Mutation screening of the mitochondrial genome using denaturing high-performance liquid chromatography

Mutation screening of the mitochondrial genome using denaturing high-performance liquid chromatography

Molecular Genetics and Metabolism 84 (2005) 61–74 Mutation screening of the mitochondrial genome using denaturing high-...

856KB Sizes 0 Downloads 3 Views

Molecular Genetics and Metabolism 84 (2005) 61–74

Mutation screening of the mitochondrial genome using denaturing high-performance liquid chromatography Andrew Biggina, Robert Henkeb, Bruce Bennettsa,c, David R. Thorburnd,e, John Christodouloua,c,¤ a

Western Sydney Genetics Program, Children’s Hospital at Westmead, Sydney, NSW, Australia b Millennium Science Pty. Ltd., 2/390 Canterbury Road, Surrey Hills, Vic., Australia c Discipline of Paediatrics and Child Health, University of Sydney, NSW, Australia d Murdoch Childrens Research Institute and Genetic Health Services Victoria, Royal Children’s Hospital, Melbourne, Vic., Australia e Department of Paediatrics, University of Melbourne, Vic., Australia Received 10 July 2004; received in revised form 15 September 2004; accepted 21 September 2004 Available online 11 November 2004

Abstract Over 170 known mutations of the mitochondrial genome are responsible for disease. Due to the unique features of mitochondrial genetics, such patients are clinically diverse and diYcult to diagnose. As pathogenic mitochondrial DNA (mtDNA) mutations are mostly heteroplasmic, denaturing high-performance liquid chromatography (DHPLC) could be used to detect these heteroplasmic species and therefore act as a rapid screening test for mtDNA mutations. The entire mitochondrial genome was ampliWed by PCR in 40 overlapping regions. In addition, known mtDNA mutants were constructed for each of these regions using a PCR-based sitedirected mutagenesis approach. These mutants were used as positive controls and showed a detection limit of 3–10% heteroplasmy by DHPLC (depending on the speciWc mutation) compared to 40% for conventional sequencing. To further validate the screening test, mtDNA from 17 patients with seven diVerent pathogenic mutations was used to compare mutation detection by DHPLC and conventional sequencing. DHPLC had a sensitivity of 88% compared to 82% for sequencing. This increased to 100% sensitivity for DHPLC when excluding the m.8993T>G mutation. DHPLC analysis is therefore a sensitive, rapid and cost-eVective method to screen for mutations in the mitochondrial genome. The role of pyrosequencing in the quantitation of mutant load for known mtDNA mutations was highlighted using the m.3243A>G mutation as an illustrative example. Pyrosequencing analysis was able to discriminate samples containing as little as 5% heteroplasmy and proved to be an accurate and reproducible method for estimation of mutant load.  2004 Elsevier Inc. All rights reserved. Keywords: Mutation detection; Mitochondrial genome; DHPLC; Pyrosequencing; Heteroplasmy; DNA

Introduction Mitochondrial DNA (mtDNA) is a 16,569 bp circle of maternally inherited, extra-nuclear, double-stranded DNA [1]. It contains 37 genes encoding two ribosomal RNAs, 22 transfer RNAs and 13 subunits of the respira-


Corresponding author. Fax: +612 9845 1864. E-mail address: [email protected] (J. Christodoulou).

1096-7192/$ - see front matter  2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ymgme.2004.09.011

tory chain [2]. In addition to its maternal inheritance, mtDNA diVers from nuclear DNA in that there are several hundreds to thousands of copies per somatic cell [3]. The exact amount of mtDNA per cell is tissue-dependent with large copy numbers found in ‘metabolically active’ tissues such as the central nervous system, skeletal muscle, and gastrointestinal system [1]. Another contrast to nuclear genes is the existence of heteroplasmy and the threshold eVect [1]. The presence of a single species of mtDNA is termed homoplasmy


A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74

with a mixture of wild-type and mutant mtDNA termed heteroplasmy. As the proportion of mutant mtDNA species increases beyond a threshold amount, clinical manifestations of the mutation become apparent—often with the severity of phenotype in proportion to the mutant load [4]. To date there have been over 170 reported mutations and over 1000 known polymorphisms of mtDNA [5]. Non-pathogenic polymorphisms are normally homoplasmic [6] whilst pathogenic mutations are mostly heteroplasmic [4]. Some mitochondrial disorders are associated with a characteristic set of clinical and pathological features [1]. However, the factors of heteroplasmy, threshold eVect and diVerential tissue distribution present a diagnostic challenge for clinicians presented with a patient harboring a mtDNA mutation. The variable age of onset, mode of presentation and rate of progression of many mitochondrial disorders makes diagnosis particularly diYcult [3]. The use of molecular genetic approaches to determine the presence of mtDNA mutations has had recent success but has often been limited to mutations that have been previously identiWed [7]. Other groups have also had limited success using techniques such as single-stranded conformation polymorphism [8] and two-dimensional electrophoresis [9]. Issues of sensitivity, speciWcity and labour intensive methodologies inherent in these techniques have been overcome by the use of denaturing high-performance liquid chromatography (DHPLC) [10]. This automated technology allows rapid detection of heterozygous and heteroplasmic mutations and has been extensively used for the diagnosis of cystic Wbrosis [11], breast cancer [12], and acute lymphoblastic leukemia [13]. In addition to these strategies, pyrosequencing has emerged as a new and accurate method for the detection of single nucleotide polymorphisms [14] and has lent itself to the detection of speciWc mtDNA mutations [15]. Several groups have recently described the use of DHPLC in screening for mitochondrial DNA mutations [16,17]. The van den Bosch study screened the entire mtDNA genome in 13 fragments and performed ‘multiplex’ DHPLC on each fragment following restriction enzyme digestion [16]. Liu et al. [17] used a more conventional DHPLC approach to investigate m.3243A>G mutations and variants of the non-coding D-loop region of mtDNA. Both of these studies illustrated the usefulness of DHPLC in mtDNA mutation detection, but were limited by the low number of individual mutations used to validate their screening strategies. Here, we describe the development and implementation of a DHPLC-based screening approach for the detection of mutations in mtDNA. The entire mitochondrial genome was divided into 40 overlapping regions and site-directed mutants were constructed for each region allowing a comprehensive assessment to be made of DHPLC sensitivity. A group of 17 patients covering

seven known mtDNA mutations were used to further validate this screening strategy. In addition, the important role of pyrosequencing in the quantitation of mutant load was highlighted using the m.3243A>G mutation as an illustrative example.

Materials and methods DNA samples Seventeen DNA samples from patients with known mtDNA mutations, covering seven individual mutations with varying mutant loads, were used to determine the sensitivity and speciWcity of the DHPLC screening test (see Table 1 for details). In addition, total DNA was isolated from healthy anonymous male donors using a salting-out method [18]. This DNA was used as a negative control and as a template to construct site-directed mutants that would provide positive-control specimens for all 40 fragments of the screening test. Primer sequences Primers (Gibco-BRL, Life Technologies) used for the ampliWcation of speciWc regions of the mitochondrial genome are shown in Table 2. These were also used in conjunction with ‘mutant’ primers (Table 3) to introduce single nucleotide substitutions for the production of speciWc site-directed mutants. Standard PCR conditions Reactions were performed in a 50 l volume using 1–50 ng template DNA, 0.1 mM dNTP (Pharmacia Table 1 Patient samples with known mtDNA mutations Mutation

Mutant load (%)

m.8993T>C m.8993T>C m.8993T>C m.8993T>C m.8993T>C m.8993T>C m.8993T>C m.8993T>G m.8993T>G m.3243A>G m.3243A>G m.13513G>A m.13513G>A m.14459G>A m.14459G>A m.8344A>G m.3303C>T

92 74 76 95 95 96 33 74 90 5 5 50 50 99 97 90 92

DNA samples were available from 17 patients, representing seven known mtDNA mutations, with varying mutant loads in each sample.

A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74


Table 2 Primers used for ampliWcation of the mitochondrial genome Fragment number


Size (bp)

Forward primer

Reverse primer

DHPLC temperatures

1a 1b 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21a 21b 22 23 24 25a 25b 26 27 28 29 30 31 32 33 34 35 36 37

16335–283 16495–451 361–921 756–1425 1234–1769 1587–2216 2105–2660 2417–3063 2986–3460 3396–3957 3815–4262 4150–4682 4656–5171 5097–5555 5468–5933 5862–6381 6361–6846 6744–7255 7178–7728 7645–8215 8204–8669 8639–9102 8899–9365 9310–9878 9754–10275 10127–10556 10290–10764 10600–11161 11142–11713 11622–12212 12174–12755 12601–13166 13099–13580 13428–13947 13715–14388 14046–14583 14434–14996 14924–15423 15260–15774 15733–16355

517 526 561 670 536 630 556 647 475 561 448 533 516 459 466 520 486 512 551 571 465 463 466 569 522 430 474 561 572 591 582 566 482 520 674 538 563 500 515 623



57, 60 57, 60 59, 60 60 59, 60 58 58, 59 59, 60 59, 60, 61 61 58, 59 55 59 60 58, 59, 60 58 61, 62 59 58, 59 60 57, 58 61 59, 60 58, 59 58 58, 59 58, 60 59, 60 58, 59 60 58 60 60 58, 59 58 56, 58 54, 55 59 58 56, 58

Forty overlapping fragments, covering the whole mitochondrial genome, were ampliWed. See Materials and Methods for details of the ampliWcation conditions. Also shown in the table are the denaturing temperatures used for DHPLC of each of these fragments. Region—region of the mitochondrial genome ampliWed by the primer set.

Biotech) and 14 pmol of the forward and reverse primer. Reactions using Taq polymerase were undertaken using 0.5 U of AmpliTaq Gold DNA polymerase enzyme (Perkin–Elmer) in its supplied buVer (15 mM Tris–HCl, pH 8, and 50 mM KCl) supplemented with 1.5 mM MgCl2. Reactions using Pfu DNA polymerase were carried out using 1 U of PfuTurbo DNA polymerase (Stratagene) in a buVer comprising 20 mM Tris–HCl, pH 8.8, 10 mM KCl, 10 mM (NH4)2SO4, 2 mM MgSO4 and 0.1 mg/ml BSA. The supplied buVer for the Pfu enzyme was not used as it contained detergents that are harmful to the DHPLC columns [19]. PCR was undertaken using the Hybaid PCR Express Thermal Cycler system (Hybaid, Middlesex, UK). Program parameters were 1 cycle of 95 °C for 5 min followed

by 30 cycles of 95 °C for 30 s, 56 °C for 30 s and 72 °C for 1 min. Following a Wnal incubation at 72 °C for 5 min the products were cooled to 4 °C. PCR products were tested by gel electrophoresis on a 2% agarose gel stained with ethidium bromide. Site-directed mutagenesis The production of speciWc mitochondrial DNA mutations was accomplished by using a PCR-based method [20]. The Wrst stage involved constructing two halves of the mutant in two separate PCRs. Fragment A was synthesised using forward (F) and mutant-reverse (MR) primers and fragment B was synthesised with mutant-forward (MF) and reverse (R) primers (Tables 2 and 3).


Table 3 Summary data for production of mtDNA mutants using site-directed mutagenesis Number


Forward position

Mutant forward primer

Reverse position

Mutant reverse primer

PCR A (F + MR)

PCR B (MF + R)

91C>T 91C>T 583G>A 1310C>T 1555A>G 1642G>A 2371T>G 2835C>T 3243A>G 3635G>A 4136A>G 4309G>A 4917A>G 5244G>A 5703G>A 5920G>A 6555A>G 6930G>A 7444G>A 7896G>A 8344A>G 8993T>G 8993T>G 9438G>A 9957T>C 10355C>A 10663T>C 10663T>C 11415C>A 12026A>G 12320A>G 12825T>G 13275A>C 13513G>A 14001A>G 14453G>A C14568T 15059G>A 15615G>A 15990C>T

(de novo) (de novo) MELAS DM DEAF MELAS (de novo) RETT CPEO LHON LHON CPEO LHON LHON CPEO Myoglobinuria (de novo) Multisys. dis. LHON Multisys. dis. MERRF NARP/LEIGH NARP/LEIGH LHON MELAS (de novo) LHON LHON (de novo) DM MM (de novo) (de novo) MELAS (de novo) MELAS LHON MM Ex. Intol. MM

81 81 571 1296 1547 1635 2361 2821 3231 3626 4121 4301 4907 5235 5691 5911 6545 6921 7433 7886 8331 8981 8981 9427 9945 10343 10653 10653 11406 12016 12314 12817 13267 13496 13991 14441 14559 15042 15604 15981


100 100 600 1326 1573 1657 2380 2850 3250 3645 4149 4320 4926 5254 5710 5930 6564 6942 7454 7905 8350 9002 9002 9446 9968 10365 10672 10672 11427 12036 12335 12836 13287 13530 14010 14462 14580 15076 15625 16000


334 174 240 571 340 71 275 434 264 249 335 171 271 158 243 68 204 199 277 261 146 363 103 137 215 239 382 72 286 415 162 236 189 103 296 417 146 153 366 268

202 370 351 130 223 582 299 243 229 331 142 382 265 321 243 470 302 334 296 330 338 121 384 452 331 214 111 508 308 197 442 350 314 452 398 143 407 382 171 375

Previously reported or de novo mutations were generated and used as positive controls for each of the PCR fragments subjected to DHPLC analysis (‘de novo’ represents a randomly selected single nucleotide substitution as no mutations have been published in these regions to-date). The mutant primers were used in conjunction with the normal primers to generate PCR products harbouring speciWc mtDNA mutations. The PCR fragment sizes for “PCR A” (using primer F and MR for each mutation) and “PCR B” (using primers MF and R) are also shown. MELAS, mitochondrial encephalopathy, lactic acidosis, and strokelike episodes; DM, diabetes mellitus; DEAF, sensorineural deafness; RETT, Rett syndrome; CPEO, chronic progressive external ophthalmoplegia; LHON, Leber hereditary optic neuropathy; myoglobinuria, recurrent episodic myoglobinuria; multisys dis, multisystem disease; MERRF, myoclonic epilepsy ragged red Wbre disease; NARP, neurogenic ataxia, retinitis pigmentosa; Leigh, Leigh disease; MM, mitochondrial myopathy; Ex. intol, exercise intolerance [5].

A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74

1a 1b 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21a 21b 22 23 24 25a 25b 26 27 28 29 30 31 32 33 34 35 36 37


A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74

PCRs were performed in a 50 l volume using 10 ng of anonymous male donor DNA, 0.1 mM dNTP (Pharmacia Biotech), 14 pmol of each of the relevant primer, 1 U of PfuTurbo DNA polymerase (Stratagene), 20 mM Tris– HCl, pH 8.8, 10 mM KCl, 10 mM (NH4)2SO4, 2 mM MgSO4, and 0.1 mg/ml BSA. As the predicted melting temperatures of the mutant primers varied from those of the standard primers, a touchdown PCR protocol was utilised. Program parameters were 1 cycle of 95 °C for 5 min followed by four cycles of 95 °C for 30 s, 65 °C for 30 s and 72 °C for 1 min. A further 10 cycles were conducted with a denaturing step of 95 °C for 30 s, an annealing step of 65 °C for 30 s (dropping by 1 °C per cycle to 56 °C) and an extension step of 72 °C for 1 min. A further 24 cycles of 95 °C for 30 s, 56 °C for 30 s, and 72 °C for 1 min were carried out before a Wnal incubation at 72 °C for 5 min. PCR products were separated by gel electrophoresis on a 2% agarose gel stained with ethidium bromide and subsequently puriWed using the Jet-Quick gel puriWcation kit (Genomed, NC, USA) according to the manufacturer’s recommendations. A summary of the primers and expected sizes of fragments A and B are shown in Table 3. The two PCR products A and B were then assayed by a DU650 spectrophotometer (Beckman Instruments) and 30 ng of each combined for a Wnal PCR. This Wnal PCR synthesis utilised the standard forward (F) and reverse (R) primers according to the methods outlined in the ‘standard PCR conditions’ described previously. The production of the required site-directed mutation was conWrmed by sequencing analysis (data not shown). Where possible the construction of site-directed mtDNA mutants were based on previously published mutations [5]. However, nine of the 40 regions to be ampliWed contained no previously reported mutation. For these regions (shown as ‘de novo’ in Table 3) a randomly selected single nucleotide substitution was introduced within the highest melting domains of the fragment. Conventional BigDye terminator sequence analysis PCR fragments were puriWed with the QIAquick PCR puriWcation kit (Qiagen) according to the manufacturer’s protocol and assayed using a DU650 spectrophotometer (Beckman Instruments). Fourteen picomoles of the forward or reverse primer and 200–400 ng of puriWed PCR product were combined in a Wnal volume of 16 l. This was sequenced at the SUPAMAC Centre, Royal Prince Alfred Hospital, Camperdown, New South Wales according to standard protocols. Heteroduplex formation PCR products were denatured at 95 °C for 10 min and then slowly cooled to 65 °C at a rate of 1 °C/min. Following a 5-min incubation at 65 °C, samples were cooled to 4 °C at a rate of 1 °C every 5 s.


DHPLC analysis DHPLC analysis was performed on an Helix DHPLC system (Varian). The stationary phase consisted of a column packed with C18 alkylated silica which binds DNA during analysis. The mobile phase consisted of two eluants. BuVer A contained triethylammonium acetate (TEAA) which interacts with the negatively charged phosphate groups on the DNA as well as the surface of the column [19]. BuVer B contained TEAA with 25% (v/v) acetonitrile and was used as a gradient to elute the bound DNA from the column. This gradient consisted of increasing the proportion of buVer B from 43 to 48% over the Wrst 30 s, increasing it again from 48 to 66% over the following 5.5 min and then maintaining it at 66% for a further 1 min. The remaining 2 min were maintained at 43% buVer B in preparation for the next sample injection. All Xow rates were at 0.45 ml/min. Successful detection of DNA heteroduplex species required the elution to be carried out at the melting temperature of the sequence in question. Temperatures for successful resolution of heteroduplexes were calculated by the DHPLC Melt Program [21] available online ( melt.html) and were also experimentally determined. Data was collected and analysed using Star Chromatography Workstation Software Version 5.51 (Varian). For comparative analysis of sequencing and DHPLC, sensitivity was deWned as the probability that a test was positive if a mutation was present. SpeciWcity was deWned as the probability a test was negative if a mutation was absent. Pyrosequencing PCRs were performed in a 50 l volume using 10 ng template DNA, 0.1 mM dNTP (Pharmacia Biotech) and 10 mol of the forward (5⬘-biotin-AGAAATAAGGCC TACTT) and reverse (5⬘-CCATGGGTATGTTGTTA AGAA) primer. Reactions were undertaken using 0.5 U of AmpliTaq Gold DNA polymerase enzyme in its supplied buVer (15 mM Tris–HCl, pH 8, and 50 mM KCl) supplemented with 1 mM MgCl2. PCR was performed using 1 cycle of 95 °C for 10 min followed by 45 cycles of 95 °C for 15 s, 53 °C for 30 s, and 72 °C for 115 s. Following a Wnal incubation at 72 °C for 5 min the products were cooled to 4 °C. PCR products were visualised by gel electrophoresis on a 2% agarose gel stained with ethidium bromide to conWrm a single product of 180 bp. PCR puriWcation and pyrosequencing were undertaken according to the manufacturer’s instructions (Pyrosequencing, Uppsala, Sweden) using 30 pmol of sequencing primer (5⬘-GCGATTACCGGGC) per reaction. Single-stranded DNA-binding protein was added to the primed DNA template prior to the pyrosequencing reaction to reduce mispriming and increase signal intensity [22].


A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74


The eVects of DNA polymerase

DHPLC of site-directed mutants

Despite multiple attempts, several regions did not yield heteroduplexes with Taq polymerase. The regions in question did result in a clean, single PCR product of the correct size and sequence analysis conWrmed that the correct product was formed (data not shown). However, despite numerous attempts under diVerent conditions these regions failed to form clear heteroduplexes with Taq polymerase (Fig. 4). Fig. 4 also illustrates DHPLC analysis of samples under identical conditions but with the use of Pfu instead of Taq DNA polymerase. The use of Pfu DNA polymerase resulted in products with a greater amplitude upon DHPLC analysis, and more easily identiWable heteroduplexes. These data led to us routinely using Pfu DNA polymerase in all subsequent reactions.

DHLPC analysis was performed at a range of temperatures on samples containing a mixture of wild-type and site-directed mutant in a 1:1 stoichiometry. Fig. 1 compares the proWles formed with these samples compared to wild-type at their established melting temperatures. These data illustrate the formation of heteroduplex species with mutant/wild-type mixtures compared to the homoduplex proWles of wild-type samples. Heteroduplex formation was consistent and pronounced for all regions of the mitochondrial genome with the exception of the m.8993T>G mutation of regions 22 and 23 that only exhibited small changes for heteroduplex detection. DHPLC analysis of 100% sitedirected mutant products revealed homoduplex proWles that were indistinguishable from wild-type (data not shown). Sensitivity of DHPLC and sequencing To compare the sensitivity of DHPLC and conventional sequencing it was Wrst necessary to produce heteroplasmic mtDNA samples. This was achieved by mixing known concentrations of site-directed mutant and wild-type puriWed PCR products to produce a series of mutant loads between 1 and 99%. These samples were subjected to both DHPLC and sequencing analysis. Fig. 2 illustrates the DHPLC analysis of the m.3635G>A mutation at diVerent mutant loads. It can be seen that mutant loads of 3% are indistinguishable from wild-type but concentrations of 6% and greater exhibit heteroduplex formation. These heteroduplexes were seen at increasing mutant load concentrations beyond 75% but were no longer detectable at loads of 95%. However, 1:1 mixing of these high mutant loads with wild-type mtDNA resulted in the re-detection of heteroduplexes. These results were typical examples of the sensitivities obtained from the diVerent regions of the mitochondrial genome. Analysis of diVerent mtDNA mutations revealed a lower sensitivity threshold of between 3 and 10% mutant load. These mirrored the associated upper threshold detection limits of 90–97% mutant load. However, mixing of high mutant loads with wild-type DNA enabled detection of mutant loads as high as 99%. Conventional (Xuorescence-based) sequencing analysis of the m.3635G>A mutation is shown in Fig. 3. At mutant loads less than 40% it was often diYcult to interpret potential heteroplasmy due to ‘background noise.’ Detection of heteroplasmy was reproducible with mutant loads of 40–100%. Interestingly, the automated base-assignment software could only reliably detect mutants with greater than 65% mutant load. These results were typical of the diVerent regions analysed.

Analysis of patient mtDNA The mtDNA from 17 patients harbouring seven known mtDNA mutations (Table 1) and 14 control samples were subjected to both DHPLC and sequence analysis. Typical analyses of these mutations are summarised in Fig. 5. Mutation detection by sequencing analysis was successful for 14 of the 17 known mutations. Those undetected included the m.8993T>C mutation with 33% mutant load (data not shown) and two m.3243A>G mutations with 5% mutant load (Fig. 5), with a calculated sensitivity of using sequencing as a screening test of 82%. Sequencing analysis was able to identify all 14 control samples with no false-positive results. Analysis of patient mtDNA using DHPLC revealed the presence of all mutation regions with the exception of the 74% (data not shown) and 90% m.8993T>G mutations (Fig. 5), yielding a calculated sensitivity of using DHPLC as a screening test of 88%. With the exclusion of the m.8993T>G mutations, this sensitivity increased to 100%. The successful detection of four of the patients with the m.8993T>C mutations depended on diluting samples with wild-type mtDNA prior to analysis. These samples had original mutant loads of 92, 95, 95, and 96%. A 99% m.14459G>A mutation also required dilution prior to detection. Interestingly, a 92% m.3303C>T mutation and 97% m.14459G>A mutation did not require any dilution for mutation detection. DHPLC also correctly identiWed the 14 control samples without any false-positive results, giving a speciWcity of 100%. Quantitation of m.3243A>G mutant load by pyrosequencing Heteroplasmic mtDNA samples with diVerent m.3243A>G mutant loads were produced by mixing

A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74


Fig. 1. DHPLC of wild-type and site-directed mutant fragments. This shows the DHPLC proWles for each of the mtDNA mutations generated by site-directed mutagenesis. The wild-type and mutant species were mixed in a 1:1 ratio, run at the temperatures speciWed in Table 2, and compared with the proWles from wild-type mtDNA alone.


A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74

each sample as indicated by a correlation coeYcient of 0.9996 (data not shown). As shown in Table 4, pyrosequencing analysis could statistically discriminate samples with 5–95% mutant loads from wild-type or 100% site-directed mutants (p < 0.05, two-tailed t test).


Fig. 2. DHPLC of m.3635G>A mutation with diVerent mutant loads. Varying amounts of mutant and wild-type species were mixed, and their DHPLC proWles compared. It can be seen that the proWle was indistinguishable from wild-type (w.t.) when the mutant load was 3% or less or 95% or more.

known concentrations of site-directed mutant and wildtype puriWed PCR products. These samples were collectively analysed by pyrosequencing (Table 4). The wild-type (0% load) and site-directed mutant samples (100% load) were used to calibrate the software so that calculated mutant loads could be produced for the remaining samples. The mutant load (mean %) as calculated by pyrosequencing was consistent with the actual mutant load of

The development of a DHPLC-based screening test for the detection of mtDNA mutations requires stringent validation and assessment of both speciWcity and sensitivity. Validation of each of the 40 regions of the screening test was undertaken by the successful detection of speciWc site-directed mutants to each of these regions. Where possible, site-directed mutagenesis was based on currently identiWed mtDNA mutations with only nine regions having no associated mutation. Mutants for these regions were constructed containing a random single nucleotide substitution within the highest melting domains. DHPLC analysis of 1:1 mixtures of wild-type and site-directed mutant revealed the presence of heteroduplex species. The successful detection of mutant mtDNA samples occurred within 2 °C of the predicted temperature (as determined by the DHPLC Melt program [21]). Comparison of the resulting heteroduplexes from each of the 40 regions revealed a marked diVerence in the proWles obtained. Detection of the m.8993T>G mutation was particularly subtle with only a slight ‘shoulder’ being detectable in region 23. It is interesting to note that the m.8993T>G mutations under investigation in this paper occur in the highest melting domain of the PCR amplicon. Variability in heteroduplex detection has been previously described [23] and has been attributed to variability in heteroduplex stability, the degree of surrounding denaturation and the nature of the mutation (including surrounding base pairs). Analysis of multiple sequence alterations in a deWned DNA region [24] has also revealed that mutations in low melting domains were easily resolved by DHPLC whilst those in high melting domains were unlikely to be detected. Although conditions were optimised for mutation detection in each given fragment, the successful detection of mutations in high melting domains and the use of a range of diVerent temperatures for heteroduplex analysis increases the likelihood of detecting mutations elsewhere in the amplicon. However, the existence of the diVerent variables outlined in the above studies [23,24] means that detection of all possible mutations within a given amplicon is not guaranteed. Another interesting observation was that the detection of some of these mutations was dependent on the use of Pfu DNA polymerase instead of Taq DNA polymerase. It is widely appreciated that the success of DHPLC in mutational analysis is dependent on the quality of PCR product [10]. Although DHPLC does not

A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74


Fig. 3. Conventional BigDye terminator sequencing of m.3635G>A mutation with diVerent mutant loads. Using automated base-assignment software, the mutation could only be reliably called correctly when the mutant load was greater than 65%, and even when chromatograph traces were examined manually, it was diYcult to identify the mutation when the mutant load was less than 40%.

require any special PCRs or conditions, reliable DHPLC analysis requires ampliWed fragments of high quality. Despite much success with Taq DNA polymerase [16,23], touchdown PCR using Pfu polymerase [25] is recommended for DHPLC. Touchdown PCR minimises mis-primed products while the proof reading Pfu polymerase minimises PCR-induced mutations. Pfu DNA polymerase exhibits the lowest error rate of any thermostable DNA polymerase [26] making it the enzyme of choice from this respect. However, it is diYcult to explain the observed results based on enzyme Wdelity alone. One would expect that if Taq polymerase did induce excessive errors during PCR, artifactual heteroduplex formation would occur rather than a complete absence of heteroduplex formation. Due to the errone-

ous results with Taq polymerase, Pfu DNA polymerase was used routinely throughout the screening test. Sensitivity of the screening test was determined using various mutations with diVerent mutant loads (formed by mixing wild-type and mutant mtDNA in various ratios). DHPLC was able to detect levels of heteroplasmy as low as 3% (m.3243A>G mutation), with all other tested regions being detectable at heteroplasmy of 5–10%. DHPLC could also detect heteroplasmy as high as 99% (with additional dilution with wild-type mtDNA). The variability in the detection sensitivity of diVerent mutations may reXect the position of the mutation and degree of surrounding denaturation as stated above [24]. In comparison with the broad detection range of DHPLC, sequencing analysis was only able to


A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74

Fig. 4. DHPLC of m.2835C>T mutation using Taq or Pfu polymerase. Under otherwise identical conditions, Pfu consistently resulted in products that generated a higher amplitude on DHPLC analysis, and more readily formed heteroduplexes when mixed with wild-type PCR products. w.t., wild-type; sdm, site-directed mutant.

detect mutant loads of 40% or greater. Furthermore, if sequencing background was suYciently high this threshold increased. The automated base-assignment software could only reliably detect mutants with greater than 65% mutant load thus emphasising the need for manual review of sequences, and the use of newer generation sequence analysis programs such as Mutation Surveyor (SoftGenetics). The entire mitochondrial genome of 17 patients (covering seven known mutations) and 14 control samples were screened using both DHPLC and sequencing analysis. Both sequencing and DHPLC analysis correctly identiWed all 14 control samples with no false positive results. The calculated speciWcity was therefore 100% for both of these methodologies. For detection of pathogenic mutations, DHPLC was superior to sequencing

with a calculated sensitivity of 88% compared to 82%. The superior sensitivity of DHPLC resulted from the ability to detect patients with low mutant loads. Such sensitivity would have been 100% upon the exclusion of the patients with m.8993T>G mutations due to the poor detection of mutations in this high melting domain area. The m.8993T>G mutation and the milder variant m.8993T>C (NARP and Leigh disease) represent the most common mtDNA mutations diagnosed in children [27]. Compared to other mtDNA mutations they show very little tissue or age dependent variation with mutant load correlating well with symptoms [28]. As high mutant load is required for clinical phenotype, sequencing can be employed for diagnostic purposes. It should be remembered that a mixture of two populations of mtDNA molecules does not always indicate a

A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74


Fig. 5. Comparison of conventional sequencing and DHPLC analyses for selected mutations. With the exception of the m.8993T>G mutation, at lower mutant loads DHPLC was more likely to identify a mutation than was direct sequencing. w.t., wild-type.

causative mutation is present. Although polymorphisms are normally homoplasmic [6], it may be possible that several mtDNA species can co-exist, each containing diVerent polymorphisms. The use of DHPLC cannot easily distinguish polymorphisms from mutations but the formation of heteroduplexes in any given region highlights this area for further investigation. Furthermore, as the detection of very high mutant loads requires mixing with wild-type mtDNA, the presence of diVerent polymorphisms may also result in heteroduplex formation in the absence of a pathogenic mutation—a phe-

nomenon that is very likely considering there are over 1000 known polymorphisms covering all regions of the mtDNA genome [29]. In an attempt to reduce the occurrence of these phenomena, the patient DNA could be mixed with a series of samples from individuals with known polymorphisms. Other important areas of consideration also include the diVerential tissue distribution of mutant mtDNA. It has previously been established that the distribution of mutated mtDNA molecules may diVer depending on the organs investigated. This is exempliWed in Kearns–Sayre


A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74

Table 4 Estimation of m.3243A>G mutant load by pyrosequencing Known mutant load (%)

Calculated mutant load (mean %)

95% ConWdence interval



0 0.5 1 2.5 5 10 20 50 75 85 95 100

0.8 0.8 1.1 2.0 4.2 7.4 18.9 49.2 75.8 85.5 95.8 100.0

0.8 to 0.8 ¡2.5 to 4.0 ¡0.5 to 2.7 ¡2.6 to 6.6 2.6 to 5.8 4.4 to 10.4 18.5 to 19.3 44.2 to 54.2 71.8 to 79.8 81.9 to 89.1 94.2 to 97.4 100.0 to 100.0

2 7 7 7 6 6 7 7 7 7 7 2

1.00 0.98 0.74 0.67 0.04 0.02 <0.01 <0.01 <0.01 <0.01 0.04 1.00

Calculated mutant load (mean and 95% conWdence interval) following pyrosequencing analysis of known mutant samples. Samples containing mutant loads between 5 and 95% were signiWcantly diVerent from either wild-type or 100% site-directed mutant samples (p < 0.05, two-tailed t test).

Syndrome patients whose circulating lymphocytes often do not harbour the characteristic mtDNA deletions [3]. To prevent misdiagnosis of a mitochondrial disorder it is therefore important to analyse mtDNA from diVerent tissues, particularly the clinically aVected tissues. However, this is only of vital importance in those conditions with a lymphocyte mutant load of less than 5–10% because of the sensitivity of DHPLC. An alternative approach to elucidate tissue-speciWc near-homoplasmic mutations could be to mix the mtDNA from diVerent tissues of the same person. For example, mixing mtDNA from muscle of a mitochondrial myopathy patient with mtDNA from lymphocytes not harbouring the mutation would make mutation detection by DHPLC more sensitive. This approach would also minimise the erroneous (polymorphismbased) heteroduplexes that are inherent when mixing mtDNA from diVerent people. However, such an approach would only be utilised if clinically indicated or if the initial DHPLC screening run was negative. DHPLC has a proven history in the diagnosis of many disorders including haemophilia [23], Marfan syndrome [30], cystic Wbrosis [11], breast cancer [12], and acute lymphoblastic leukemia [13]. In all of these studies, DHPLC has been shown to be superior to alternative methodologies for reasons of sensitivity, time or labour resources. DHPLC is also at least an order of magnitude less expensive than sequencing [31]. The use of DHPLC for the screening of mitochondrial DNA mutations has been recently described [16,17]. The Wrst of these papers ampliWed the mtDNA genome in only 13 fragments and used DHPLC analysis following restriction enzyme digestion of these fragments. This study used only two mutations (m.3243A>G and m.8344A>G) to test the sensitivity and speciWcity of the DHPLC system and could only detect three mutations (m.3302A>G, m.3271T>C, and m.9176T>C) from six patients with mitochondrial disease [16]. Despite using 13 fragments rather than 40, the necessity for

restriction enzyme digestion and the wider range of temperatures required in the van den Bosch study do not make it more time or cost eVective. Liu et al. [17] used only two mtDNA mutations for optimization of DHPLC mutation detection and were only concerned with screening for m.3243A>G mutations and variants of the non-coding D-loop region of mtDNA. Pyrosequencing has recently been shown to be a sensitive and reproducible method for detection of mtDNA polymorphisms [15]. Using the m.3243A>G mtDNA mutation as an illustrative example, pyrosequencing was able to detect samples with 5% mutant load or greater (incidentally, DHPLC analysis of this region also showed a detection limit of 5% mutant load indicating that pyrosequencing is as sensitive as DHPLC for detection of this speciWc mutation). Although it is not currently cost-eVective or time-eYcient to have a complete screening strategy based on pyrosequencing, this emphasises the role for pyrosequencing in the quantitation of mutant load for known mtDNA mutations. In addition to pyrosequencing, a number of alternative strategies have also been developed that could be utilised to identify speciWc mtDNA mutations. Mismatch repair detection (MRD) is a technique that is capable of detecting a single mismatch in a DNA fragment as large as 10 kb in size and is a powerful method for high throughput genotyping and mutation detection [32]. Alternatively, sequencing by hybridisation (SBH) is an indirect sequencing method that allows the development of high-throughput, low-cost, miniaturised sequencing processes on arrays of DNA samples or probes [33]. These techniques are both highly sensitive and speciWc and can allow detection of mutations as low as 1% [34]. The development and implementation of the DHPLC-based screening strategy described in this paper therefore allows a rapid, cost-eVective, and sensitive method for the detection of mitochondrial mutations. Moreover, it represents the most comprehensive validation of a screening test for the entire mitochondrial

A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74

genome described to-date. Despite the apparent superiority of DHPLC over conventional sequencing both methodologies have the potential to miss certain mutations, including large deletions. The most eYcient screening strategy would involve DHPLC analysis of ‘unmixed’ patient mtDNA and conventional sequencing of heteroduplexes and/or known homoplasmic mutations if clinically indicated. For deletions, Southern blotting or long-range PCR would be required. Pyrosequencing could also be directly employed to detect mutations aVecting nucleotide 8993 as DHPLC cannot readily detect T>G mutations at this position. If these initial strategies fail to identify a mtDNA mutation then further DHPLC screening of mixed patient/maternal mtDNA or mixed patient muscle/lymphocyte mtDNA may help to elucidate an underlying mtDNA mutation. Once the mutation is identiWed, the mitochondrial mutant load can be quickly and accurately determined by pyrosequencing.

Acknowledgments This research was supported by an Australian National Health and Medical Research Council (NHMRC) project grant, the Cecilia Kilkeary Foundation, and the Children’s Hospital Fund of the Children’s Hospital at Westmead. D.R.T. is a NHMRC Senior Research Fellow.

References [1] S. DiMauro, E.A. Schon, Mitochondrial DNA mutations in human disease, Am. J. Med. Genet. 106 (2001) 18–26. [2] S. Anderson, A.T. Bankier, B.G. Barrel, M. DeBruijin, A.R. Coulson, J. Drouin, I.C. Eperon, D.P. Nierlich, B.A. Roe, F. Sanger, P.H. Schreier, A. Smith, R. Staden, I.G. Young, Sequence and organization of the human mitochondrial genome, Nature 290 (1981) 457–465. [3] A. Munnich, P. Rustin, Clinical spectrum and diagnosis of mitochondrial disorders, Am. J. Med. Genet. 106 (2001) 4–17. [4] S.M.D. DiMauro, C.T. Moraes, Mitochondrial encephalomyopathies, Arch. Neurol. 50 (1993) 1197–1208. [5] MITOMAP (D.C. Wallace, M.T. Lott, “MITOMAP: A Human Mitochondrial Genome Database, 2004.). [6] M. Zeviani, C. Antozzi, Mitochondrial disorders, Mol. Hum. Reprod. 3 (1997) 133–148. [7] C.-Y. Lu, D.-J. Tso, T. Yang, Y.-J. Jong, Y.-H. Wei, Detection of DNA mutations associated with mitochondrial diseases by Agilent 2100 bioanalyzer, Clin. Chim. Acta 318 (2001) 97–105. [8] C. Eng, J. Vijg, Genetic testing, Nat. Biotechnol. 15 (1997) 422–426. [9] N.J. Van Orsouw, X. Zhang, J.Y. Wei, D.R. Johns, J. Vijg, Mutational scanning of mitochondrial DNA by two-dimensional electrophoresis, Genomics 52 (1998) 27–36. [10] W. Xiao, P.J. Oefner, Denaturing high-performance liquid chromatography: a review, Hum. Mut. 17 (2001) 439–474. [11] C. Le Marechal, M.P. Audrezet, I. Quere, O. Raguenes, S. Langonne, C. Ferec, Complete and rapid scanning of the cystic Wbrosis transmembrane conductance regulator (CFTR) gene by denatur-








[19] [20] [21]











ing high-performance liquid chromatography (D-HPLC), Hum. Genet. 108 (2001) 290–298. C. Eng, L.C. Brody, T.M.U. Wagner, P. Devilee, J. Vijg, C. Szabo, S.V. Tavtigian, K.L. Nathanson, E. Ostrander, T.S. Frank, Interpreting epidemiological research: blinded comparison of methods used to estimate the prevalence of inherited mutations in BRCA1, J. Med. Genet. 38 (2001) 824–833. U. Zur Stadt, J. Rischewski, R. Schneppenheim, H. Kabisch, Denaturing HPLC for identiWcation of clonal T-cell receptor gamma rearrangements in newly diagnosed acute lymphoblastic leukemia, Clin. Chem. 47 (2001) 2003–2011. H. Fakhrai-Rad, N. Pourmand, M. Ronaghi, Pyrosequencing: an accurate detection platform for single nucleotide polymorphisms, Hum. Mut. 19 (2002) 479–485. H. Andreasson, A. Asp, A. Alderborn, U. Gyllensten, M. Allen, Mitochondrial sequence analysis for forensic identiWcation using pyrosequencing technology, Biotechniques 32 (2002) 124–126. B.J. Van Den Bosch, R.F. de Coo, H.R. Scholte, J.G. Nijland, R. van Den Bogaard, M. de Visser, C.E.M. de Die-Smulders, H.J.M. Smeets, Mutation analysis of the entire mitochondrial genome using denaturing high-performance liquid chromatography, Nucleic Acids Res. 28 (2000) E89. M.-R. Liu, K.-F. Pan, Z.-F. Li, Y. Wang, D.-J. Deng, L. Zhang, Y.Y. Lu, Rapid screening mitochondrial DNA mutation by using denaturing high-performance liquid chromatography, World J. Gastroenterol. 8 (2002) 426–430. S.A. Miller, D.D. Dykes, H.F. Polesky, A simple salting out procedure for extracting DNA from human nucleated cells, Nucleic Acids Res. 16 (1988) 1215. Varian Inc, Standard Operating Procedure. Rev 7. 2001. K.E. Gustin, R.D. Burk, A rapid method for generating linker scanning mutants utilizing PCR, Biotechniques 14 (1993) 22–24. A.C. Jones, J. Austin, N. Hansen, B. Hoogendoorn, P.J. Oefner, J.P. Cheadle, M.C. O’Donovan, Optimal temperature selection for mutation detection by denaturing HPLC and comparison to single-stranded conformation polymorphism and heteroduplex analysis, Clin. Chem. 45 (1999) 1133–1140. M. Ronaghi, Improved performance of pyrosequencing using single-stranded DNA-binding protein, Anal. Biochem. 286 (2000) 282–288. M.C. O’Donovan, P.J. Oefner, S.C. Roberts, J. Austin, B. Hoogendoorn, C. Guy, G. Speight, M. Upadhyaya, S.S. Sommer, P. McGuYn, Blind analysis of denaturing high-performance liquid chromatography as a tool for mutation detection, Genomics 52 (1998) 44–49. T.R. Skopek, W.E. Glaab, J.J. Monroe, K.L. Kort, W. Schaefer, Analysis of sequence alterations in a deWned DNA region: comparison of temperature-modulated heteroduplex analysis and denaturing gradient gel electrophoresis, Mutat. Res. 430 (1999) 13–21. R.H. Don, P.T. Cox, B.J. Wainwright, K. Baker, J.S. Mattick, ‘Touchdown’ PCR to circumvent spurious priming during gene ampliWcation, Nucleic Acids Res. 19 (1991) 4008. J. Cline, J.C. Braman, H.H. Hogrefe, PCR Wdelity of Pfu DNA polymerase and other thermostable DNA polymerases, Nucleic Acids Res. 24 (1996) 3546–3551. S.L. White, V.R. Collins, R. Wolfe, M.A. Cleary, S. Shanske, S. DiMauro, H.H. Dahl, D.R. Thorburn, Genetic counselling and prenatal diagnosis for the mitochondrial DNA mutations at nucleotide 8993, Am. J. Hum. Genet. 65 (1999) 474–482. D.R. Thorburn, H.H. Dahl, Mitochondrial disorders: genetics, counselling, prenatal diagnosis and reproductive options, Am. J. Med. Genet. 106 (2001) 102–114. A.M. Kogelnik, M.T. Lott, M.D. Brown, S.B. Navathe, D.C. Wallace, MITOMAP: a human mitochondrial genome database— 1998 update, Nucleic Acids Res 26 (1998) 112–115. W.O. Liu, P.J. Oefner, C. Qian, R.S. Odom, U. Francke, Denaturing HPLC-identiWed novel FBN1 mutations, polymorphisms, and


A. Biggin et al. / Molecular Genetics and Metabolism 84 (2005) 61–74

sequence variants in Marfan syndrome and related connective tissue disorders, Genet. Test. 1 (1997) 237–242. [31] T.M.U. Wagner, D. Stoppa-Lyonnet, E. Fleischmann, D. Muhr, S. Pagès, T. Sandberg, V. Caux, R. Moeslinger, G. Langbauer, A. Borg, P. Oefner, Denaturing high-performance liquid chromatography (DHPLC) detects reliably BRCA1 and BRCA2 mutations, Genomics 62 (1996) 369–376. [32] M. Faham, D.R. Cox, A novel in vivo method to detect DNA sequence variation, Genome Res. 5 (1995) 474–482.

[33] R. Drmanac, S. Drmanac, G. Chui, R. Diaz, A. Hou, H. Jin, P. Jin, S. Kwon, S. Lacy, B. Moeur, J. Shafto, D. Swanson, T. Ukrainczyk, C. Xu, D. Little, Sequencing by hybridisation (SBH): advantages, achievements, and opportunities, Adv. Biochem. Eng. Biotechnol. 77 (2002) 75–101. [34] H. Fakhrai-Rad, J. Zheng, T.D. Willis, K. Wong, K. Suyenaga, M. Moorhead, J. Eberle, Y.R. Thorstenson, T. Jones, R.W. Davis, E. Namsaraev, M. Faham, SNP discovery in pooled samples with mismatch repair detection, Genome Res. 14 (2004) 1404–1412.