QUANTITATION OF HETEROPLASMIC MITOCHONDRIAL MUTATIONS 193Three microliters <strong>of</strong> Streptavidin Sepharose HP (Amersham,Piscataway, NJ) was added to 37 l binding buffer (10 mMTris-HCl pH 7.6, 2 M NaCl, 1 mM ethylenediaminetetraaceticacid [EDTA], 0.1% Tween 20) <strong>and</strong> mixed with 20 L PCRproduct <strong>and</strong> 20 l high purity water for 10 min at room temperatureusing a Variomag Monoshaker (Camlab, Cambridge,UK). The beads containing the immobilized templates werecaptured onto the filter probes after applying the vacuum <strong>and</strong>then washed with 70% ethanol for 5 sec, denaturation solution(0.2 M NaOH) for 5 seconds <strong>and</strong> washing buffer (10 mM Tris-Acetate pH 7.6) for 5 seconds. The vacuum was then released<strong>and</strong> the beads released into a PSQ 96 Plate Low containing 45l annealing buffer (20 mM Tris-Acetate, 2 mM MgAc 2 pH7.6), 0.3 M sequencing primer. For the T8993C/G assay, wefound it necessary to add 1 l single str<strong>and</strong>ed binding protein(Promega, 2.2 g/l) to eliminate secondary structure in thetemplate DNA. The samples were heated to 80°C for 2 min <strong>and</strong>then allowed to cool to room temperature.Pyrosequencing reactions <strong>and</strong> data analysis. Pyrosequencingreactions were performed according to the manufacturer’sinstructions using the PSQ 96 SNP Reagent Kit, which containedthe enzyme <strong>and</strong> substrate mixture <strong>and</strong> nucleotides. Assayswere performed using the nucleotide dispensation ordersshown in Table 1. The sample genotype <strong>and</strong> percent heteroplasmywere determined using the allele frequency quantification(AQ) function in the SNP S<strong>of</strong>tware (Biotage AB). Sampleswere considered to have the mutation if the value <strong>of</strong> percentheteroplasmy was greater than three st<strong>and</strong>ard deviations fromthe mean value obtained from the normal replicates.Threshold detection <strong>of</strong> A3243G mutationDNA samples with known levels <strong>of</strong> heteroplasmy were preparedto determine the lowest level <strong>of</strong> the A3243G mutationthat could be detected reliably. In 2004, Urata <strong>and</strong> colleaguesanalyzed 40 healthy donors <strong>and</strong> MELAS patients using peptidenucleic acid probe-directed PCR clamping <strong>and</strong> demonstratedthat a mutational load as low as 0.1% can be clinically significant.Therefore, the detection <strong>of</strong> low frequency heteroplasmyfor this mutation is particularly important. DNA from a MELASpatient was amplified using the primers 5 tgcagccgctattaaaggtt3 <strong>and</strong> 5 ggttcggttggtctctgcta 3 (amplified region 3014–3894). The resulting 880-bp amplicon was cloned into the vectorpCR2.1 (Invitrogen, Carlsbad, CA) <strong>and</strong> several colonieswere sequenced to identify 2 clones; one with a wild-type genotype<strong>and</strong> the other with the MELAS mutation. DNA from eachclone was quantified <strong>and</strong> diluted to a final concentration equivalentto 2 10 4 mitochondrial genomes per microliter (dilutedin 10 ng salmon sperm DNA). The wild-type <strong>and</strong> mutated DNAwere mixed to generate samples with the A3243G mutationpresent at levels ranging from 1%–100%. Each sample was analyzedin triplicate using the Pyrosequencing assay, fluorescent<strong>and</strong> nonfluorescent PCR-RFLP (using identical primersets), <strong>and</strong> the detection threshold for each technique was determined.RESULTSSensitivity, specificity, <strong>and</strong> reproducibilityTo determine the background levels for each Pyrosequencingassay, 15 normal DNA samples were analyzed in six independentexperiments. The mean <strong>and</strong> st<strong>and</strong>ard deviation valuesfor intra <strong>and</strong> inter assay analysis are shown in Table 2. Nodifferences were observed between the interassay <strong>and</strong> intra-assayanalysis, which suggests that the background variation seenin these assays is consistent <strong>and</strong> that the assays are robust. Figure1 shows the genotype <strong>and</strong> percent heteroplasmy obtainedfrom the 50 patient DNA samples using the Pyrosequencing assayscompared to the results obtained using the routine PCR-RFLP method for all mutated samples. Error bars for the Pyrosequencingdata indicate the st<strong>and</strong>ard deviation for triplicateanalysis. Samples were considered to harbor the mutation if theAQ value obtained was greater than three st<strong>and</strong>ard deviationsTABLE 2. INTERASSAY AND INTRA-ASSAY ANALYSIS OF FIFTEEN NORMAL SAMPLES TESTED IN SIX INDEPENDENT EXPERIMENTSTO DETERMINE LEVEL OF BACKGROUND DETECTION OF MUTANT ALLELE FOR EACH ASSAYInterassay analysis (n 15) Intra-assay analysis (n 6)St<strong>and</strong>ardSt<strong>and</strong>ardMutation Mean deviation Mean deviation Mean 3SDA3460G 2.86 1.62 2.88 1.62 7.72LHONG11778A 0.05 0.14 0.06 0.22 0.72LHONT14484C 2.13 0.63 2.13 0.98 5.07LHONA3243G 0.06 0.07 0.06 0.11 0.39MELASA8344G 0.13 0.32 0.13 0.50 1.63MERRFT8993G/C 0.01 0.02 0.01 0.03 0.04NARP/LeighsSD, st<strong>and</strong>ard deviation.
194WHITE ET AL.FIG. 1. Percentage heteroplasmy detected by the Pyrosequencing assays compared to the diagnostic result obtained using nonfluorescentpolymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP). A: G3460A. B: G11778A. C:T14484C <strong>and</strong> fluorescent PCR-RFLP. D: A3243G. E: T8993G/C. F: A8344G. The x axis shows the sample number <strong>and</strong> the yaxis shows the percentage heteroplasmy detected. In graphs (a) to (c) the 100% heteroplasmy obtained for PCR-RFLP indicatesthat no wild-type product was viewed on a 3% agarose gel after restriction enzyme digestion.from the mean value obtained from the normal replicated samples(Table 2). The genotypes obtained using the Pyrosequencingassays were 100% concordant with those obtained usingPCR-RFLP <strong>and</strong> the levels <strong>of</strong> heteroplasmy detected usingboth techniques were essentially identical. Of the 50 patientDNA samples 13 were found to have 1 <strong>of</strong> the 3 LHON mutations,10 had the A3243G MELAS mutation, 4 were positivefor the MERRF A8344G mutation, 4 carried the NARP/LeighsT8993G/C mutation, <strong>and</strong> no mutation was detected in 19 samples.None <strong>of</strong> the samples had the G8994A polymorphism. ThePyrosequencing assays were 100% sensitive <strong>and</strong> 100% specifictaking into account the respective background values. The determination<strong>of</strong> the level <strong>of</strong> heteroplasmy using the Pyrosequencingassays was highly reproducible. For the mutated samplesanalyzed in triplicate, no deviation from the mean wasobserved for the 3 LHON assays <strong>and</strong> the coefficient <strong>of</strong> variancefor the 3243, 8344, <strong>and</strong> 8993 assays was 1.2–52.7,1.65–5.47, <strong>and</strong> 0.6–3.5 respectively. Representative pyrogramsobtained for a mutated sample from each assay are shown inFigure 2.Threshold detection <strong>of</strong> A3243G mutationThe lowest level <strong>of</strong> detection <strong>of</strong> the A3243 mutation for thePyrosequencing, nonfluorescent PCR-RFLP <strong>and</strong> fluorescentPCR-RFLP assay was determined by generating a st<strong>and</strong>ardcurve from cloned wild-type <strong>and</strong> mutated DNA samples thathad been mixed to generate samples with the A3243G mutationpresent at levels <strong>of</strong> 0%, 1%, 2.5%, 5%, 10%, 20%, 30%,40%, 50%, 60%, 70%, 80%, 90%, <strong>and</strong> 100%. Figure 3 showsthe st<strong>and</strong>ard curves for the Pyrosequencing <strong>and</strong> fluorescentPCR-RFLP assays for identical samples analyzed in triplicate.A level <strong>of</strong> 1% heteroplasmy was reliably detected with a meanAQ value <strong>of</strong> 1.73 (st<strong>and</strong>ard deviation 0.84). Results from thefluorescent PCR-RFLP show that 5% heteroplasmy is the limit<strong>of</strong> detection for this technique. In contrast to the expected linearrelationship between observed <strong>and</strong> expected heteroplasmylevels for the Pyrosequencing assay an apparent quadratic relationshipwas seen for PCR-RFLP. This may reflect the underrepresentation<strong>of</strong> heteroplasmy detected because <strong>of</strong> the formation<strong>of</strong> heteroduplexes that cannot be cut by the restrictionenzyme. Analysis <strong>of</strong> the same PCR products by agarose gelelectrophoresis detected the mutation with a sensitivity <strong>of</strong> only20% (data not shown) which is consistent with other studies(Hancock et al., 2002).Cost-effectiveness <strong>and</strong> speed <strong>of</strong> analysisCostings for the Pyrosequencing assays <strong>and</strong> fluorescentPCR-RFLP using 2004 list prices were comparable to the costper sample being £1.20 (GBP) <strong>and</strong> £1.17, respectively, excludingsystem costs <strong>and</strong> machine maintenance contracts. Af-