Detection and quantification of the A3243G mutation of mitochondrial DNA by ligation detection reaction

Molecular and Cellular Probes (1998) 12, 273–282 Article No. ll980191

Detection and quantification of the A3243G mutation of mitochondrial DNA by ligation detection reaction M. Nigou,1 B. Parfait,2 E. Clauser1 and J. L. Olivier1∗ 1

Laboratoire commun de biologie mole´culaire, Hoˆpital Saint Antoine, Paris, France and 2 Laboratoire de Ge´ne´tique, Hoˆpital Necker, Paris, France (Received 9 February 1998, Accepted 19 May 1998)

The A3243G mutation of mitochondrial DNA is associated to the MELAS syndrome and to transmitted forms of diabetes mellitus. This mutation exists in a heteroplasmic state and can be present at a minor and hardly detectable level. The aim was to design a method which could be applied to large series of samples and could provide rapid, sensitive and quantitative detection of this mutation in the wild-type mitochondrial DNA background. The ability of ligation detection reaction (LDR) to satisfy these objectives was evaluated. Ligation detection reaction was performed on a model template composed of mixtures of various proportions of plasmids bearing the wildtype or mutant mitochondrial DNA sequence. Radiolabelled or fluorescent primers and the wildtype and mutant LDR products were separated by electrophoresis on conventional denaturating gel or on an Applied Biosystem 373. The ratios of mutant/wild-type products were consistent with the initial ratios of the plasmids in the template. The sensitivity and accuracy of the fluorescence and isotopic detection methods were similar. The detection limit of mutant DNA was 10% of total mitochondrial DNA. The percentage of mutant DNA in DNA samples extracted from leukocytes of 19 patients having the mutation at different levels, was evaluated by fluorescent or isotopic LDR.  1998 Academic Press

KEYWORDS: mitochondrial DNA, diabetes, MELAS, ligation chain reaction, polymerase chain reaction, point mutation.

INTRODUCTION Several point mutations of mitochondrial DNA have been correlated with various encephalomyopathy syndromes including LHON (Leber’s hereditary optic neuropathy), MERRF (myoclonus epilepsy associated with ragged red fibers) and MELAS (myopathy, encephalopathy, lactic acidosis and stroke-like episodes; min 540,000). MELAS syndrome is associated with a G→A mutation at position 3243 in the tRNALeu(UUR) gene. This is also the binding site of a transcription termination factor. The 3243 G→A mutation is heteroplasmic, as are many other mitochondrial point mutations and there are large

variations in the mutant DNA of different affected patients and of tissues in individual patients. The degree of correlation between the proportion of mutant mitochondrial DNA and the severity of the clinical and biochemical phenotype is unclear. Mariotti et al.1 found that the percentage of the A3243G mutation in a series of 22 patients with MELAS syndrome was correlated with the specific activity of respiratory chain complex I. However, no correlation was found with the clinical features except lactic acidosis which was present in all the patients. MELAS syndrome can occur in young

∗ Author to whom all correspondence should be addressed at: Laboratoire commun de biologie mole´culaire, Service de Biochimie B, Hoˆpital Saint Antoine, 184 rue du Faubourg Saint Antoine, 75012 Paris France.

0890–8508/98/050273+10 $30.00/0

 1998 Academic Press

274

M. Nigou et al.

patients as well as those older than 25 and the percentage of mutated DNA is correlated with the age of onset.1 Several studies have suggested that the A3243G mutation may also be a pathogenic factor for non-insulin-dependent diabetes mellitus.2,3 Kadowaki et al.4 found the mutation in 6% of patients with IDDM and family history of diabetes and 2% of patients with NIDDM and family history of diabetes. They concluded that diabetes mellitus associated with the A3243G mutation constitutes a subtype of diabetes in Japan. Therefore it is interesting to evaluate the frequency of this mutation and its predictive value in European populations. The percentage of mutant mitochondrial DNA may be low in peripheral leukocytes but high in poorly accessible tissues such as muscle and brain.5 Thus, a rapid, sensitive and specific method is required to detect minor mutation rates in large populations. The A3243G mutation creates an ApaI restriction site, which has made it possible to quantify the mitochondrial DNA heteroplasmy by conventional restriction fragment length polymorphism (RFLP) and Southern blot analysis. However, partial digestion by ApaI and variations in efficiency of transfer and hybridization makes sensitive and accurate quantification difficult. The combination of RFLP analysis with polymerase chain reaction (PCR) improves sensitivity but there is still a risk of partial digestion and this may be increased by heteroduplex formation during PCR. Polymerase chain reaction-single strand conformation polymorphism (SSCP) has also been used to detect the A3243G mutation6 and other mitochondrial point mutations5 and to determine the level of heteroplasmy. The frequency of mutations around the tRNALeu gene is high and mutations around position 3243 associated with the MELAS syndrome have also been found.7–9 The mutation causing the migration shift cannot be identified by SSCP, so subsequent microsequencing must also be used to determine the nature of the mutation. An automated non-radioactive method was devised based on the ligation detection reaction (LDR) to detect and quantify the percentage A3243G mutation in the wild-type mitochondrial DNA background. Ligation detection reaction has mostly been used to detect point mutations in nuclear DNA and is based on the alternative ligation of one of two upstream oligonucleotides to a common downstream oligonucleotide. The upstream oligonucleotides each have a different nucleotide at the 3′ end which is the point of the mutation. In this study, the sensitivity and accuracy of isotopic and fluorescent LDR was evaluated in a model system and then the method

assessed using DNA from patients affected by the MELAS syndrome.

MATERIALS AND METHODS Materials Taq ligase was provided by New England Biolabs (USA). Taq polymerase, T4 ligase and polynucleotide kinase, and dNTP were from Pharmacia (Uppsala, Sweden). c-32P ATP was obtained from Amersham (Buckinghamshire, UK). 6-FAM-derived and unlabelled oligonucleotides were provided by GENSET (Paris, France). Genomic DNA from leukocytes of patients affected by the MELAS syndrome were kindly provided by Professor A. Mu¨nnich (Necker Medical School, Paris, France).

Plasmid constructions The plasmid MT2 containing the 2240–4200 region of wild-type mitochondrial DNA, was a gift from Professor Malthiery and Dr Raynier (Biochemistry Department, Angers Medical School, France). This plasmid was used in the production of the MT3 plasmid containing the 3015–3468 sequence of mitochondrial DNA with the A3243G mutation. MT3 was produced by PCR site-directed mutagenesis using OCN1 and OCN3 oligonucleotides or OCN2 and OCN4 oligonucleotides as primers (Fig. 1). OCN3 is the antisense and OCN4 the sense strand of the 3230–3256 region of mitochondrial DNA containing the A3243G mutation, so overlapping 250 and 248 bp length amplicons were obtained by 30 cycles of PCR (denaturation 95°C 1 min, annealing 55°C 1 min, 72°C 1 min). The amplicons were purified by electroelution and aliquots were mixed with OCN1B and OCN2B primers (Fig. 1) and used for a second round of PCR. The mixed primary amplicons were denatured at 95°C for 5 min, re-annealed by cooling at 55°C for 5 min and extended for 1 min at 72°C. Thirty cycles of PCR were then performed in the same conditions as for the first round of PCR to obtain a 473 bp fragment. The final amplicon was ligated into the HindIII and XbaI sites of the pUC19 vector to produce the MT3 plasmid.

Polymerase chain reaction (PCR) and ligation detection reaction (LDR) Genomic DNA (500 ng), MT2 or MT3 plasmids (5 fmoles total) were used as a template to amplify the

Ligation detection reaction for A3243G mitochondrial DNA mutation

275

Fig. 1. Sequences and locations of the oligonucleotides used in the plasmid constructions, polymerase chain reaction (PCR) and ligation detection reaction (LDR) procedures. The OCN series of oligonucleotides was used for PCR, the LSseries mainly for ligation reactions. OCN1B contains the sequence of OCN1 and a HindIII linker sequence and OCN2B contains the sequence of OCN2 and an XbaI linker sequence.

3015–3377 region of mitochondrial DNA in a 50 ll reaction volume containing 0·5 l OCN1 and OCN5 oligonucleotides, 100 l dNTP, 1·5 m MgCl2, 10 m Tris-HCl pH 9, 50 m KCl and 0·5 U Taq polymerase. The DNA mixture was denatured by heating to 94°C for 4 min and 30 cycles of PCR (denaturation 94°C 1 min, annealing 55°C 1 min, extension 72°C 1 min) were performed on a Perkin Elmer 480 thermocycler. The efficiency of the PCR was checked and 1 ll of the amplicon (approximately 50 fmoles) was added to a 20 ll reaction mixture containing 50 m TrisHCl pH 7·5, 10 m MgCl2, 10 m DDT, 1 m ATP, 25 lg BSA, 80 U Taq ligase and 300 fmoles of the oligonucleotides used for LDR. LD3W oligonucleotide (50 pmoles) was labelled in presence of 50 lCi c-32P ATP (Specific activity 3000 Ci/mmole) using 10 U T4 polynucleotide kinase for 1 h at 37°C. Distilled water was added to the reaction mixture to give a final volume of 50 ll and the specific activity of the labelled LD3W oligonucleotide was evaluated by spotting 1 ll onto a PEI cellulose TLC and separating the probe from free phosphate and ATP by chromatography in 0·5  ammonium bicarbonate. The spot corresponding to the probe was cut and counted in a LS6000SC scintillation counter. The probe was purified on a sephadex G50 minicolumn, 300 fmoles of the radiolabelled LD3W oligonucleotide (200,000 cpm) were added to the LDR reaction mixture with the unlabelled LS1W and LS2M oligonucleotides (300 fmoles) for isotopic LDR. The

fluorescent LS1Wfam, LS2Mfam analogues (300 fmoles) were added with the unlabelled LD3W oligonucleotide (300 fmoles) for non-isotopic LDR. Ten cycles of LDR were performed (denaturation 94°C 1 min, annealing and ligation 45°C 4 min) except when indicated. For isotopic detection, 10 ll of LDR product was loaded on a 15% acrylamide/bisacrylamide (40:2) gel containing 7  urea and subjected to electrophoresis at 30 mA for 30 min on a Biorad minigel apparatus. An autoradiograph was produced by exposing X-ray film to the gel at −80°C for 1 h. The bands corresponding to the wild-type and mutated LDR products were excised, put in a scintillation vial and the cerenkov units were counted on a LS6000SC Beckman scintillation counter to determine the degree of heteroplasmy. An Applied Biosystem 373 automatic sequencer was used for detection of fluorescence. Ligation detection reaction products (5 ll) were loaded on a 6% acrylamide/ bisacrylamide (40:2) gel containing 7  urea and subjected to electrophoresis at 30 mA for 3 h.

Evaluation of the intra-assay and inter-assay standard deviations DNA samples (500 ng each) from four patients were amplified in a single series of PCRs to evaluate intraassay standard deviation. Each PCR product was used

M. Nigou et al.

for five LDRs with radiolabelled LD3W primer or 6FAM-labelled LS1W and LS2M oligonucleotides (10 reactions in total). For evaluation of the inter-assays standard deviation, human genomic DNA (500 ng) was used as a template in six separate series of PCRs followed by LDRs with either radiolabelled LW3D primer (three experiments) or LS1Wfam and LS2Mfam oligonucleotides (three experiments). The results of fluorescent and isotopic LDR were analysed by repeated measures analysis of variance to determine whether the two detection methods gave significantly different results.

(a) 1

2

PM

276

3

4

154 bp

(b) 1

2

3

4

RESULTS 48 bp

The aim of this investigation was to find a method for rapid detection of the A3243G mitochondrial mutation and accurate determination of the level of heteroplasmy in large series of samples. Polymerase chain reaction was performed using primers sharing the same antisense primer (OCN5) but two different sense primers. LS1W for the wildtype genotype and LS2M for the mutant genotype (Fig. 1). The 3′ ends of the sense primers LS1W and LS2M were positioned on the 3243 site of the mutation to impede amplification in case of mismatch with the template. The LS1W oligonucleotide was also seven bases longer than LS2M to make it possible to distinguish the two PCR products after gel electrophoresis if both genotypes were present in the template. Both primer pairs produced a band of 150–160 bp using a wild-type genomic DNA (Fig. 2a). Moreover different non-specific products were also observed. Similar results were obtained when the annealing temperature was raised to 60°C or the MT2 plasmid was used as template (data not shown). This meant that PCR could not be used to analyse the A3243G mutation. Therefore, a new oligonucleotide LD3W, corresponding to the 3244–3264 region, was designed and used with LS1W and LS2M oligonucleotides to evaluate detection and quantification of the A3243G mutation by LDR. Polymerase chain reaction was performed on wild-type genomic DNA (Fig. 2b) or the MT2 and MT3 plasmids (Fig. 2c) using OCN1 and OCN5 primers. Aliquots of the PCR products were used for LDR with the 32P 5′-end labelled LD3W oligonucleotide and unlabelled LS1W or LS2M oligonucleotides. The wild-type primers (LS1W and LD3W) detected a 48 base wild-type fragment (Fig. 2b, lanes 1 and 2; Fig. 2c, lanes 5 and 6) but not the mutant 41 base product (Fig. 2c, lanes 7 and 8). The LS2M and LD3W primers detected the 41 base product (Fig. 2c, lanes 3 and 4) but not the wild-type sequence (Fig. 2c, lanes 1 and 2).

(c) 1

2

3

4

5

6

7

8 48 b 41 b

Fig. 2. Detection of the A3243G mutation of mitochondrial DNA by polymerase chain reaction (PCR) and ligation detection reaction (LDR). (a) PCR was performed with the LS1W (wild-type) (lanes 1 and 3) or LS2M (mutant) (lanes 2 and 4) oligonucleotides as sense primers on human wild-type genomic DNA. OCN5 oligonucleotide was used as an antisense primer and two different genomic DNA samples from a patient without MELAS syndrome (wild-type at position 3243) were used as template (0·5 lg genomic DNA). Thirty PCR cycles were performed (denaturation 94°C, annealing 55°C and extension 72°C for 1 min each). (b) LDR was performed on 50 fmoles of wild-type product using radiolabelled LD3W oligonucleotide and either LS1W (lane 1, 2) or LS2M (lane 3, 4) oligonucleotides. Ten cycles of LDR were performed on the samples in lanes 2 and 4, 20 cycles on those in lanes 1 and 3. (c) LDR was performed on 50 fmoles of product amplified from the wild-type HT2 plasmid (lanes 5–8) or the mutant MT3 plasmid (lanes 1–4) using radiolabelled LD3W oligonucleotide and either LS1W (lanes 1, 2, 5, 6) or LS2M (lanes 3, 4, 7, 8) oligonucleotides. Ten cycles were performed on the samples in lanes 2, 4, 6, 8 and 20 cycles on those in lanes 1, 3, 5, 7.

The specificity of the LDR was not affected by increasing the number of LDR cycles to 20 or 30. Varying the quantity of PCR product used as a template for LDR from 10 to 250 fmoles did not affect the specificity of LDR and the percentage of bound oligonucleotides did not appear to increase above 50 fmoles of PCR product in the presence of 60 fmoles

Ligation detection reaction for A3243G mitochondrial DNA mutation

277

Fig. 3. Linearity of ligation detection reaction (LDR) performed on mixtures of wild-type and mutant polymerase chain reaction (PCR) products. Amplicons corresponding to the 3015–3358 region of mitochondrial DNA were produced from MT2 or MT3 plasmids using OCN1 and OCN5 oligonucleotides and mixed in various amounts before LDR (lanes 3–7). Polymerase chain reaction was carried out on MT3 plasmid and LDR was subsequently performed using pure LS1W (lane 1) or a mixture of LS1W and LS2M oligonucleotides (lane 2). MT2 plasmid was used for PCR and discriminant LS1W oligonucleotide alone (lane 7) or mixed with LS2M primer (lane 9) was used for LDR.

of oligonucleotides (data not shown). Increasing the amount of oligonucleotides to 600 fmoles did not affect the specificity or the ratios of bound/free oligonucleotides although it did increase the yield of LDR products and shortened the exposure time required to detect them (data not shown). Ligation detection reaction remained specific when annealing temperature was lowered to 37°C. Some mitochondrial-like sequences are found in the nuclear genome.10 Thus, PCR reactions, which are performed before LDR, can lead to inaccurate estimations of the amount of mutant DNA and the choice of the primers is critical to avoid this. No amplicons were produced when the OCN1 and OCN2 primers were used on DNA extracted from rho0 HeLa cells (data not shown). The authors investigated whether the amounts of mutant and wild-type LDR product varied linearly depending on the amount of PCR product used as template. Polymerase chain reaction products amplified from the MT2 or MT3 plasmids were mixed in various amounts from 1 to 10 fmoles (10 fmoles in total) and used as template in LDR with radiolabelled LD3W oligonucleotide and unlabelled LS1W or LS2M oligonucleotides. The calculated ratios of the 48 base/ 41 base LDR products were the same as the theoretical ratios of wild-type and mutant PCR substrates (Fig. 3). This shows that the method is competitive and linear. The effects of PCR on the linearity of detection were also examined. MT2 and MT3 plasmids were mixed in various ratios and 10 fmoles of each mixture was used for PCR and LDR. Wild-type 48 base and mutant 41 base products were separated by electrophoresis on 15% acrylamide minigels and the bands were excised and counted. There was no difference in the intensity of the wild-type and mutant signals when the detection primers LS1W and LS2M were

used together or separately in LDR (Fig. 4a). There was a good correlation between the percentage of 41 base LDR product detected and the theoretical amount of mutant MT3 plasmid (Fig. 4b) showing that the result obtained with PCR and LDR in these conditions accurately reflected the mutant/wild-type ratio of the initial template despite the exponential nature of the PCR process. No mutant product was detected in this isotopic system if less than 10% mutant MT3 plasmids were used as template. Genescan analysis (Perkin Elmer—Applied Biosystems Division) was tested to see if it was possible to avoid use of radioactivity and to improve LDR product size determination. Labelled oligonucleotide LD3W was used in the isotopic detection system. It bound to both the mutant and wild-type sequences and was therefore incorporated in both the mutant and wild-type products. Ligation detection reaction discriminates mismatches at the 3′ end of the upstream primer but not at the 5′ end of the downstream primer (data not shown). Only 3′-end labelled fluorescent oligonucleotides were available to us, so two different 5′ 6-fam-labelled discriminant primers LS1Wfam and LS2Mfam were used with the unlabelled common LD3W oligonucleotide. The LS1Wfam and LS2Mfam oligonucleotides were purified and their specific fluorescence was checked. The use of fluorescent primers had no effect on the specificity of the reaction (Fig. 5a). A good correlation was also observed between the percentage of mutant signal obtained after PCR-LDR and the amount of MT3 plasmids initially incorporated in the template, using mixtures of MT2 and MT3 plasmids for PCR-LDR (Fig. 5b). No mutant LDR product was detected if total plasmid DNA contained less than 10% MT3 mutant plasmid. This limit of sensitivity was similar to that observed with the isotopic detection technique.

M. Nigou et al.

278 (a) 1 2

3

4

5 6

7

8

9

WT M

Free (b) 100

% Mutant LDR product

y = 18.74x + 0.86 2 R = 0.986 75

50

25

0

1.25 2.5 3.75 MT3 plasmid (fmoles)

5

Fig. 4. Isotopic quantification by polymerase chain reaction-ligation detection reaction (PCR-LDR) of the A3243G mutation in mixtures of wild-type MT2 and mutant MT3 plasmids. MT2 and MT3 plasmids were mixed in various proportions and 5 fmoles of each mixture was used for PCR with the OCN1 and OCN5 oligonucleotides. (a) 5 fmoles MT2 (lanes 1–3) or MT3 (lanes 7–9) plasmid or MT2/MT3: 1:1 mixtures (lanes 4–6) were used for PCR and then for LDR with LS1W (lanes 1, 4, 7), LS2M (lanes 2, 5, 8) or both primers (lanes 3, 6, 9) in presence of radiolabelled LD3W oligonucleotide. WT indicates the wild-type 48 base and M the mutant 41 base product. Free indicates the radiolabelled LD3W not incorporated into ligation proportion of MT3 plasmid in the initial PCR template. The proportion of mutant LDR product was calculated using the sum of the radioactivity incorporated in the 41 and 48 base bands.

The experiments described above were performed using model templates consisting of plasmids containing wild-type or mutant mitochondrial DNA sequences. This method was also applied to DNA samples from 19 patients with MELAS syndrome. These patients had between 20 and 60% mutant DNA as detected by LDR using the radiolabelled LD3W primer. Four DNAs were selected with low (sample 6) or high (samples 7, 9 and 11) levels of mutant DNA to compare the results obtained by isotopic and fluorescence-based LDR and to evaluate the intraassay standard deviation. The percentages of mutant DNA recorded were 19·9±2·6 for sample 6, 38·7±3·9 for sample 7, 44·6±1·4 for sample 9 and

33·8±1·3 for sample 11 using the radioactive technique (Fig. 6). Similar percentages of mutant DNA were recorded with the fluorescent method (13·7±1 for sample 6, 40·9±0·3 for sample 7, 36·2±1·3 for sample 9 and 33·8±1·3 for sample 11; Fig. 6). The mean intra-assay deviations were 0·8% for fluorescence-based and 2·5% for isotopic detections which suggests that discrepancies in the estimation of the mutant DNA level may result mainly from PCR. The entire series of DNAs (19 samples) were used for separate assays of PCR-LDR using radiolabelled primers (three experiments) of 6-FAM labelled LS1W and LS2M primers (three experiments) to evaluate the interassay deviations and the correlation between the results of the isotopic and fluorescence methods. Performing different PCR reactions on the same samples before LDR resulted in higher standard deviations (Fig. 7). The mean standard deviation was higher for fluorescent LDR than for isotopic detection (9·3% compared to 3%). The two detection systems gave consistent percentages of mutant DNA for 16 of the 19 samples studied (Fig. 7). Differences greater than 15% were observed between two measurements of mutant DNA in three patients. The fluorescent method gave a higher value in two cases and the isotopic method in one case. Statistical analysis by the repeated measures analysis of variance method showed there was no significant difference between the fluorescence and isotopic detection systems. The authors were unable to detect the A3243G mutation by isotopic LDR in DNA samples from 100 obese patients with or without associated diabetes mellitus but without MELAS syndrome. These results show that the LDR-PCR combined method provides specific, rapid and quantitative detection of the A3243G mutation of mitochondrial DNA.

DISCUSSION The sequential use of PCR and LDR combines sensitivity and specificity. Polymerase chain reaction alone can lead to false positive results in the presence of mismatches between the 3′ ends of the primers and the template depending on the type of mismatch.11 It was found that LDR accurately detected as little as 10% of mitochondria with the A3243G mutation while PCR alone gave false positive results. The G/T mismatch has been described as the least well discriminated by thermostable ligase,12 but no nonspecific ligation product was observed when the annealing temperature was reduced to 37°C, or the number of LDR cycles increased to 20. Concentrations of target DNA 50 times higher and oligonucleotides seven times higher than Barany12 were

Ligation detection reaction for A3243G mitochondrial DNA mutation

279

(a) 1 29 39 49 59 69

2 29 39 49 59 69

3 29 39 49 59 69

A

B

C

WT (b)

M WT

M

100

% Mutant LDR product

y = 17.05x + 0.55 2 R = 0.945 75

50

25

0

1.25

2.5 MT3 plasmid (fmoles)

3.75

5

Fig. 5. Quantification by polymerase chain reaction-ligation detection reaction (PCR-LDR) of the A3243G mutation in mixtures of wild-type MT2 and mutant MT3 plasmids using fluorescent LDR primers. MT2 and MT3 plasmids were mixed in various proportions and 5 fmoles of each mixture was used for PCR with the OCN1B and CN5 oligonucleotides. (a) 10 ng MT2 (lanes 1A, 1B, 1C) or MT3 (lanes 3A, 3B, 3C) plasmids or MT2/MT3 1:1 mixture (lanes 2A, 2B, 2C) were used for PCR and then for LDR with LS1Wfam (lanes 1A, 2A, 3A), LS2Mfam (lanes 1B, 2B, 3B) or both primers (lanes 1C, 2C, 3C) in the presence of unlabelled LD3W oligonucleotide. WT indicates the wild-type LDR, product, M the mutant product. (b) Quantification of the proportion of mutant LDR product in relation to the proportion of MT3 plasmid in the initial PCR template. The percentage of mutant LDR product was calculated using the sum of the fluorescences of the mutant and wild-type products which are 41 and 48 bases long, respectively.

used in order to detect and quantify a rare product in a complex background. The ligation temperature was also lowered to increase the linearity of the reaction. The mutation specific LDR oligonucleotide

was designed seven bases shorter than the wild-type primer to enable the products to be distinguished easily. Length differences in the discriminant oligonucleotides were also used by Eggerding et al.13 to

M. Nigou et al.

280 32 P Wildtype 6-FAM

AB CD

32

Mutant

P

6-FAM

A B CD

0 15 30 45 60

0 15 30 45 60

Patient 6

Patient 7

100 6-FAM

% Wild-type or mutant DNA

90 80

WT P M

70

32

60

A B CD

50 40

A B CD

0 15 30 45 60

0 15 30 45 60

Patient 9

Patient 11

30 6-FAM 20 10 0

P WT M

32

6

7 9 Patients

11

Fig. 6. Intra-assay variation of the quantification of the A3243G mutation by ligation detection reaction (LDR) in DNA samples from four patients affected by MELAS. One polymerase chain reaction (PCR) with OCN1 and OCN5 primers was performed on each DNA sample and the product was used in the same assay for 10 LDRs using either unlabelled LS1W, LS2M oligonucleotides and 32P 5′-end labelled LD3W primer (five reactions), or unlabelled LD3W primer and LS1Wfam, LS2Mfam oligonucleotides (five reactions). (a) Estimates of the mutant and wild-type DNA percentages calculated as the average of five LDR reactions in each detection system for each patient. Error bars correspond to the standard deviations. (b) Typical genescan patterns are shown for each patient. Peak A is the free LS1Wfam oligonucleotide and peak B is the free LS2Mfam oligonucleotide. Peak C is the wild-type 48 base product and peak D is the mutant 41 base LDR product. Autoradiographs of the five isotopic LDR are shown below the genescan pattern for each patient.

detect the quantify mutations of the CFTR gene. Differences could theoretically introduce some discrepancy in the formation of the wild-type or mutant product. However, the percentage of mutant LDR product detected was directly related to the amount of plasmid bearing the mutation added as template, especially when isotopic detection was used. The sequences of the LS1W and LS1Wfam oligonucleotides contained five more A and two more G/C than those of the LS2M and LS2Mfam oligonucleotides. This did not have a major effect on Tm (59°C instead of 64°C). Ligation detection reaction was performed at a reaction temperature 14–19°C below the Tm, which maintained the specificity of the LDR at this level of sensitivity. The detection limit for mutant DNA in the wild-type background was 10% using a radiolabelled common oligonucleotide or two discriminant fluorescent primers. This is equivalent to 3×109 mutant copies in the 50 fmoles of PCR product. Much higher sensitivity can be obtained by using thermostable ligases. Ka¨lin et al.14 detected 250 copies of the Ha-ras codon 61 mutation by performing 20–40 cycles of ligation and using two complementary oligonucleotides in

addition to the discriminant and common primers. These modifications converted LDR to LCR (ligation chain reaction). Barany12 used 6×108 target molecules in LDR and 6×106 target molecules in LCR to evaluate the specificity of these methods. Ligation chain reaction is a potentially exponential reaction like PCR and less well suited for quantitative assays. Isotopic LDR is easy to perform and provides rapid, accurate results. However, fluorescent primers have several advantages such as safe handling and stability and they also provide quantitative results. The use of various fluorophores also makes the detection of different products in the same reaction possible. 5′ labelled fluorescent primers were originally used by Landegren et al.15 to demonstrate the bS mutation by LDR with thermolabile T4 ligase. In this study we show that fluorescent primers can be used in LDR to accurately evaluate the level of heteroplasmy of mitochondrial DNA for the A3243G mutation. The specific fluorescence of both the mutant and wildtype discriminant primers must be carefully checked to avoid an underestimation of one of the ligation products. Fluorescence-based LDR gave estimates of

Ligation detection reaction for A3243G mitochondrial DNA mutation

281

Fig. 7. Inter-assay variation of the quantification of the A3243G mutation by ligation detection reaction (LDR) in DNA samples from 19 patients affected by MELAS (myopathy, encephalopathy, lactic acidosis and stroke-like episodes). Each DNA sample was used for six polymerase chain reactions (PCRs) with OCN1 and OCN5 primers. Aliquots of each PCR product were used as template for LDR using either unlabelled LS1W, LS2M oligonucleotides and 32P 5′-end labelled LD3W primer, or unlabelled LD3W primer and LS1Wfam, LS2Mfam oligonucleotides. Three independent isotopic and three independent fluorescence LDR were performed on each template. Error bars correspond to the standard deviations.

the level of the A3243G mutation consistent with the values obtained with isotopic detection. There was a lower estimation of mutant DNA by fluorescent LDR for only one patient (patient 6) of the three for whom differences greater than 15% were observed between methods. This may be due to a slightly lower sensitivity of fluorescent LDR than of isotopic detection. Heteroplasmy could thus be underestimated when the percentage of mutant DNA reaches the limit of sensitivity. Ligation detection reaction does not require preliminary extraction procedures, unlike PCR-RFLP systems. Heteroduplex formation does not impede mutation detection. The presence of point mutations upstream or downstream from the mutation target site has little effect on the annealing temperature and therefore on ligation unlike SSCP in which the pattern can be greatly changed by such a mutation. In the same assays Eggerding et al.13 showed that multiple mutations of the cystic fibrosis transmembrane regulator (CFTR) gene could be performed in a single assay. This approach could also be applied to the simultaneous detection of other mutations of mitochondrial DNA. As these mutations may be rare and very diverse in many populations, LDR-PCR can rapidly and accurately detect and quantify them.

ACKNOWLEDGEMENTS The authors thank Dr Raynier and Professor Malthiery (Angers Medical School, Biochemistry Department, Angers, France) for the gift of the MT2 plasmid and Professor A. Mu¨nnich (Necker Medical School, Genetics Department, Paris, France) for having provided the DNA samples with the A3243G mutation. The authors thank Dr G. Thomas (Saint Antoine Medical School, Biochemistry Department, Paris, France) for the use of DNA from obese patients and Dr F. Carrat (Saint Antoine Medical School, Biostatistics Department, Paris, France) for his help in statistical analysis. The authors also thank Dr Julie Knight for her editorial help.

REFERENCES 1. Mariotti, C., Savarese, N., Suomalainen, A. et al. (1995). Genotype to phenotype correlations in mitochondrial encephalomyopathies associated with the A3243G mutation of mitochondrial DNA. Journal of Neurology 242, 304–12. 2. Van den Ouweland, J. M. W., Lenkes, H. H. P. J., Ruitenbeek, W. et al. (1992). Mutation in mitochondrial tRNALeu(UUR) gene in a large pedigree with maternally transmitted type II diabetes mellitus and deafness. Nature Genetics 1, 368–71. 3. Gerbitz, K. D., Paprotta, A., Jaksch, M., Zierz, S. & Drechsel, J. (1993). Diabetes mellitus is one of the

282

4.

5.

6.

7.

8.

9.

M. Nigou et al. heterogeneous phenotypic features of a mitochondrial DNA point mutation within the tRNALeu(UUR) gene. FEBS Letters 321, 194–96. Kadowaki, T., Kadowaki, H., Mori, Y. et al. (1994). Diabetes mellitus is one of the heterogeneous phenotypic features of a mitochondrial DNA point mutation within the tRNALeu(UUR) gene. New England Journal of Medicine 330, 962–68. Sherrat, E. J., Thomas, A. W., Gagg, J. W. & Alcolado, J. C. (1995). Nonradioactive characterization of lowlevel heteroplasmic mitochondrial DNA mutations by SSCP-PCR enrichment. Biotechniques 20, 430–32. Tanno, Y., Yoneda, M., Tanaka, K. et al. (1995). Quantitation of heteroplasmy of mitochondrial tRNA (Leu (UUR)) gene using PCR-SSCP. Muscle and Nerve 18, 1390–97. Goto, Y., Nonaka, I. & Horai, S. (1991). A new mitochondrial mutation associated with mitochondrial myopathy encephalopathy, lactic acidosis and stroke-like episodes (MELAS). Biochimica Biophysica Acta 1097, 238–40. Goto, Y., Tojo, M., Tohyama, J., Horai, S. & Nonaka, I. (1991). A novel point mutation in the mitochondrial tRNALeu(UUR) gene in a family with mitochondrial myopathy. Annals of Neurology 31, 672–75. Hao, H. & Moraes, C. T. (1996). Functional and molecular mitochondrial abnormalities associated with a

10.

11.

12.

13.

14.

15.

C→T transition at position 3256 of the human mitochondrial genome. Journal of Biological Chemistry 271, 2347–52. Tsuzuki, T., Nomiyana, H., Setoyama, C., Maeda, S. & Shimada, K. (1983). Presence of mitochondrialDNA like sequences in human nuclear DNA. Gene 25, 223–29. Newton, R. C. (1995). Mutational analysis: known mutations. In PCR2. A Practical Approach (McPherson, M. J., Hames, B. D. & Taylor, G. R., ed.) pp 219–252. Oxford, New York, Tokyo: Oxford University Press. Barany, F. (1991). Genetic disease detection and DNA amplification using cloned thermostable ligase. Proceedings of the National Academy of Science USA 88, 189–93. Eggerding, F. A., Iovannisci, D. M., Brinson, E., Grossman, P. & Winn-Deen, E. S. (1995). Fluorescencebased oligonucleotide ligation assay for analysis of cystic fibrosis transmembrane conductance regulator gene mutations. Human Mutation 5, 153–65. Ka¨lin, I., Shephard, S. & Candrian, U. (1992). Evaluation of the ligase chain reaction (LCR) for the detection of point mutations. Mutation Research 283, 119–23. Landegren, U., Kaiser, R., Sanders, J. & Hood, L. (1988). A ligase-mediated gene detection technique. Science 241, 1077–80.