High-throughput genotyping systems promise to be an efficient means of identifying susceptibility genes involved in the etiology of non-Mendelian disorders. Adequate amounts of high-quality DNA are essential, however, for large-scale genotyping studies (1). The supply of genomic DNA is frequently limited, and the quality of DNA obtained from oral buccal swabs or Guthrie cards has not been thoroughly evaluated for high-throughput single-nucleotide polymorphism (SNP) genotyping (2)(3). Whole-genome amplification (WGA) technologies offer the opportunity to expand DNA from depleted biological samples. The first generation of WGA strategies (i.e., PCR-based methods) (4)(5), however, was limited by substantial amplification bias and incomplete coverage of genetic markers (6)(7). Recently, new strategies for WGA, such as multiple displacement amplification (MDA) or OmniPlex® WGA technology (Rubicon Genomics) have been developed. MDA is an isothermal amplification with the bacteriophage φ29 DNA polymerase (6)(8), whereas OmniPlex uses in vitro libraries with fragmented DNA (∼1.5 kb) to amplify the entire genome by PCR (9).
To apply WGA technology to BeadArray™ genotyping (Illumina), the utility of MDA and/or OmniPlex on DNA samples derived from lymphoblast cells has been evaluated (9)(10). In this study, we determined the genotyping success rate and reliability of 2 MDA variants (8)(11) and OmniPlex with and without 7-deaza-dGTP, using buccal swabs, whole blood, dried blood spots, and sheared genomic DNA on 1260- and 1228-SNP BeadArray panels. The 7-deaza-dGTP nucleotide analog was included in an attempt to ensure amplification of GC-rich DNA.
After Institutional Review Board approval and informed consent were obtained, DNA samples were collected from participants in a study of oral clefts (12). Genomic DNA samples were prepared from peripheral blood by protein precipitation (13). Buccal cells were obtained by rubbing the inside of cheeks with a brush (Medical Packaging), and DNA was extracted with 600 μL of 50 mmol/L NaOH (2). For blood spot samples dried on Guthrie cards, two to three 2-mm circles punched from the blood spots with a micro-punch (Fitzco) were placed in 5 g/100 mL Chelex® 100 chelating resin (Bio-Rad Laboratories) (14) and treated as described (blood spot A) (15) or stored at −20 °C for ∼3 years (blood spot C). As an alternative method, 2 to 3 circles punched from the blood spots were placed in a cell lysis solution (10 μL of 0.2 mol/L KOH at 65 °C for 10 min), neutralized with 10 μL of 0.2 mol/L tricine, and diluted with 80 μL of sterile H2O (blood spot O). The paper punch was cleaned with compressed air to avoid cross-contamination. To evaluate the effect of the molecular weight of the template DNA, genomic DNA was randomly sheared with DNase I into fragments of ∼20, 5, and 2 kb (16). The concentrations of DNA samples were determined by PicoGreen® dsDNA Quantitation (Molecular Probes).
The minimum amount of template DNA for MDA was assessed over a 105 range of genomic DNA (3 pg to 300 ng). Whether MDA products were human DNA or artifacts of WGA was determined by PCR with primers specific to a human DNA sequence (LMX1B exon 3) (17) and a short tandem repeat marker (AFM143×d12) and separation by 2% agarose gel electrophoresis. DNA samples (∼1 ng) and both positive and negative controls were amplified by use of a GenomiPhi™ or TempliPhi™ Kit according to the manufacturer’s instructions (Amersham Biosciences). A 50-ng aliquot of DNA at 1–2 ng/μL for each sample, identical to that used for MDA, was sent to Rubicon Genomics for amplification with and without 7-deaza-dGTP.
A panel of 1260 SNP markers for testing WGA methods (plate 1) was selected from a subset (GS0005002-OPA) of the Linkage III SNP panel generated by Illumina, and a more reliable 1228-SNP panel was selected for testing sources of DNA and sheared DNA (plate 2). Plate 1 consisted of 18 unamplified genomic DNA reference samples and the same samples amplified by each of 4 WGA methods. Plate 2 consisted of 38 unamplified genomic DNA samples and 51 samples amplified by GenomiPhi. The 2 sets of samples, including duplicates, were genotyped at the Johns Hopkins SNP Genotyping Core by the BeadArray method (Sentrix® Array Matrix) (1).
The success rate was computed as (1 − proportion of missing genotypes) ×100 (%), where the proportion of missing genotypes was calculated as the number of genotypes with a Gene Call score <0.25 over the total number of genotypes. After SNPs with a Gene Call score <0.25 were dropped, reliability was estimated with the proportion of mismatches in genotypes between unamplified and amplified genomic DNA, with and without consideration of allele loss. The distributions of SNP genotypes were analyzed by different data quality thresholds: those containing all data, a moderately conservative group (median Gene Call score >0.4), and a very conservative group (>0.5). The significance of differences between the 2 compared groups was tested by a paired t-test or 2-sided t-test (18). Statistical analyses were performed by Intercooled STATA 7.0 (STATA Co.).
Input of genomic DNA >0.03 ng yielded sufficient and reliable amplified product for a run of BeadArray genotyping requiring 3 μg by either MDA method. Mean (SE) yields were 377 (11), 217 (10), 175 (17), and 125 (12) ng/μL from each method for GenomiPhi, TempliPhi, OmniPlex without 7-deaza-dGTP, and OmniPlex with 7-deaza-dGTP, respectively. With DNA template from sources other than whole blood, however, more input DNA (>0.5 ng) was required.
As shown in Table 1⇓ , among 22 680 genotypes, 98.9 (1.4)% were successfully obtained from unamplified DNA, whereas the success rate decreased among amplified DNA [mean (SE), 90 (4.1)% for the GenomiPhi and 64.1 (8.3)% for the OminPlex with 7-deaza-dGTP]. Samples amplified by MDA showed less variability and higher reliability [0.21 (0.2)% and 0.29 (0.2)%, respectively] than those amplified by OmniPlex [0.54 (0.4)% and 1.33 (0.7)%, respectively] among 12 442 SNPs successfully typed by all 4 methods. GenomiPhi and TempliPhi were not significantly different, whereas other pairs showed significant differences (P <0.01). The percentage of mismatches for which an AB genotype was called either AA or BB (i.e., indicative of allele loss) was uniformly low for 3 of the groups (0.13%–0.16%), with the exception of the OminPlex with 7-deaza-dGTP (0.38%) method (Fig. 1⇓ ).
Regarding sources of DNA, the success rate of genotyping was 99.98% for unamplified genomic DNA, and the rates decreased with amplified DNA, particularly in DNA amplified from blood spots (99.5%–31.2%; Table 1⇓ ). DNA amplified from buccal swabs (0.35% without 2 outliers) gave low reliability similar to DNA amplified from whole blood (0.07%). DNA amplified from blood spots, however, gave less reliability (14.5%–44.9%), more allele loss (11.8%–41.3%), and more variability across individuals, for all extraction methods (Table 1⇓ ).
The size of DNA fragments was >23 kb in most genomic DNA samples, but it was ∼20 kb in buccal swabs and ranged from 2 to 10 kb in blood-spot samples. Two buccal swab samples yielding low reliability contained predominantly low–molecular-weight DNA (2–20 kb). Both success rates and reliability of unamplified DNA (99.98% success rate) and unfragmented DNA amplified from whole blood (99.3% success rate and 0% mismatches) were high among 1228 SNPs. These rates decreased, however, for sheared DNA samples (98.5%, 94.4%, and 76.4% success rate; 2.4%, 8.2%, and 28.9% mismatches for 20, 5, and 2 kb, respectively). Increasing the threshold of data quality (median Gene Call score >0.5) had a minimal effect on the success rate of genotyping, but it reduced the number of genotypes obtained from blood spots (see Table 1⇓ in the Data Supplement that accompanies the online version of this Technical Brief at http://www.clinchem.org/content/vol51/issue8/) and sheared DNA samples without any increase in reliability (data not shown).
Previous studies have reported high genotype concordance rates (>99.8%) with the use of MDA or OmniPlex with genomic DNA (9)(10)(19)(20)(21) and have suggested that MDA carried out directly from crude biological samples can be used immediately for subsequent assays with no need to measure DNA concentration or purification (6)(22). In our study, however, both success rate and reliability were highest for DNA amplified by GenomiPhi and lowest for DNA amplified by OmniPlex with 7-deaza-dGTP. Amplification of DNA from buccal swabs was less successful and less reliable than DNA from whole blood but superior to DNA from dried blood spots, despite the use of 3 independent extraction methods. The reduced success rate and reliability may result from its low molecular weight because sheared genomic DNA (<20 kb) also gave a low success rate and reliability.
The successful amplification of human DNA requires a small amount of genomic template of appropriate quality of input DNA (e.g., 0.5 ng of unfragmented DNA). Template DNA for MDA, therefore, should be shown to be >20 kb by visualization of DNA on an agarose gel, and whether DNA amplified by MDA is human or bacterial DNA must be determined by amplification of human DNA sequences by PCR before subsequent analyses, particularly when there is very little DNA. In most cases, DNA from other sources is not as good a template for WGA as whole blood and also cannot be used without amplification for BeadArray assay. In future experiments, we will consider the cut point of 0.25 for Gene Call scores adequate for BeadArray genotyping.
In conclusion, our study suggests that MDA represents an efficient and reliable method to maximize DNA resources from whole blood and buccal swabs for BeadArray SNP genotyping. DNA samples that are isolated from dried blood spots or are degraded and fragmented, however, are inappropriate for MDA.
This research was supported by Grants P60-DE13078 and R01-DE014581 from the National Institute of Dental & Craniofacial Research. We are grateful to Dr. M.D. Falin for helpful comments.
- © 2005 The American Association for Clinical Chemistry