient in whole genome amplification from a single cell (25). However, stoichiometric amplification of any complex DNA sample containing multiple fragments of various sizes and sequence compositions still remains a problematic goal (26). Here, we report the diverse parameters of amplification tested, and the optimized protocol successfully established. This approach efficiently amplifies as little as 1 ng of genomic DNA, and faithfully preserves the initial ratios observed with BAC-array CGH, allowing the reliable detection of one-copy-level variations among the amplified material.
MATERIALS AND METHODS
Cell lines
Human cell lines, all derived from neuroblastoma tumours (NJB, LAN1, LAN5, IGRN91, IMR32, SKNSH, SKNAS, SKNBE-2, CLB-GE and CLB-RE) were obtained from different laboratories, and cultured in RPMI-1640 medium supplemented with 10% foetal bovine serum, 2 mM L-glutamine, 100 U/ml penicillin and 100 mg/ml streptomycin. Genomic DNA was extracted using the DNeasy tissue kit (Qiagen) according to the manufacturer's instructions.
Construction of BAC/PAC microarray
BAC and P1-derived artificial chromosome (PAC) clones were selected from Pieter De Jong libraries (http://www.chori.org/bacpac/) according to their position on UCSC human genome . Clones were grown in 96-well blocks containing 1.2 ml of 2YT medium (Sigma) supplemented with 12.5 μg/ml chloramphenicol or 25 μg/ml kanamycin, for BACs or PACs, respectively, and DNA extracted according to the protocol used at Genoscope (Evry, France) (28). The authenticity of the clones was verified by fluoresence in situ hybridization (FISH) to normal human metaphase preparations. Valid BAC/P1 DNAs were amplified by ligation-mediated PCR as published by Snijders et al. (3) with the use of Taq DNA polymerase (MBI Fermentas). PCR products were ethanol-precipitated and re-dissolved in H2O–50% dimethyl sulfoxide. DNA solutions (final concentrations 800 ng/μl) were arrayed onto amino-silane-coated glass slides (Corning GAPS II) in six replicates using a Microgrid Pro spotter (BioRobotics Ltd).
Genome-wide amplification of specimen DNA
The amplification procedure used a ligation-mediated PCR technique. Multiple parameters were varied in order to identify the conditions most suited to fully quantitative amplification of all genomic regions. The details of these tests are presented in the results. The optimal protocol finally selected was the following. Human large-size genomic DNA was digested by Mse1 (2 U; New England Biolabs) in One-Phor-All Buffer PLUS (Amersham) and, after thermal inactivation of enzyme, directly processed to the following step. DNA concentration was determined on a Nanodrop spectrophotometer (Nanodrop Technologies, Inc.) and the digested DNA was adjusted to the concentration of 1 ng/μl. Ligation was performed by mixing 1 μl of the diluted DNA with 3.5 μl H2O, 0.5 μl of 10x One-Phor-All Buffer PLUS, 0.5 μl of 100 μM primer L6 (5'-CACTAGGAGCTGGCAGATCGTACATTGA-3'; Invitrogen) and 0.5 μl of 100 μM primer C8 (5'-TATCAATGTACGA-3'). After heating for 1 min at 65°C, the temperature was lowered to 15°C, ramping at 1°C/min. A mixture containing 2 μl H2O, 1 μl of T4 DNA ligase (5 U/μl; MBI Fermentas) and 1 μl of 10 mM rATP (Promega) was then added, and the ligation was performed overnight at 15°C. The ligation products were incubated for 4 min at 65°C and stored at 37°C. For PCR amplification, a 40 μl mixture, consisting of 31.2 μl H2O, 4.5 μl Taq PCR Buffer10x (MBI Fermentas), 2 μl of dNTPs (10 mM each; Amersham), 0.5 μl of 100 μM primer L6, 0.8 μl of 100 mM MgCl2 and 1 μl of Taq DNA polymerase (5 U/μl; MBI Fermentas), was similarly pre-warmed to 37°C and added to the ligation mixture. The 50 μl reaction mixture was incubated for 3 min at 68°C before starting the PCR programme: for 14 cycles: 94°C for 50 s, 65°C for 1 min and 68°C for 3 min. Amplification products were purified with Qiaquick gel extraction kit (Qiagen) and eluted in 40 μl. DNA concentration was measured and the size of amplified material determined by agarose gel electrophoresis.
DNA labelling for array-CGH
Test and reference DNAs were labelled by random priming in 25 μl reaction volumes. Briefly, a 17.5 μl pre-mixture was prepared containing between 0.4 and 0.6 μg of genomic Mse1 digested DNA (possibly amplified), and a final concentration of 0.2 mM random decamers solution (Invitrogen) in Klenow fragment exo(–) buffer. After denaturing DNA for 5 min at 95°C, two different procedures were tested. In the first one, only one fluorescent dNTP was used. A 7.5 μl mixture was added, leading to a final labelling reaction containing 0.2 mM dATP, 0.2 mM dGTP, 0.2 mM TTP, 0.1 mM dCTP, 0.1 mM Cy3 or Cy5-dCTP (Amersham) and 12.5 U Klenow fragment (5 U/μl; MBI Fermentas). In the second one, both fluorescent dCTP and dUTP similarly labelled were added to the same reaction. The labelling mixture finally contained 0.2 mM dATP and dGTP, 0.075 mM dCTP and TTP, 0.05 mM Cy3- or Cy5-dCTP, 0.05 mM Cy3- or Cy5-dUTP and 12.5 U Klenow fragment. After labelling, unincorporated nucleotides were removed by use of the Nucleospin extract kit (Macherey-Nagel).
Array hybridization, imaging and data analysis
Labelled test and reference DNAs were mixed with 80 μg of Cot-1 DNA (Roche), 80 μg of salmon sperm DNA (Gibo-BRL), 80 μg of yeast tRNA (Sigma) and ethanol-precipitated. The pellet was dissolved in 9 μl H2O and 21 μl of 1.43x hybridization buffer, leading to a final composition of 50% formamide, 2x SSC, pH 7, 10% dextran sulphate and 4% SDS. Probes were denatured for 5 min at 73°C, and incubated at 37°C for 30 min to allow blocking of repetitive sequences. Microarray slides were briefly hydrated with steam and irradiated at 150 mJ (GS Gene Linker, Bio-Rad). Arrays were pre-hybridized for 30 min at 50°C in 30 ml of 3.5x SSC, 4% SDS and 10 mg/ml BSA (Sigma). After two rinses of 1 min in H2O and isopropanol, respectively, the slides were dried by centrifugation. Hybridization was performed under a 24 x 50 mm coverslip, in hybridization chambers (Corning) incubated for 72 h at 37°C. Slides were washed at room temperature for 2 min in 0.5x SSC 0.3% Igepal (Sigma), at 72–75°C for 2 min 30 s in the same solution, at room temperature for 10 min in 0.1x SSC and 0.1% SDS, and finally dried by centrifugation.
Arrays were scanned with an Agilent scanner. The UCSF ‘SPOT’ software (29) was used to locate spots automatically on the Cy3 and Cy5 image acquisitions and to calculate fluorescence ratios. A dedicated array-CGH Excel Macro was developed, which automatically analyses and summarizes the results as follows: (i) averages the ratios of the replicates and calculates the standard deviation; (ii) rejects individual spot data based on several criteria (including too weak fluorescent signals); (iii) adjusts Cy5/Cy3 ratios such that ratios of normal genomic regions are always equal to 1, despite variations in dye labelling efficiency; and (iv) plots data relative to the position of the clones on human genome (according to July 2003 UCSC cartography).
Conventional CGH on metaphase chromosome preparations and FISH
CGH was performed as described previously (30) with slight modifications. Briefly, 400 ng of tumour DNA and 200 ng of reference DNA (sexed normal human genomic DNA; Promega) were labelled by random priming in the presence of Alexa 488-dUTP or Alexa 594-dUTP (Molecular Probes), respectively. Probes were ethanol-precipitated with 20 μg of Cot-1 DNA (Roche), and hybridized to metaphase preparations for 3 days at 37°C. The slides were washed for 2.5 min in 0.5x SSC and 0.3% Igepal (Sigma) at 72°C, followed by 5 min in 1x PBS at room temperature. Images were acquired with a Vysis station and analysed with Quips XL imaging software.
FISH experiments were performed as described previously (31), except that probes were labelled by random priming in the presence of Alexa 488-dUTP or Alexa 594-dUTP.
RESULTS
Validation of BAC-array and array-CGH procedure
As an initial step, the BAC-array was tested for its ability to measure directly the genome copy number in normal cells and known cancer cell lines. Control co-hybridizations (n = 4) of normal human genomic male and female DNAs, alternatively labelled with Cy3 and Cy5, respectively, regularly displayed a signal-to-background ratio around 7. Only microarray elements with reference fluorescence intensity at least 3-fold over the background were taken into account. Of the 1578 array features printed on the slides, 98% passed this selection criterion. Hybridization to heterologous DNA (tomato BAC clone DNA spotted as control) gave a fluorescent signal twice that of slide background. Very homogeneous fluorescence ratios centring on 1 were observed on human array elements corresponding to autosomal regions, with all of the normalized ratios falling within the range 0.83–1.20. The standard deviation was 0.038, ranging from 0.021 to 0.052 according to the experiments. The single-copy difference tested on spots corresponding to X chromosome DNA, led to a mean male/female ratio of 0.65 instead of the theoretical value 0.5.
Tumour cell (n = 10) versus normal DNA co-hybridizations were also performed on the BAC-array, and at the same time tumour genomes were characterized using conventional CGH and FISH experiments. An example of such comparative analysis is presented in Figure 1. The results obtained from all 10 cell lines exhibited a very good correlation between the various techniques, with only 5–6% of spots giving discordant results in genomic status evaluation. Symmetrical ‘dye swap’ experiments, with tumour DNA being alternatively labelled with Cy5 or Cy3, as well as repetition of analogous experiments on array batches spotted from fully independent amplification series of BAC DNAs, similarly led to a reproducibility of 94–95% in the detection of single-copy gains and losses. However, while the reproducibility was even better with pseudodiploid tumour cell lines, the percentage of discrepancies increased when working with pseudotriploid cells. Such pathological genomes are common in tumour specimens. Careful examination of the results indicated that a small number of BAC clones, generally containing a biased base composition, were responsible for this decreased reliability. To limit the consequences of these GC% variations in human DNA regions, the labelling of test and reference DNAs was modified. Although it was initially performed in the presence of the single Cy-dCTP nucleotide, subsequent labelling reactions contained both cyanine-labelled dCTP and dUTP nucleotides (see Materials and Methods). This modification allowed a more homogeneous labelling of the various genomic regions, leading to an overall increase in ratio reliability (Table 1). All the results obtained in these control experiments indicate that the array and the procedure allow the reproducible detection of one-copy-level variations in specimen DNA, in a pseudodiploid or pseudotriploid genomic background.
Figure 1. Agreement between results from array-CGH, conventional CGH and FISH techniques, in genomic status evaluation of NJB tumour cell line. (A) Conventional CGH profile showing genomic regions gained (green) and lost (red) in the tumour DNA. (B) FISH experiment with two different 17q BAC clones hybridized to a tumour metaphase preparation. The first probe (RP11-51F16), labelled in red, contains 17q23.3 material, and indicates one-copy gain in the tumour genome. The second probe (RP11-115C3), labelled in green, is specific for the 17q24.2 region, and shows two additional copies of the region in tumour DNA. (C) Detail of an array-CGH image. Tumour DNA is labelled in red and normal DNA in green. The red spots indicate the amplification of 2p24 region coding MYCN oncogene. (D) The evaluation of DNA copy number in genomic regions of NJB cell line, according to the ratio obtained from the array-CGH experiment. Array-CGH results are fully consistent with the observations made with other techniques.
Table 1. Concordance between array-CGH and conventional CGH: effect of labelling conditions
Parameter optimization for whole-genome amplification
Human genomic DNA was restricted with Mse1, giving fragments in the range of 70–2000 bp, with the bulk of the material between 200 and 1000 bp as expected (25). As BAC DNAs spotted on the array had been amplified by a ligation-PCR procedure using the previously described adapter (3,25), a new adapter, devoid of any potential cross-hybridization was designed to amplify DNA specimens. A 28 nt primer (called L6) with a base composition of 50% GC and a complementary 13mer oligonucleotide called C8 were selected. After ligation of this L6/C8 adapter, whole genomic fragments were submitted to various conditions of amplification.
Several publications (32,33) suggest that ThermoSequenase could favourably replace Taq polymerase in DOP–PCR whole-genome amplification. These polymerases were then compared in their ability to synthesize DNA products of the same composition as the starting material, after an adapter ligation reaction. A test plasmid DNA was digested by Mse1 to obtain clear-cut class sizes of DNA fragments ranging from 250 to 800 bp, and diluted to 1 ng/μl. After the adapter ligation reaction, aliquots were amplified using either Taq polymerase or ThermoSequenase (Figure 2A). The results reproducibly showed a low efficiency of ThermoSequenase amplification in comparison with Taq polymerase, with amounts of synthesized material almost 3–4-fold lower. In addition, ThermoSequenase appeared unable to amplify DNA fragments smaller than 300 bp, even in the presence of primer concentrations reaching 16 μM. Taq polymerase, in contrast, amplified fragments of all size. The comparison of the two polymerases was repeated on genomic DNA with analogous results.
Figure 2. Optimization of several parameters of the ligation-mediated PCR amplification. The protocol detailed in Materials and Methods was applied in every case with the exception of the parameter analysed. Amplified DNAs are characterized by agarose gel electrophoresis. (A) Amplifications of a PUC18 plasmid using two different DNA polymerases and increasing concentrations of primer. Lanes 1–3: Amplifications using ThermoSequenase; lanes 4–6: amplifications using Taq polymerase; lanes 1 and 4: 4 μM primer L6; lanes 2 and 5: 8 μM primer L6; lanes 3 and 6: 16 μM primer L6; lane 7: starting material (pUC18 plasmid digested with Mse1); and lane 8: molecular weight markers (mixture of Lambda DNA/HindIII fragments, and PhiX174/HaeIII fragments). (B) Amplifications of a BAC clone (RP11-111M5) with Taq polymerase, to determine the most suitable ligation buffer and the optimal number of PCR cycles. Lanes 1–3: ligation performed in the presence of 0.5x One-Phor-All buffer; lanes 5–7: ligation performed in the presence of 1x ligation buffer; lanes 1 and 5: 14 PCR cycles; lanes 2 and 6: 17 PCR cycles; lanes 3 and 7: 20 PCR cycles; and lane 4: molecular weight markers. (C) Test of increasing concentrations of primers during the PCR step. Lane 1: 1 μM primer; lane 2: 2 μM primer; lane 3: 4 μM primer; and lane 4: 8 μM primer. (D) Size distribution of the DNA before and after amplification by the optimized protocol. Lane 2: 1 μg of BAC digested DNA; lane 3: 1 μg of PCR-amplified BAC DNA; lane 4: 400 ng of digested LAN5 cell line DNA; lane 5: 400 ng of LAN5 amplified DNA; and lanes 1 and 6: molecular weight markers.
To avoid template loss, restriction enzyme digest, adapter ligation and PCR amplification reactions were linked, without intermediate purification steps. In order to check for potential negative interference between ligation buffer and optimal PCR conditions, two ligation conditions were tested on a BAC-digested DNA (RP11-111M5), one in classical ligation buffer and the second in 0.5x One-Phor-All Buffer. Subsequent amplifications gave very analogous qualitative results, with a quantitative advantage with the latter buffer (Figure 2B, lanes 1 and 5). One-Phor-All Buffer was then used for ligation.
Comparative amplifications were performed in the presence of various primer L6 concentrations, ranging between 1 and 16 μM (Figure 2A and C). The best amplification of the various class sizes of DNA fragments was observed with primer concentrations of 1 and 2 μM. The latter concentration was finally used because it allows the synthesis of a larger quantity of DNA. Primer concentrations of 4 μM and above led to preferential amplification of short DNA fragments, to the detriment of fragments longer than 600 bp.
Genomic specimen DNA has to be amplified as much as possible. However, great care was taken to avoid potential hybridization of DNA fragments originating from different chromosomes (by means of the adapter sequence), especially during the last cycles of amplification. To analyse this particular point, the digested and ligated BAC DNA was amplified for 14, 17 and 20 cycles. As can be seen in Figure 2B, after 14 cycling amplifications, a banding profile analogous to that observed with the original material was still present. However, after 17 cycles of PCR, a smear of increasingly higher molecular weight DNA progressively replaced the characteristic DNA banding of the starting DNA, suggesting physical association with irrelevant fragments.
Characteristics of amplified DNA using the optimized protocol
Starting with 1 ng of digested genomic DNA in every case, 12 independent amplifications were performed according to the optimized protocol detailed in Materials and Methods. After purification of the amplified DNA on small columns, 2.6–3.2 μg of material was consistently obtained, with a mean of 2.9 μg for all 12 experiments. Analysis of the size of amplified material showed a very good correlation with starting material (see Figure 2D, lanes 4 and 5) with apparently similar distribution. Only very small fragments ranging from 70 to 150 bp were absent in the amplified material, but this was partly related to the fact that during the ligation step primers of 28 bp are added at both ends of the DNA fragments. Accordingly, no DNA fragments <125 bp were expected in the amplified material. However, in order to define more quantitatively the efficiency of amplification of DNA fragments of various class sizes, the optimized protocol was applied to digested BAC DNA (Figure 2D, lanes 2 and 3). The amount of each of the DNA fragments before and after amplification was assessed and the results are shown in Table 2. It can be seen that the amplification is not fully homogeneous, since the amplification ratio ranged from 0.4 to 1.3 according to the size of DNA fragments. The most efficient amplification was observed for DNA fragments between 155 and 1100 bp. Amplification efficiency declined for smaller as well as larger fragments. However, 70–75% of initial material was amplified very homogeneously, with the amplification ratio ranging from 0.9 to 1.3.
Table 2. Size influence on the efficiency of amplification of DNA fragments in a complex DNA mixture
Validation of the amplification protocol using array-CGH hybridization
An aliquot of 1 ng of the genomic DNA from the 10 cell lines used previously was amplified 3000-fold according to the optimized protocol, and the 10 non-amplified as well as the 10 corresponding amplified DNAs were hybridized to the BAC-array. Examination of the absolute fluorescence values retained on tomato BAC clones showed non-specific background levels completely identical between the two series (means of background values differed by 3% between amplified and non-amplified DNAs), demonstrating an absence in any non-specific hybridization related to primer L6.
Several authors characterizing amplified genomic DNA with conventional CGH experiments (33,34) as well as the cDNA-array CGH technique (20) reported a distortion of ratios when the amplification procedure was only applied to the test sample. We wondered whether the same phenomenon would occur with our amplification procedure. In order to verify this point, conventional CGH and array-CGH were performed, with non-amplified reference DNA co-hybridized with amplified test DNA. Conventional CGH results showed occasional deviations of the ratios at several chromosome loci in comparison with the profiles obtained from non-amplified test DNA. Moreover, array-CGH experiments on five cell lines confirmed this bias. Variable discordance rates were observed, reaching 20% in some cases, while only 5% of incorrect ratios were obtained without amplification. When test and reference DNAs were similarly amplified, the distortions disappeared and the ratios complied with those observed with non-amplified DNA, both with conventional CGH and array-CGH techniques.
The 10 cell lines were then analysed by array-CGH, using both test and reference DNAs amplified 3000-fold. The reliability of the amplification method was assessed by comparing each pair of ratios of each BAC clone from the amplified and non-amplified datasets. A linear regression performed on 2473 pairs of ratio values corresponding to genomic regions bearing low-copy-number variations led to a correlation coefficient of 0.959, which was quite similar to the correlation coefficient of 0.957 obtained from direct replicate experiments (Figure 3A). When ratios from all genomic regions were considered, including those coding for in vivo amplified oncogenes, the correlation coefficients increased further, reaching 0.990 and 0.986, respectively. The scatter plot of the ratios obtained from in vitro amplified versus non-amplified DNA (focusing on low-copy-variation regions) is presented in Figure 3B. The low scattering of the dots and the slope of 0.936, very close to 1, attest to the fidelity of the amplification method. Two examples of the accuracy of the amplification are presented in Figure 4. These diagrams show that ratios obtained with and without amplification of DNAs are almost perfectly superimposed, indicating that the amplification has retained the copy number variations present in the initial DNA specimens. To further verify this point, the genomic copy number determined by array-CGH with amplified DNA samples was exhaustively compared with the results from conventional CGH. The results, summarized in Table 3, confirm the accuracy of the optimized amplification protocol, and the reliability of the results, even in the most difficult case of pseudotriploid genomic status.
Figure 3. Comparison of array-CGH data obtained from a series of 10 neuroblastoma cell lines. (A) Correlation coefficients determined from (i) two series of non-amplified DNAs (Direct) using alternative labelling of tumour DNAs with Cy3 or Cy5 nucleotides and (ii) series of non-amplified versus amplified tumour DNAs. A total of 2473 pairs of ratios were considered, all corresponding to genomic regions of low-level variations. (B) Scatter plot of ratios obtained from array-CGH analysis of non-amplified versus amplified DNAs, compiling the results of regions of low-level variations in all neuroblastoma cell lines. The tight distribution of points along the fitted line illustrates the high fidelity of the amplification.
Figure 4. Focus on individual ratio values obtained from array-CGH analyses, with or without amplification of starting genomic DNA. (A) Fluorescence ratios (log 2) observed for the SKNAS cell line as test DNA, with BAC clones ordered from chromosome1p to chromosome Y. (B) Analogous results for IMR32 cell line. The superimposed values underline the fidelity of amplification.
Table 3. Concordance between array-CGH and conventional CGH: effect of DNA amplification
DISCUSSION
The aim of the experiments reported here was to amplify several 1000-fold, and in a nearly stoichiometric way, all regions of a large genome, in order to preserve the one-copy-level differences present in the original DNA specimen. Various amplification methods have already been proposed. However, the results of Daigo et al. (35) and Veltman et al. (5) showed a poor fidelity of DOP–PCR amplification for array-CGH measurements of genomic DNA copy numbers. Multiple displacement amplifications demonstrated their reproducibility for gene dosage alterations of 3-fold or more, but not for one-copy-level variations (19–20). Similarly, Ren et al. (36) and Liu et al. (37), as well as Tanabe et al. (38) proposed amplification methods starting from randomly shared genomic DNA, within the framework of Chromatin Immunoprecipitation (ChIP) studies. The use of ligation-mediated PCR, or of T7-based linear amplifications, appeared to give reliable results for ratios in the range of 1–25, but few if any analyses were performed for ratios close to 1. The ligation PCR-based amplification method proved to give valuable results in measurements of one-copy-level variations using conventional CGH (25,39). However, BAC-arrays are much more demanding, since conventional CGH averages the ratios obtained from 10 to 20 Mb regions on chromosome arms, whereas BAC-arrays explore regions 150 kb in size. Here we propose a protocol of whole-genome amplification based on ligation-mediated PCR, with optimized conditions specifically adapted to the sensitivity of the array-CGH technique. The principle of ligation-mediated PCR amplification offers several advantages over other techniques. First, it uses the same primer for all fragments to be amplified. This circumvents the bias amplification problems related to the variable affinity of random or degenerated primers towards their respective targets. Second, complementary sequences, present at both ends of each DNA strand, favour an internal pairing of the smallest molecules, which counterbalances their otherwise preferential duplication during the PCR process (26). Finally, cycling induces a synchronized duplication of the various genomic DNA fragments present in starting material. This is not the case with multiple displacement amplifications, where the first duplicated regions may be, in turn, rapidly used as template. The three elements favour quantitative and homogeneous whole-genome amplification, provided no limiting elements interfere with the procedure. Our results clearly confirm this observation, as the whole starting DNA fragments appeared to be amplified with similar efficiencies, the bulk of which were between 0.9 and 1.3.
Critical aspects of the protocol described herein, in relation to its use before array-CGH experiments, are the following. The adapter ligation reaction was performed at low-DNA template concentrations, using rather high adapter concentrations, to avoid any potential ligation events between genomic fragments from different chromosomes. Here, the use of 1 ng of genomic DNA template and 5 μM of the adapter led to concentration ratios in the range of 5000/1 for adapter ends and genomic fragment ends, respectively. The second crucial step of this protocol was the number of PCR cycles, in relation to the primer concentration. In order to obtain amplified DNA of the same size as starting DNA, a maximum concentration of 2 μM of primers should be used in the PCR. In these conditions, a reproducible synthesis of 3 μg of DNA was achieved from 1 ng of starting material, after a 14-cycle Taq PCR amplification. This DNA amount corresponds to the upper limit for safe amplification, because in these conditions 25% of the initial primer was incorporated in the neosynthesized DNA. When additional cycling steps were scheduled, a rapidly high proportion of these molecules hybridized between themselves, instead of hybridizing free primers. Thus, if the starting DNA has not been damaged during storage, and if no stray impurities are present in the initial sample, a better security margin would be obtained if only 13 amplification cycles were scheduled, in order to synthesize between 1 and 2 μg of amplified genomic DNA. The present results also confirm the high processivity of Taq polymerase when used in its optimal conditions, since the observed duplication factor was 0.89, in comparison with the theoretical factor of 1, if all molecules were duplicated at each cycling. Taking this doubling factor into account, the present protocol may be adapted. Cycling number might be increased when starting with <1 ng of genomic DNA, which approximately corresponds to 100 cells. Performing the ligation reaction with >1 ng could rapidly lead to an increasing number of false results, especially if the cycling number is not reduced in proportion. Ligation reactions using too much starting genomic DNA (in comparison with the adapter concentration) or too many PCR cycles would lead to unwanted physical association with genomic fragments originally far away in the genome. As a consequence of this potential problem, ratios of altered genomic regions progressively evolve towards normality.
To check this particular point, we used a series of 10 altered genomes to test the amplification procedure proposed. All the validation experiments, performed on arrays of 6 replicates of 250 BAC clones, demonstrated the faithful character of the results obtained from amplified genomic DNAs. No false hybridization occurred between the two primers used to amplify arrayed BAC DNAs and specimen DNAs, respectively. Pairs of ratios from direct and amplified genomic DNAs were very close or superimposed on diagrams corresponding to each cell line. The correlation coefficient of 0.96, determined from regions bearing small copy number variations, as well as the narrow distribution of each pair in the scatter plot of the data, strengthened the validity of the procedure. Finally, the comparison of the results obtained from array-CGH and conventional CGH, showing >90% concordance for individual BAC loci, even in a pseudotriploid context, firmly attests to the sound nature of the procedure. The observed reproducibility is even better than expected, in relation with the slight inhomogeneity of amplification of the DNA according to its size. Our results showed that DNA fragments of 70–150 bp, as well as those >1100 bp, were amplified with an efficiency about one-half of that observed for intermediate fragments. Together, these fragments represent 25–30% of the starting genomic DNA. Their decreased amplification might have generated bias in the ratios obtained from amplified DNA, but interestingly, this was not the case. Three explanations, either alone or in combination, may be proposed. The first is that an analogous amplification process is applied to the test, reference and BAC-arrayed DNAs. Thus, the same slight amplification bias on all these DNAs should neutralize each other, with no consequence for the fluorescence ratios. The second parameter deals with the fact that array-CGH is performed in the presence of high amounts of Cot1-DNA, to block repetitive sequences. The very small and too large fragments, present after Mse1 digestion of genomic DNA, could be specially enriched in repetitive sequences. If it was the case, their decreased amplification should be neutral for the accurate genomic copy number evaluation. Repetitive sequences represent 44.8% of the human genome (40). The third factor of reliability might reside in the fact that the amplified DNA comprises a mixture of fragments independently amplified. Several hundreds of them hybridize to each of the BAC clones spotted on the array. The fluorescence recorded on each spot represents the sum of these individual elements, and the small differences in the efficiency of amplification between the DNA fragments might counterbalance each other.
In conclusion, array-CGH on BAC clones is currently the best procedure for measuring precisely the genome copy number at a wide range of loci. In many circumstances, there is insufficient DNA to be characterized with this technique. Several procedures of whole-genome amplification have been developed in the last decade but none has demonstrated its reliability for subsequent one-copy-level variation analyses on BAC-arrays. Recent publications suggest that every whole-genome amplification procedure has specific advantages and pitfalls. At present, the multiple displacement amplification seems the most accurate procedure to be used before genotyping and gene sequencing. However, the optimized protocol of ligation-mediated PCR amplification presented here should be the most reliable procedure in measuring one-copy-level variations in minute amounts of genomic DNA with BAC-array CGH technology.
ACKNOWLEDGEMENTS
We are grateful to D. Fauvet, A. Pitaval, Z. Skalli, P. Coullin, B. Clausse, M.A. Dillies-Pelletier, V. Lazar, H. Ripoche and P. Dessen for helpful advice, and to V. Combaret for some tumour cell lines. Warmest thanks are due to L. Saint Ange for assistance with editing and to Prof. G. Lenoir for his support. This work was funded by grants from the CNRS, Institut Gustave Roussy, the Association pour la Recherche sur le Cancer (ARC) and Paris 11 University. M.G.-B. is supported by a doctoral fellowship from the CNRS.
REFERENCES
Solinas-Toldo,S., Lampel,S., Stilgenbauer,S., Nickolenko,J., Benner,A., Dohner,H., Cremer,T. and Lichter,P. ( (1997) ) Matrix-based comparative genomic hybridization: biochips to screen for genomic imbalances. Genes Chromosomes Cancer, , 20, , 399–407.
Pinkel,D., Segraves,R., Sudar,D., Clark,S., Poole,I., Kowbel,D., Collins,C., Kuo,W.L., Chen,C., Zhai,Y. et al. ( (1998) ) High resolution analysis of DNA copy number variation using comparative genomic hybridization to microarrays. Nature Genet., , 20, , 207–211.
Snijders,A.M., Nowak,N., Segraves,R., Blackwood,S., Brown,N., Conroy,J., Hamilton,G., Hindle,A.K., Huey,B., Kimura,K. et al. ( (2001) ) Assembly of microarrays for genome-wide measurement of DNA copy number. Nature Genet., , 29, , 263–264.
Bruder,C.E., Hirvela,C., Tapia-Paez,I., Fransson,I., Segraves,R., Hamilton,G., Zhang,X.X., Evans,D.G., Wallace,A.J., Baser,M.E. et al. ( (2001) ) High resolution deletion analysis of constitutional DNA from neurofibromatosis type 2 (NF2) patients using microarray-CGH. Hum. Mol. Genet., , 10, , 271–282.
Veltman,J.A., Schoenmakers,E.F., Eussen,B.H., Janssen,I., Merkx,G., van Cleef,B., van Ravenswaaij,C.M., Brunner,H.G., Smeets,D. and van Kessel,A.G. ( (2002) ) High-throughput analysis of subtelomeric chromosome rearrangements by use of array-based comparative genomic hybridization. Am. J. Hum. Genet., , 70, , 1269–1276.
Vissers,L.E., de Vries,B.B., Osoegawa,K., Janssen,I.M., Feuth,T., Choy,C.O., Straatman,H., van der Vliet,W., Huys,E.H., van Rijk,A. et al. ( (2003) ) Array-based comparative genomic hybridization for the genomewide detection of submicroscopic chromosomal abnormalities. Am. J. Hum. Genet., , 73, , 1261–1270.
Hackett,C.S., Hodgson,J.G., Law,M.E., Fridlyand,J., Osoegawa,K., de Jong,P.J., Nowak,N.J., Pinkel,D., Albertson,D.G., Jain,A. et al. ( (2003) ) Genome-wide array CGH analysis of murine neuroblastoma reveals distinct genomic aberrations which parallel those in human tumors. Cancer Res., , 63, , 5266–5273.
O'Hagan,R.C., Brennan,C.W., Strahs,A., Zhang,X., Kannan,K., Donovan,M., Cauwels,C., Sharpless,N.E., Wong,W.H. and Chin,L. ( (2003) ) Array comparative genome hybridization for tumor classification and gene discovery in mouse models of malignant melanoma. Cancer Res., , 63, , 5352–5356.
Locke,D.P., Segraves,R., Carbone,L., Archidiacono,N., Albertson,D.G., Pinkel,D. and Eichler,E.E. ( (2003) ) Large-scale variation among human and great ape genomes determined by array comparative genomic hybridization. Genome Res., , 13, , 347–357.
Locke,D.P., Segraves,R., Nicholls,R.D., Schwartz,S., Pinkel,D., Albertson,D.G. and Eichler,E.E. ( (2004) ) BAC microarray analysis of 15q11-q13 rearrangements and the impact of segmental duplications. J. Med. Genet., , 41, , 175–182.
Ishkanian,A.S., Malloff,C.A., Watson,S.K., DeLeeuw,R.J., Chi,B., Coe,B.P., Snijders,A., Albertson,D.G., Pinkel,D., Marra,M.A. et al. ( (2004) ) A tiling resolution DNA microarray with complete coverage of the human genome. Nature Genet., , 36, , 299–303.
Ludecke,H.J., Senger,G., Claussen,U. and Horsthemke,B. ( (1989) ) Cloning defined regions of the human genome by microdissection of banded chromosomes and enzymatic amplification. Nature, , 338, , 348–350.
Telenius,H., Carter,N.P., Bebb,C.E., Nordenskjold,M., Ponder,B.A. and Tunnacliffe,A. ( (1992) ) Degenerate oligonucleotide-primed PCR: general amplification of target DNA by a single degenerate primer. Genomics, , 13, , 718–725.
Saunders,R.D., Glover,D.M., Ashburner,M., Siden-Kiamos,I., Louis,C., Monastirioti,M., Savakis,C. and Kafatos,F. ( (1989) ) PCR amplification of DNA microdissected from a single polytene chromosome band: a comparison with conventional microcloning. Nucleic Acids Res., , 17, , 9027–9037.
Lichter,P., Ledbetter,S.A., Ledbetter,D.H. and Ward,D.C. ( (1990) ) Fluorescence in situ hybridization with Alu and L1 polymerase chain reaction probes for rapid characterization of human chromosomes in hybrid cell lines. Proc. Natl Acad. Sci. USA, , 87, , 6634–6638.
Zhang,L., Cui,X., Schmitt,K., Hubert,R., Navidi,W. and Arnheim,N. ( (1992) ) Whole genome amplification from a single cell: implications for genetic analysis. Proc. Natl Acad. Sci. USA, , 89, , 5847–5851.
Walker,G.T., Fraiser,M.S., Schram,J.L., Little,M.C., Nadeau,J.G. and Malinowski,D.P. ( (1992) ) Strand displacement amplification—an isothermal, in vitro DNA amplification technique. Nucleic Acids Res., , 20, , 1691–1696.
Lizardi,P.M., Huang,X., Zhu,Z., Bray-Ward,P., Thomas,D.C. and Ward,D.C. ( (1998) ) Mutation detection and single-molecule counting using isothermal rolling-circle amplification. Nature Genet., , 19, , 225–232.
Dean,F.B., Hosono,S., Fang,L., Wu,X., Faruqi,A.F., Bray-Ward,P., Sun,Z., Zong,Q., Du,Y., Du,J. et al. ( (2002) ) Comprehensive human genome amplification using multiple displacement amplification. Proc. Natl Acad. Sci. USA, , 99, , 5261–5266.
Lage,J.M., Leamon,J.H., Pejovic,T., Hamann,S., Lacey,M., Dillon,D., Segraves,R., Vossbrinck,B., Gonzalez,A., Pinkel,D. et al. ( (2003) ) Whole genome analysis of genetic alterations in small DNA samples using hyperbranched strand displacement amplification and array-CGH. Genome Res., , 13, , 294–307.
Hosono,S., Faruqi,A.F., Dean,F.B., Du,Y., Sun,Z., Wu,X., Du,J., Kingsmore,S.F., Egholm,M. and Lasken,R.S. ( (2003) ) Unbiased whole-genome amplification directly from clinical samples. Genome Res., , 13, , 954–964.
Bown,N., Cotterill,S., Lastowska,M., O'Neill,S., Pearson,A.D., Plantaz,D., Meddeb,M., Danglot,G., Brinkschmidt,C., Christiansen,H. et al. ( (1999) ) Gain of chromosome arm 17q and adverse outcome in patients with neuroblastoma. N Engl. J. Med., , 340, , 1954–1961.
Pollack,J.R., Sorlie,T., Perou,C.M., Rees,C.A., Jeffrey,S.S., Lonning,P.E., Tibshirani,R., Botstein,D., Borresen-Dale,A.L. and Brown,P.O. ( (2002) ) Microarray analysis reveals a major direct role of DNA copy number alteration in the transcriptional program of human breast tumors. Proc. Natl Acad. Sci. USA, , 99, , 12963–12968.
Trotman,L.C., Niki,M., Dotan,Z.A., Koutcher,J.A., Cristofano,A.D., Xiao,A., Khoo,A.S., Roy-Burman,P., Greenberg,N.M., Dyke,T.V. et al. ( (2003) ) Pten dose dictates cancer progression in the prostate. PLoS Biol., , 1, , E59.
Klein,C.A., Schmidt-Kittler,O., Schardt,J.A., Pantel,K., Speicher,M.R. and Riethmuller,G. ( (1999) ) Comparative genomic hybridization, loss of heterozygosity, and DNA sequence analysis of single cells. Proc. Natl Acad. Sci. USA, , 96, , 4494–4499.
Shagin,D.A., Lukyanov,K.A., Vagner,L.L. and Matz,M.V. ( (1999) ) Regulation of average length of complex PCR product. Nucleic Acids Res., , 27, , e23.
Kent,W.J., Sugnet,C.W., Furey,T.S., Roskin,K.M., Pringle,T.H., Zahler,A.M. and Haussler,D. ( (2002) ) The human genome browser at UCSC. Genome Res., , 12, , 996–1006.
Roest Crollius,H., Jaillon,O., Dasilva,C., Ozouf-Costaz,C., Fizames,C., Fischer,C., Bouneau,L., Billault,A., Quetier,F., Saurin,W. et al. ( (2000) ) Characterization and repeat analysis of the compact genome of the freshwater pufferfish Tetraodon nigroviridis. Genome Res., , 10, , 939–949.
Jain,A.N., Tokuyasu,T.A., Snijders,A.M., Segraves,R., Albertson,D.G. and Pinkel,D. ( (2002) ) Fully automatic quantification of microarray image data. Genome Res., , 12, , 325–332.
Marchio,A., Meddeb,M., Pineau,P., Danglot,G., Tiollais,P., Bernheim,A. and Dejean,A. ( (1997) ) Recurrent chromosomal abnormalities in hepatocellular carcinoma detected by comparative genomic hybridization. Genes Chromosomes Cancer, , 18, , 59–65.
Regnier,V., Meddeb,M., Lecointre,G., Richard,F., Duverger,A., Nguyen,V.C., Dutrillaux,B., Bernheim,A. and Danglot,G. ( (1997) ) Emergence and scattering of multiple neurofibromatosis (NF1)-related sequences during hominoid evolution suggest a process of pericentromeric interchromosomal transposition. Hum. Mol. Genet., , 6, , 9–16.
Kuukasjarvi,T., Tanner,M., Pennanen,S., Karhu,R., Visakorpi,T. and Isola,J. ( (1997) ) Optimizing DOP-PCR for universal amplification of small DNA samples in comparative genomic hybridization. Genes Chromosomes Cancer, , 18, , 94–101.
Huang,Q., Schantz,S.P., Rao,P.H., Mo,J., McCormick,S.A. and Chaganti,R.S. ( (2000) ) Improving degenerate oligonucleotide primed PCR-comparative genomic hybridization for analysis of DNA copy number changes in tumors. Genes Chromosomes Cancer, , 28, , 395–403.
Voullaire,L., Wilton,L., Slater,H. and Williamson,R. ( (1999) ) Detection of aneuploidy in single cells using comparative genomic hybridization. Prenat. Diagn., , 19, , 846–851.
Daigo,Y., Chin,S.F., Gorringe,K.L., Bobrow,L.G., Ponder,B.A., Pharoah,P.D. and Caldas,C. ( (2001) ) Detection of aneuploidy in single cells using comparative genomic hybridization. Degenerate oligonucleotide primed-polymerase chain reaction-based array comparative genomic hybridization for extensive amplicon profiling of breast cancers: a new approach for the molecular analysis of paraffin-embedded cancer tissue. Am. J. Pathol., , 158, , 1623–1631.
Ren,B., Robert,F., Wyrick,J.J., Aparicio,O., Jennings,E.G., Simon,I., Zeitlinger,J., Schreiber,J., Hannett,N., Kanin,E. et al. ( (2000) ) Genome-wide location and function of DNA binding proteins. Science, , 290, , 2306–2309.
Liu,C.L., Schreiber,S.L. and Bernstein,B.E. ( (2003) ) Development and validation of a T7 based linear amplification for genomic DNA. BMC Genomics, , 4, , 19.
Tanabe,C., Aoyagi,K., Sakiyama,T., Kohno,T., Yanagitani,N., Akimoto,S., Sakamoto,M., Sakamoto,H., Yokota,J., Ohki,M. et al. ( (2003) ) Evaluation of a whole-genome amplification method based on adaptor-ligation PCR of randomly sheared genomic DNA. Genes Chromosomes Cancer, , 38, , 168–176.
Stoecklein,N.H., Erbersdobler,A., Schmidt-Kittler,O., Diebold,J., Schardt,J.A., Izbicki,J.R. and Klein,C.A. ( (2002) ) SCOMP is superior to degenerated oligonucleotide primed-polymerase chain reaction for global amplification of minute amounts of DNA from microdissected archival tissue samples. Am. J. Pathol., , 161, , 43–51.
Lander,E.S., Linton,L.M., Birren,B., Nusbaum,C., Zody,M.C., Baldwin,J., Devon,K., Dewar,K., Doyle,M., FitzHugh,W. et al. ( (2001) ) Initial sequencing and analysis of the human genome. Nature, , 409, , 860–921.(Marine Guillaud-Bataille, Alexander Vale)