Rapid parallel mutation scanning of gene fragments using a microelectronic protein–DNA chip format

Abstract

We have developed a method for the de novo discovery of genetic variations, including single nucleotide polymorphisms and mutations, on microelectronic chip devices. The method combines the features of electronically controlled DNA hybridisation on open-format microarrays, with mutation detection by a fluorescence-labelled mismatch- binding protein. Electronic addressing of DNA strands to distinct test sites of the chip allows parallel analysis of several individuals, as demonstrated for mutations in different exons of the p53 gene. This microelectronic chip-based mutation discovery assay may substitute for time-consuming sequencing studies and will complement existing technologies in genomic research.

INTRODUCTION

After the completion of the human genome project, the major goal of pharmacogenomic research will be the continuous discovery of genetic variations within the human population and the subsequent correlation of these variations with particular phenotypes (1). This may allow the development of better diagnostic tools and safer medications and will help in the study of gene biological functions.

Large-scale sequencing projects are in progress that aim to draw a map of the most frequently occurring single nucleotide polymorphisms (SNPs), the type of mutation that is expected to contribute most to the genetic variation within the human population (2–4). Several applications, however, require a more targeted approach in which only a limited number of genes from many individuals are scanned for SNPs that occur at medium to low frequency within the human population. If, for instance, the phenotype of a group of patients suggests that a particular molecular pathway is defective, only those genes that are involved in that pathway might be scanned for potential SNPs. Several methods for the de novo identification of SNPs and other genetic variations have been developed to circumvent the time-consuming direct sequencing step (5), but none of them is compatible with a rapid screening microarray format. In particular, several protocols use the ability of mismatch repair proteins of the mutS family to bind to mutant DNA (6). In these protocols, mismatches are generated by the hybridisation of a DNA that serves as a reference sequence, with the complementary sequence from a different individual; if the two sequences differ, a mismatch is generated, which is discriminated from a perfectly matched DNA by stronger mutS binding (7). As DNA microarrays are the method of choice for the parallel analysis of multiple sequences, it was a challenge to combine this feature of mutS with DNA microarray technology for use in de novo identification of genetic variations. For rapid detection of known SNPs, an electronic microarray-based technology has already been successfully established (8). Here we present a SNP and mutation discovery protocol that advantageously combines electronically controlled DNA hybridisation with mutS protein binding, to allow scanning of DNA for genetic variations.

MATERIALS AND METHODS

Labelling of mutS with Cy5-succinimidylester

Labelling was performed for 30 min at 20°C in a 10 ml reaction containing 1 mg Escherichia coli mutS (Genecheck, Fort Collins, CO) and 12.5 µM Cy5 monofunctional dye (Amersham Pharmacia Biotech, Amersham, UK) in buffer A (10 mM HEPES pH 7.9, 50 mM KCl, 5 mM MgCl₂, 1 mM DTT, 0.2 mM phenylmethylsulphonyl fluoride, 10% v/v glycerol). The Cy5mutS protein was subsequently dialysed against buffer A and stored at –80°C.

Surface plasmon resonance (SPR) analysis of Cy5mutS

The experimental set-up was as described (9). The 5′-biotinylated oligonucleotide ‘sense’ (5′-AAGCATACGGAAGTTAAAGTGCGGATCATCTCTAGCCA-3′) was loaded onto the chip surface first. Complementary oligonucleotides were subsequently loaded. Their names indicate the type of mismatch that they generate in combination with the oligonucleotide sense. Complementary oligonucleotides were: A/T (wild-type), 5′-TGGCTAGAGATGATCCGCACTTTAACTTCCGTATGC-3′; A/A, 5′-TGGCTAGAGATGATCCGCACATTAACTTCCGTATGC-3′; A/C, 5′-TGGCTAGAGATG ATCCGCACCTTAACTTCCGTATGC-3′; A/G, 5′-TGGCTAGAGATGATCCGCAATTTAACTTCCGTATGC-3′; C/C, 5′-TGGCTAGAGATGATCCCCACTTTAACTTCCGTATGC-3′; C/T, 5′-TGGCTAGAGATGATCCGCCCTTTAACTTCCGTATGC-3′; G/G, 5′-TGGCTAGAGATGATC CGCAGTTTAACTTCCGTATGC-3′; G/T, 5′-TGGCTAGAGATGATCCGCGCTTTAACTTCCGTATGC-3′; T/T, 5′-TGGCTAGAGATGATCCGCTCTTTAACTTCCGTATGC-3′.

After heteroduplex formation, 0.5 µg Cy5mutS was resuspended in buffer A containing 0.01% Tween-20 and applied to the chip surface. Changes in the relative resonance units were measured after a washing step with 200 µl of buffer A containing 0.01% Tween-20.

PCR amplification

PCR amplification of individual p53 exons was performed in 100-µl reactions containing 5 U Pfu Hotstart DNA polymerase (Stratagene, La Jolla, CA), 0.8 µl dNTPs (25 mM each), 10 µl 10× Pfu buffer, 200 ng genomic DNA and 0.5 µl each primer (100 µM). Primer pairs were 5′-CCTTACTGCCTCTTGCTT-3′ and 5′-TGAATCTGAGGCATAACT-3′ for exon 8 and 5′-TGCAGTTATGCCTCAGATTC-3′ and 5′-TCCACTTGATAAGAGGTC-3′ for exon 9. After an initial denaturation at 95°C for 3 min, we performed 35 amplification cycles, each consisting of 30 s at 95°C, 30 s at 60°C and 1 min at 72°C.

Template genomic DNA from MCF-7, MOLT-4, 293 and SW-480 human cell lines was obtained from the DSMZ (Braunschweig, Germany). For the amplification of test DNA (the DNA suspected of containing a mutation), each reaction contained an unconjugated primer and a Cy3-labelled primer, yielding test PCR products that were Cy3 labelled either on the coding or non-coding strands, respectively. The reference DNA was amplified using the same primer combinations with a biotin label instead of Cy3, and by using MCF-7 genomic DNA as template. This cell line encodes a wild-type form of p53 (10). PCR products were purified and desalted using a kit (Qiagen, Hilden, Germany).

Genetic variation discovery on DNA microarrays

NanoChip^™ Cartridges harbouring microelectronic 100 test site arrays obtained from Nanogen Europe BV (Helmond, The Netherlands) were processed on the Molecular Biology Workstation from the same manufacturer. The principles of this technology have been described previously (8,11). Oligonucleotides were identical to those analysed by SPR experiments, with the exception that the complementary oligonucleotides were labelled with Cy3. All oligonucleotides were diluted to 100 nM in 50 mM histidine. Heteroduplex DNA was generated by addressing the biotinylated oligonucleotide sense for 60 s at 2.1 V, and the complementary strands for 120 s at 2.1 V, to distinct test sites of the array.

For the analysis of PCR products, biotinylated single-stranded reference DNA was isolated using streptavidin-coated Dynabeads M-280 (Dynal, Hamburg, Germany) as recommended by the manufacturer and purified using a kit (Qiagen). Reference DNA and test PCR product were both supplemented with 50 mM histidine, denatured for 5 min at 95°C and subsequently addressed for 60 s (reference DNA) or 180 s (test DNA) at 2.1 V.

After electronic hybridisation, the arrays were first incubated with 10 U mung bean nuclease (New England Biolabs, Frankfurt, Germany) for 45 min at 30°C and then for 1 h at room temperature with 20 mM Tris–HCl pH 7.6, 50 mM KCl, 5 mM MgCl₂, 0.01% Tween-20 and 3% (w/v) bovine serum albumin (BSA) (Serva, Heidelberg, Germany). An aliquot of 500 ng Cy5mutS was diluted in 100 µl of the same buffer containing 1% BSA and incubated with the array for 1 h at room temperature. Before scanning for Cy3 and Cy5 fluorescence, the chip surface was washed with 15 ml of 20 mM Tris–HCl pH 7.6, 50 mM KCl, 5 mM MgCl₂, 0.1% Tween-20.

RESULTS

Assay overview

Microelectronic arrays allow a site-specific and individualised hybridisation of DNA strands (8,11). Hence, they are ideal tools for the generation of heteroduplex DNA by first addressing a biotinylated reference DNA with a known sequence to an individual test site, and subsequently addressing a Cy3-labelled, complementary test DNA from, for example, a patient sample to the same pad (Fig. 1). If the test DNA contains a mutation when compared to the reference DNA, the heteroduplex will contain a mismatch. This mismatch should be detectable by a fluorescence-labelled mutS protein that binds to DNA containing a mismatch at any position of the DNA sequence (Fig. 1). Since long DNA fragments can be analysed simultaneously, this assay resembles high throughput sequencing, yet substantially reduces the bioinformatics. In addition to the parallel analysis of multiple genes, one major benefit of this active chip technology is that several individuals can be analysed on the same chip, which is not possible on passive microarrays.

Figure 1.

Schematic representation of the mutS chip assay. (I) Biotinylated reference strands (e.g. PCR products) are first addressed to individual test sites of the array using electronic biassing. (II) Cy3-labelled complementary test strands are ‘electronically’ hybridised to the reference strands, thereby generating heteroduplex DNA. (III) The Cy5mutS protein binds preferentially to mismatched heteroduplex DNA. Hybridisation and binding events are monitored by fluorescence scanning of the array.

SNP detection on electronic microchips

In order to generate a functionally active Cy5-labelled mutS protein, we tested several reaction conditions using E.coli mutS and the Cy5 active ester, and obtained a Cy5mutS protein with mismatch-binding activity, as assessed by band shift experiments (data not shown).

Many genetic variations result from C→T mutation, and the corresponding G:T mismatch which is formed by hybridisation of the mutated DNA strand with the complementary wild-type strand is a preferred binding motif for mutS (7). To test if this mutation is detectable by the assay, heteroduplex DNA was generated on a microarray by first loading the biotinylated reference DNA oligonucleotide sense to all test sites of the array. Subsequently, the perfectly matching complementary test strand or a complementary test strand containing a C→T mutation was hybridised to individual test sites of the array (Fig. 2A). After incubation with Cy5mutS and a washing step, the array was scanned for Cy3 (DNA) and Cy5 (mutS) fluorescence. Uniform Cy3 fluorescence indicates even hybridisation of the different test strands (Fig. 2B). In contrast, the Cy5 fluorescence was mainly localised to the test sites harbouring the mutation (Fig. 2C), demonstrating that Cy5mutS preferentially binds to the mutant DNA.

Figure 2.

SNP detection on electronically active DNA microarrays. (A) Chip configuration. A 100 position chip was loaded with the oligonucleotide ‘sense’. The complementary oligonucleotides G/T and A/T were then addressed to the indicated positions, generating matched (A:T) or mismatched (G:T) DNA. (B) Cy3 (green, DNA) and (C) Cy5 (red, protein) fluorescence image of the microarray after DNA addressing and mutS incubation.

To check which mutations can be detected by Cy5mutS, the ability of the protein to bind to all possible mismatches was assessed by SPR analysis (Fig. 3A and B). The G:T, G:G, A:A, C:A and C:T mismatches that represent C→T, C→G, T→A, G→A and G→T mutations showed good binding signals for the protein, compared to the perfectly matched (A:T) DNA. Although the A:G, C:C and T:T mismatches that represent T→G, G→C and A→T mutations were poorly bound by the protein, their complementary mismatch sequences C:T, G:G and A:A showed good signals (Fig. 3A). Therefore, all base substitution mutations can be detected if both complementary strands are analysed by the assay (Fig. 3C). In addition to its practical value, this feature of mutS may reflect a biological significance.

Figure 3.

Binding specificity of Cy5mutS. (A) SPR analysis. Heteroduplexes containing the indicated mismatches or the complete match were loaded onto a Biacore SA chip and incubated with Cy5mutS. The values represent the amount of mutS protein bound to the heteroduplexes as resonance units after a washing step. Each value represents the average of three different experiments. A perfectly matched homoduplex DNA (A:T) served as a control. (B) A typical SPR sensogram showing Cy5mutS binding to DNA containing a G:T (red), a G:G (green) or a C:C (orange) mismatch or to perfectly matched control DNA (blue). (C) All eight possible mismatches can be called (red), as each mismatch can be transformed into its complementary equivalent by simply analysing the complementary strand by the assay.

We next tested if Cy5mutS can bind to different mismatched duplexes that were electrically hybridised on the microarray surface. After incubation with Cy5mutS, the chip was scanned for Cy3 and Cy5 fluorescence (Fig. 4A). The Cy5 fluorescence representing mutS binding to the individual mutations (Fig. 4B) strongly resembled the data obtained by SPR analysis (Fig. 3A). Hence all eight different mismatches can be detected using the rules shown in Figure 3C. The SPR and electronic microarray experiments yielded quantitatively different results for the A:A and G:G mismatches (Figs 3A and 4B). However, the different binding conditions for mutS binding to the different ligand surfaces lead to similar, albeit in some cases different, data, making it difficult to compare the two sets of data directly.

Figure 4.

Detection of different mismatches on chip surfaces. (A) The same heteroduplexes as analysed in the experiment shown in Figure 3A were generated on a chip surface by electronic loading of the oligonucleotide sense, followed by row-by-row hybridisation of the complementary Cy3-labelled strands that generate the indicated mismatches. Lane 1, A:A; lane 2, A:C; lane 3, A:G; lane 4, A:T; lane 5, C:C; lane 6, C:T; lane 7, G:G; lane 8, G:T; lane 9, T:T; lane 10, sense only. After incubation with Cy5mutS and washing, Cy3 (DNA, left) and Cy5 (protein, right) fluorescence values were determined. A representative chip image is shown. (B) Mean Cy5 fluorescence on pads encoding double-stranded oligonucleotides encoding the indicated mismatches. A perfectly matched homoduplex DNA (A:T) served as the control. The values represent the average of at least six different individual values.

p53 multi-sample mutation detection

One key feature of our assay is the ability to scan sequences from several individuals in parallel. To demonstrate this, exons 8 and 9 of the p53 gene from several human cancer cell lines were individually amplified by PCR yielding amplicons of 237 and 158 bp in size, respectively, and analysed for mutations.

The reference DNA for each exon was obtained by PCR amplification using genomic DNA from the p53 wild-type cell line MCF-7 (10) as template. Two reactions were performed for each exon to label either the coding or the non-coding strand with biotin. The biotinylated strands were then further purified to yield single-stranded DNA, and immobilised in rows to the array. The complementary Cy3-labelled test strands from the cell lines MOLT-4, SW-480 and 293 were generated by PCR and electronically hybridised to their corresponding reference DNA. Prior to the incubation with Cy5mutS, unhybridised single-stranded DNA was destroyed by on-chip digestion with mung bean nuclease to reduce non-specific Cy5mutS binding. After incubation with Cy5mutS and a washing step, the DNA was checked for even hybridisation (Cy3 fluorescence, Fig. 5A) and mutations were detected by Cy5 fluorescence scanning (Fig. 5B). We could detect mutations in exons 8 and 9 of SW-480 cells, but none in the other cell lines tested. Direct sequencing verified the p53 status of these cell lines (not shown), which was recently published in part (12,13), confirming the data obtained by the mutS chip experiment.

Figure 5.

Parallel mutation scanning of p53 exons 8 and 9 from human cell lines. Biotinylated sense and antisense strands of the exons were generated by PCR using a wild-type p53 gene as template, and addressed row-by-row as follows. Lane 1, exon 8 antisense; lane 2, exon 8 sense; lane 3, exon 9 antisense; lane 4, exon 9 sense. Subsequently, the complementary PCR amplified strands from the indicated cell lines were addressed to generate heteroduplex DNA. (A) The Cy3 image of the array indicates even hybridisation. (B) Staining of the array with Cy5mutS revealed a mutation in both exons 8 and 9 of the cell line SW-480.

DISCUSSION

The mutS chip assay allows a robust detection of genetic variations, demonstrated by using model oligonucleotides and exons of the p53 gene from several human cell lines. As sequencing of PCR products for mutation detection is time-consuming and costly, the mutS chip may be used to circumvent sequencing in many screening cases. Potential fields for the use of the assay range from pharmacogenomic patient screening to drug target validation studies and cancer diagnostics. Future individualised molecular diagnostics needs such assays valid for any haplotype variation. Other microarray-based mutation detection technologies have been developed that allow either targeted detection of a specific number of known base substitutions from many patients (8,11) or the analysis of a gene from a single patient (14). The mutS chip complements these technologies, as a SNP first discovered by the mutS chip will lead to the design of a targeted SNP assay.

A prerequisite of the assay is the individualised hybridisation event on each test site, provided through the microelectronic DNA chip platform. Apparently, the salt-free hybridisation conditions used are also beneficial for the accuracy of heteroduplex generation, thereby reducing background Cy5mutS staining. Moreover, hybridisation is achieved faster than with passive arrays, considerably shortening assay duration. We were able to detect a single point mutation in amplicons as long as 700 bp by this method (data not shown).

Our new technology complements the existing mutation reporting (8,11) and gene expression (15) protocols on electronic DNA microarrays, providing a single experimental platform for gene-based testing.