Long-Extension PCR to Detect Deleted Mitochondrial DNA Molecules Is Compromized by Technical Artefacts

Biochemical and Biophysical Research Communications 254, 507–514 (1999) Article ID bbrc.1998.9975, available online at http://www.idealibrary.com on

Long-Extension PCR to Detect Deleted Mitochondrial DNA Molecules Is Compromized by Technical Artefacts Olli A. Kajander,* ,† Tarja A. Kunnas,* ,† Markus Perola,‡ Sanna K. Lehtinen,† ,§ Pekka J. Karhunen,* ,† and Howard T. Jacobs† ,§ ,¶,1 *Department of Forensic Medicine, University of Tampere, Finland; †Tampere University Hospital, Tampere, Finland; ‡Department of Human Molecular Genetics, National Public Health Institute, Helsinki, Finland; §Institute of Medical Technology, University of Tampere, Finland; and ¶Institute of Biomedical and Life Sciences, University of Glasgow, Scotland

Received December 7, 1998

Long-extension PCR (LX-PCR), followed by Southern hybridization to probes for two different regions of the mitochondrial genome, was used to evaluate the presence of deleted mtDNA molecules in heart muscle samples from alcoholic cardiomyopathy patients compared with age-matched controls. Two different primer pairs capable of amplifying the entire genome, as well as a variety of other primer pairs predicted to amplify the genome in large, overlapping fragments, were tested. Products indicating the presence of a variety of subgenomic, deleted molecules were detected in variable amounts from patient and control myocardial samples alike. Most of these hybridized with a probe for the 16S/ND1 region, but not with a probe for the ND4/ND5 region that is commonly deleted. Dilution of a given template DNA in which deleted products were prominent resulted in the disappearance of the subgenomic bands in favour of the full-length, undeleted product. Therefore, the appearance and amount of such products is subject to template concentration or quality. The results indicate that the application of LX-PCR to the detection and quantitation of deleted mtDNAs is inherently unreliable, and findings using this technique should be treated with caution unless supported by an independent method. © 1999 Academic Press

Heteroplasmic deletions of mitochondrial DNA are associated with a variety of human pathological states. In some instances, for example the group of sporadic mitochondrial encephalomyopathies including KearnsSayre syndrome, deleted mtDNA molecules are generally regarded as causal of disease (1), although definitive proof of this is lacking. In such disorders, a single, 1 To whom correspondence should be addressed. Institute of Medical Technology, University of Tampere, PO Box 607, 33101 Tampere, Finland. Fax: 1358-3-215-7731. E-mail: [email protected].

clonally expanded deletion product is found in abundance in the affected tissues, most commonly skeletal muscle. In some other disorders, including a number of mitochondrial syndromes showing autosomal inheritance (2– 4) many different deleted molecules have been found in each affected individual, which is also the case in disorders such as inclusion body myositis, where the presence of deleted mtDNA is arguably a consequence rather than a cause of the primary pathology (5–7). The accumulation of multiple deleted species of mtDNA has also been reported as a concomitant of aging in humans (8, 9) and experimental animals (10, 11). Cardiomyopathy is a feature of some mitochondrial syndromes involving deleted mtDNA molecules (4), and deleted mtDNAs have also been found in cases of isolated cardiomyopathy (12–16), mainly of the dilated type (14). The quantitative significance of such lesions has, however, been challenged by other investigators, who failed to find a meaningful association between the disorder and two specific, mtDNA deletions that could also be detected at a low level in control heart tissue by a PCR-based method (17). Deleterious point mutations of mtDNA have been reported in other cases of cardiomyopathy, including both hypertrophic (18, 19) and dilated (20) forms. The most generally accepted and reliable method for detecting and quantifying the full range of deleted mtDNA molecules is Southern blotting. However, the method is dependent upon both the quality and quantity of DNA available. The DNA must be pure enough to be efficiently digested with diagnostic restriction enzymes, must be sufficiently intact that the strandlength is significantly greater than that of the largest restriction fragments to be detected, and must be available in sufficient quantities (typically a minimum of 1–5 mg per track) to give a reasonable autoradiographic or phosphorescent signal after hybridization, given that deleted molecules may individually represent only

507

0006-291X/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.

Vol. 254, No. 2, 1999

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

a minor fraction of the total. In practice these conditions are difficult to satisfy, especially when all that is available is a small biopsy sample, or post-mortem tissue which may have been preserved under suboptimal conditions. For these reasons, many investigators have preferred PCR methods as a way of detecting deleted molecules. One particular technique which is becoming widely used for this purpose is long-extension PCR (LX-PCR). This relies upon progressive increases in the extension time in succeeding synthesis cycles, to favour the net synthesis of longer DNA molecules (8, 21). This application typically uses adjacent, ‘outwardly oriented’ mtDNA primers, that are, in principle, capable of amplifying a linear product equivalent to the entire mitochondrial genome, or any subgenomic-sized deletion product that actually contains the priming site. Although the method is not strictly quantitative, since it still favours the synthesis of shorter over longer products, it has been adopted as a method for profiling the full set of deletion products that may be present in a given template. More conventional PCR approaches are capable of detecting just one or a small subset of the many thousands of possible deletions. In this study we set out to evaluate whether deleted mtDNAs are associated with alcoholic cardiomyopathy (ACM), one type of dilated cardiomyopathy. Mitochondrial damage is a feature of the disorder (22), and in animal models, mitochondrial changes in the heart have been reported to be associated both with acute and chronic alcohol administration (23), although the literature on this point is far from unanimous (24). Acute ethanol administration results in a decreased synthesis of ventricular mitochondrial proteins in vivo (25) and acute ethanol treatment of cultured cardiomyocytes produces metabolic and ultrastructural (26) changes to mitochondria. Chronic ethanol treatment of rats results in altered levels and defective regulation of cardiac mitochondrial ATP synthase (27), upregulation of respiratory chain complexes (28), altered mitochondrial transcript levels (29) and long-term changes in heart mitochondrial ultrastructure interpreted as damage resulting from prolonged oxidative stress (30). Multiple, deleted species of mtDNA have recently been reported in liver of some chronic alcohol abusers (31), based on results obtained using a PCR approach. In principle, finding an association between mtDNA deletions and the disorder may indicate a mechanism of toxicity of the drug, or else a predisposing etiological factor. Because only relatively small amounts of DNA of variable quality were available for the study, we decided to adopt the LX-PCR approach described above. To ensure that all replication-competent, deleted mtDNAs could be detected, we followed reasoning used by others, and designed primers located within the conserved portion of the control region, required for the initiation of transcription and replication. Further-

more, to guard against ‘primer-specific’ artefacts we also designed primers from another region of the genome, located within the 16S rRNA gene, which were tested in parallel. Except for those molecules in which the region of the second priming site is deleted, the set of products synthesized, if meaningful, would be expected to be very similar in the two cases. In addition, we amplified overlapping, large segments of the genome, using various pairs of non-adjacent primers, each of which would be predicted to give rise to a subset of deletion-derived products compatible with the patterns of products obtained using the full genomelength primers. Because the method is at best semi-quantitative, we applied a number of additional controls and criteria for evaluating our data, as follows. Firstly, if the appearance of products corresponding with deleted molecules is truly associated with this disorder, then they should be found in myocardial material from ACM patients but not from age-matched controls without any history of alcohol abuse or any signs of heart disease, or at any rate should be present at a systematically higher prevalence in ACM patients than in controls. Secondly, if the products are truly derived from mtDNA, and are not the result of aberrant priming events on nuclear DNA, then the products should hybridize at high stringency with mtDNA probes. Thirdly, most products should fail to hybridize with a probe for the region of the mitochondrial genome between the two origins that is commonly deleted, but conversely should hybridize with probes for the rarely deleted portion of the genome in the region of the rRNA genes. Fourthly, if this is to be applied as a semi-quantitative assay for deleted mtDNAs, the appearance and relative amount of the deleted products should be independent of template quality or concentration. We systematically applied these tests and controls in our analysis. We successfully amplified products that appear to derive from bona fide deleted mtDNAs of the type reported elsewhere. However, the results indicate that the method is severely compromized by technical artefacts. No evidence was obtained indicating an association between mtDNA deletions and ACM, and the appearance and abundance of deletion products was strongly influenced by template DNA concentration or quality. We conclude that this method cannot be used to infer the presence or amount of deleted mtDNA molecules in human tissue samples, unless backed up by a totally independent type of analysis, such as Southern hybridization or DNA sequencing. MATERIALS AND METHODS Patients and tissue samples. The material analyzed consists of 8 cases of alcoholic cardiomyopathy (age 38 – 65 years, mean 47.4 years) and 10 controls (age 34 – 69, mean 51.3 years), who were excluded from having significant heart disease or any known history

508

Vol. 254, No. 2, 1999

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS TABLE 1

Oligonucleotide Primers for LX-PCR

#

Oligonucleotide

Sequence 59 3 39

59 co-ordinate #

Location

OK1H OK2L FR6H OK6L FR7L FR31H FR32L FR51H FR44L

TCGCACCTACGTTCAATATTACAGGCGAAC TAAATAATAGGATGAGGCAGGAATCAAAGACA GGTGCAGCCGCTATTAAAGGTTCGT ATCGGGATGTCCTGATCCAACATCG CCGATCAGGGCGTAGTTTGAGTTTG CTTCCCACAACACTTTCTCGGCCTA GTAAAGGATGCGTAGGGATGGGAGG ACTTCTAGCAAGCCTCGCTAACCTC TTTGGGTTGTGGCTCAGTGTCAGTT

160 161 3012 3013 3698 7178 7840 11836 12553

ncr ncr 16S 16S ND1 COI COII ND4 ND5

See Ref. 42.

of alcohol abuse. Samples of heart muscle were obtained from medico-legal autopsies performed in 1991–1992. DNA extraction. DNA was extracted and purified from frozen cardiac tissue by standard methods at the Department of Forensic Medicine, University of Helsinki. Essentially, the frozen cardiac tissue block (270°C) was cut into thin pieces with a sterile blade in the freezer. The sample was then dissolved in 2 ml of “digestion buffer” (containing 25 mM EDTA, pH 8.0 and 75 mM NaCl). 0.5 mg (0.5 ml) proteinase K and 0.1 ml 20% SDS were then added and the solution was incubated 12 hours at 37°C. The samples were then loaded in Phase Lock Gel II B Light-tubes (5 prime 3 prime, USA). 2 ml of phenol and 2 ml of chloroform/isoamylalcohol was then added and the tubes were shaken well and centrifuged at 2000 rpm for 5 min. The aqueous phase was transferred to a sterile tube, 0.2 ml 2 M KCl was added and the tube was shaken gently. DNA was precipitated by adding 5 ml ice-cold 80% ethanol. The precipitated DNA was collected with a closed Pasteur pipette, air-dried and redissolved in 0.5 ml of TE-buffer. The concentration was measured spectrophotometrically. Long extension PCR (LX-PCR). Two different sets of primers were used to amplify the full mitochondrial genome (see Tables 1 and 2 and Fig. 1). Several subgenomic mtDNA fragments were also amplified in other reactions, using various primer combinations (detailed in Table 2). The enzyme used in all PCR reactions was DyNAzyme EXT DNA polymerase (Finnzymes Oy, Espoo, Finland), which has a weak 39 3 59 proofreading and 59 3 39 exonuclease activity. A biphasic hot start was performed using DynaWax (Finnzymes) and thin-walled 0.6 ml tubes (Robbins Scientific Corp., USA). The 50 ml PCR reaction consisted of 0.6 mM each primer, 500 mM each dNTP, 1 3 EXT buffer supplied with the enzyme (50 mM Tris-HCl, pH 9.0 at 25°C, 15 mM (NH 4) 2SO 4, 0.1% Triton X-100), 2.5 mM MgCl 2, 2% DMSO and 1 unit DNA polymerase, plus template DNAs as indicated in the figure legends. Cycle conditions were as follows. An initial denaturation of 2 min at 93°C was followed by 10 cycles of denaturation for 30 sec at 93°C, annealing for 1 min at 60°C and primer extension for 12 min at 68°C. During the following 20 cycles primer extension time was increased by 20 sec each cycle and an additional extension of 7 min at 68°C was carried out after the last cycle. The PCR products were separated on standard 0.7% agarose gels, stained in ethidium bromide solution and visualized under UV-illumination. Southern blotting. Before blotting, the agarose gel was soaked for 10 min in 0.25 M HCl, twice for 20 min in 0.5 M NaOH/1.5 M NaCl and twice for 20 min in 0.5 M Tris-HCl/1.5 M NaCl, pH 7.4. DNA was transferred to a MagnaCharge nylon membrane (Micron Separation Inc., Westborough, USA) and UV-linked. The filter was pre-wetted prior to hybridization in 4 3 SSC, 0.1% SDS for 20 min and prehybridized 1 hour at 50°C in the hybridization solution, which contained 2.5% dextran sulphate, 0.1 mg/ml denatured, sonicated,

salmon sperm DNA, 0.025 M sodium phosphate buffer, pH 6.8, (PB) all in 4 3 SSC. The hybridization was carried out at 68°C for 15 hours. The blot was washed in solutions prewarmed to 68°C: 20 min in 5 3 SSC, 0.1% SDS, 0.025 M PB, twice for 20 min in 1 3 SSC, 0.1% SDS, 0.025 M PB, and twice for 20 min in 0.2 3 SSC, 0.1% SDS, 0.025 M PB. The hybridization signal was detected by autoradiography. Probes for southern blotting. Amplified fragments of mtDNA were generated by PCR, using a DNA template prepared from one of the control heart muscle samples. 690 bp and 717 bp fragments, receptively from the 16S/ND1 and ND4/ND5 areas, were synthesized using primers FR6H/FR7L and FR51H/FR44L (see Table 1, also Ref. 32). The products were isolated and purified from agarose with Wizard PCR preps (Promega). The fragments were labelled using Oligolabelling Kit (Pharmacia Biotech) and [a- 32P]dCTP (Amersham, 3000 Ci/mmol) as the label.

RESULTS LX-PCR detects ‘deleted mtDNAs’ in both ACM and control heart samples. LX-PCR was carried out using two sets of adjacent, outwardly oriented primers (see Fig. 1, plus Tables 1 and 2) located, respectively, in the non-coding region between LSP and O H (primers OK1H/OK2L) and in the gene for 16S rRNA (primers

TABLE 2

Predicted PCR Products Synthesized by LX-PCR Primer pair

Sequence amplified

Length of product (bp)

Product ID #

OK1H 3 OK2L FR6H 3 OK6L FR6H 3 OK2L FR31H 3 FR7L FR51H 3 FR32L OK1H 3 FR44L FR31H 3 OK2L FR6H 3 FR44L FR51H 3 FR 7L OK1H 3 FR32L FR31H 3 FR44L FR6H 3 FR32L

161–160 3013–3012 3013–160 7178–3698 11836–7840 161–12553 7178–160 3013–12553 11836–3698 161–7840 7178–12553 3013–7840

16569 16569 13716 13089 12573 12392 9551 9540 8431 7679 5415 4827

ns ns 7 ns ns 6 5 4 3 2 1 ns

509

#

As denoted on Fig. 3. ns 5 not shown on Fig. 3.

Vol. 254, No. 2, 1999

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

FIG. 1. Schematic map of the 16.6 kb human mitochondrial genome, indicating the positions of tRNA genes (filled circles) and the protein-coding and rRNA genes that they intersperse, using the standard nomenclature, plus the non-coding region (ncr). Arrowheads indicate the positions of the various primers used for LX-PCR, heavy-strand primers denoted by open arrows, and light-strand primers by filled arrows. For details of primers see Table 1.

FR6H/OK6L). The template DNAs tested were from 8 ACM patient hearts and 10 control hearts, all obtained post mortem. A sample of the data is shown as Fig. 2, which presents an agarose gel and Southern hybridization analysis of the products. All samples tested yielded the 16.6 kb monomeric band representing the

intact mitochondrial genome, as well as other products that were present in varying amounts relative to the full-length band. The pattern of bands generated by each template DNA using a given primer pair was qualitatively similar. Furthermore, some of the bands generated by the two primer pairs were similar in size, notably a prominent band of about 4.0 kb, plus an array of bands in the 5–12 kb range, suggesting that they are not the result of mis-priming, but must represent molecules already present in the template DNA preparation. Other bands, especially those of 3 kb or less, were unique to one or other primer pair. The latter finding is not surprising, since if these short products represent bona fide deleted mtDNA molecules they cannot contain both priming sites unless they each carry two separate deletions, the two primer pairs being 3 kb apart on the genomic map (see Fig. 1). Note also that bands shorter than 6 kb cannot represent molecules in which both replication origins have been retained, unless such molecules also contain more than one deletion. Southern hybridization was carried out sequentially using two probes, one for the junction region of 16S rRNA and ND1 (Fig. 2b), the other for the junction region of ND4 and ND5 (Fig. 2c). The former probe hybridized at high stringency both to the full genomelength PCR product, as well as to most of the other discrete products visible on the gel, the only notable exception being some short products of less than 2 kb. The prominent 4 kb band generated using both primer pairs was detected, indicating that it probably repre-

FIG. 2. LX-PCR using 2 alternate sets of ‘full genome’ primers: OK1H/OK2L (NCR primers), located in the non-coding region (see Table 1 and Fig. 1) and FR6H/OK6L (16S primers), located in the 16S rRNA gene. Template DNAs, 30 ng of which were used in each reaction, were from the hearts of 3 controls (denoted a, b and c) plus 4 ACM patients (denoted d, e, f and g). Panel (a) shows the agarose gel stained with ethidium bromide, and panels (b) and (c) the same gel blot-hybridized with probes for the 16S/ND1 and ND4/ND5 regions, respectively (see text for details of probes). M denotes the marker track (phage lambda HindIII digest, sizes as shown). The prominent 4 kb product generated by both primer pairs and hybridizing with the 16S/ND1, but not the ND4/ND5 probe is shown by a double arrowhead. 510

Vol. 254, No. 2, 1999

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

sents a continuous stretch of mtDNA spanning the control region and rRNA genes, but little else. One much shorter band (approx. 1 kb), generated using the NCR primer pair only, surprisingly hybridized to the 16S/ND1 probe, even though it is much too short to represent a continuous stretch of mtDNA spanning from the priming site to the probe region 3 kb away. If it is a bona fide deleted mtDNA it must carry at least two large deletions, removing most of the genome. The ND4/ND5 probe detected the full genome-length product, but almost no other discrete bands, except some minor products visible only on very long exposure (data not shown). The two probes were of similar specific radioactivity, and the exposures for Fig. 2b and 2c are similar, hence it may be concluded that the discrete products in the 5–12 kb range generated by these primer pairs represent molecules in which the ND4/ ND5 region is deleted. This portion of the genome is commonly deleted in mitochondrial encephalomyopathies characterized both by single, clonal deletions (1), as well as multiple deletions (2), and probably reflects some aspect of the deletion mechanism and/or the presumed dispensability of this region for DNA replication, since it lies between the replication origins for the two strands. Two striking conclusions emerge from these findings. Firstly, the properties of the PCR products detected in these experiments conform with previous reports of deleted mtDNAs in a variety of contexts. Secondly, these deletion products are, in our hands, routinely generated both from control and disease (in this case ACM) templates, with no evidence that they are more prominent in the latter. LX-PCR using different sub-genomic primer pairs generates discordant results. Next we used a series of primer pairs predicted to generate overlapping, subgenomic products from intact mtDNA, as well as shorter products from any deleted molecules in which both priming sites remained intact. The predicted fulllength products were generated in every case, along with a set of shorter products specific to each given primer pair. Once again, these putative deletion products were generated using all templates tested, but in widely varying relative amounts, according to the primers used. A sample of the data is presented in Fig. 3, using one control and one ACM heart DNA. The discordance between primer pairs is illustrated, for example, by the fact that the primers generating PCR product 7 give rise mainly to a diverse set of deletion products that hybridize with the 16S/ND1 probe, whereas the primers that generate the mainly overlapping PCR product 6 give rise almost exclusively to full-length product on hybridization with the same probe. A similar discordance may be noted between the results obtained for the overlapping PCR products 2 and 4. In general, primer pairs generating the longest

fragments from intact mtDNA were the most susceptible to revealing the putative deletion products (Fig. 3 and other data, not shown). As in the case of full genome-length amplification, essentially similar patterns of bands were generated by all templates, with no evidence for any product specific to ACM samples that was not also seen in at least some controls. A few prominent, specific products in the 1–2 kb size range were typically generated from each primer pair. Although the amplified regions overlap, there is little obvious coherence in the patterns of products generated, although their diversity and generally low abundance may mask this. When hybridized with the same probes as used earlier, the presumed deletion products again failed to react with the ND4/ND5 probe (Fig. 3c), except for some prominent, short fragments obtained using a specific primer pair (FR51H/FR7L). Since analogous products were absent from the full-length LX-PCR reactions (Fig. 2c) we may conclude that they were the result of primer-specific mis-priming events. By contrast, the 16S/ND1 probe again reacted with most of the deletion products (Fig. 3b) obtained using the primers spanning the largest distances, although many shorter fragments failed to hybridize. The specificity of the probes is illustrated by their complete failure to hybridize to one or two PCR products in each case, as predicted from their map positions. The overall conclusion from these experiments is that the abundance and even the qualitative patterns of ‘deletion products’ obtained using different primer pairs is enormously variable, even using the same template DNAs, but that mis-priming artefacts are relatively rare, although not negligible. Appearance of deleted mtDNA products depends on template DNA concentration. The variable efficiencies with which deleted mtDNAs were detected by LXPCR, both in a template-dependent fashion in full genome-length amplifications, and also in a primerspecific fashion, crudely related to the length of the predicted, intact mtDNA-derived product in subgenomic amplifications, led us to examine the effects of template concentration on the generation of deletionrelated products. All the reactions shown in Figs. 2 and 3 used 30ng of DNA template in a 50 ml reaction, which gave a substantial, though variable yield of both fulllength and deletion products. We therefore evaluated the effects of using substantially more and less template DNA, by carrying out a series of LX-PCR reactions with different primer pairs, on a five-fold dilution series of various template DNAs. Products were again analyzed by agarose gel electrophoresis and blot hybridization to the 16S/ND1 and ND4/ND5 probes. Essentially the same result was obtained in every case, regardless of the primer pair and template DNA used, as illustrated in Fig. 4 for the OK1H/

511

Vol. 254, No. 2, 1999

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

FIG. 3. LX-PCR using 7 different pairs of ‘sub-genomic’ primers, as detailed in Tables 1 and 2. PCR products 1 through 7 were generated using the following primer combinations: 1—FR31H/FR44L, 2—OK1H.FR32L, 3—FR51H/FR7L, 4 —FR6H/FR44L, 5—FR31H/OK2L, 6 —OK1H/FR44L, 7—FR6H/OK2L. The positions of the primers and the products generated are schematized on the arbitrarily linearized mtDNA map as shown. Products marked with an asterisk are those used as hybridization probes. 30 ng of each of two template DNAs that had also been used in the experiment shown in Fig. 2, i.e. control sample c and ACM sample e, were used. Panel (a) shows part of the agarose gel, stained with ethidium bromide, panels (b) and (c) the same gel blot-hybridized with probes for the 16S/ND1 and ND4/ND5 regions, respectively (see text for details of probes). Positions of markers (phage lambda HindIII digest) were as indicated.

OK2L primer pair and one of the ACM patient DNAs. At very high template DNA concentrations the full genome-length product was completely absent, even on long autoradiographic exposure. By contrast, the shorter, ‘deletion’ products observed previously were prominently synthesized under these conditions. Five-fold dilution of the template DNA restored the pattern seen earlier under the same conditions, with the full-length band now appearing as the main product. A further 5-fold dilution of the template DNA resulted in the almost complete elimination of the deletion products, with a corresponding enhancement of the full-length band. At the next and subsequent dilutions, synthesis failed completely, yielding only primer multimers. DISCUSSION The properties of the LX-PCR products detected here are similar to those identified elsewhere as representing deleted mtDNA molecules. However, we first consider two alternative hypotheses regarding their origin: mis-priming artefacts and nuclear pseudogenes.

As already indicated, the patterns of bands in the 4 –12 kb region synthesized using two completely different ‘full-genome’ primer pairs located 3 kb apart is extremely similar, arguing strongly against mis-priming. The only differences are seen in products that are shorter than the distance between the priming sites, which is expected. Furthermore, the fact that the 16S/ ND1 probe, but not the ND4/ND5 probe, detected the deletion products created using the NCR primer pair also argues against mis-priming, since the two probes regions are located at similar distances from the priming site, implying that any mis-priming would have to be highly directional. Putative deletion products from sub-genomic primer pairs are harder to interpret, since any deleted mtDNA molecule with only one breakpoint within the amplified region will not generate a product. Therefore, non-coherent patterns of products are not necessarily indicative of mis-priming, although some of them could be accounted for thus. Nevertheless, despite the use of 9 different primers in various combinations, only one probable case of mis-priming, although some of them could be accounted for thus. Nevertheless, despite the use of 9 different primers in

512

Vol. 254, No. 2, 1999

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

FIG. 4. LX-PCR using ‘full genome’ primer pair OK1H/OK2L, with template DNA from an ACM patient (denoted f in Fig. 2). Successive five-fold dilutions of template DNA were added to the reactions, each carried out in a final volume of 50 ml, as follows: lane 1—150 ng, lane 2—30 ng, lane 3— 6 ng, lane 4 —1.2 ng, lane 5— 0.25 ng, lane 6 — 0.05 ng (numbers rounded taking account of standard pipetting accuracy). Panel (a) shows the agarose gel stained with ethidium bromide, and panels (b) and (c) part of the same gel blothybridized with probes for the 16S/ND1 and ND4/ND5 regions, respectively (see text for details of probes). M denotes the marker tracks (1 kb ladder and phage lambda HindIII digest, sizes as shown). The 16.6 kb full genome-length product is arrowed.

various combinations, only one probable case of mispriming artefacts was revealed in the study. Nuclear pseudogenes (33–35) are harder to exclude, since some of them may be recently derived from mtDNA, thus capable of hybridizing at high stringency. Many such sequences are amplified from r 0 cells by conventional PCR with commonly used ‘mtDNAspecific’ primers (36), and this has recently confounded attempts to investigate disease-association of mtDNA polymorphisms (37, 38). Nevertheless, LX-PCR using full-genome length primers would only generate a product from such pseudogenes if they were present in tandem arrays. In this case, we would expect ladders of regularly spaced bands, which is not the case in the gels and blots shown in Figs. 2 and 3. Except in the unlikely event that all nuclear pseudogenes detected are present in just two tandem copies, we can therefore exclude them. A further argument is that the regions of the genome represented as nuclear pseudogenes would have to be highly non-random, completely excluding the ND4/ND5 region. However, the ND4 gene was the first to be shown to be present as a nuclear copy in human DNA (35). Such pseudogenes are thus unlikely to be the source of the bands generated in our study. Starting from myocardial DNA of ACM patients, we successfully generated LX-PCR products indicative of the presence of multiple mtDNA deletions. However, they were similarly detected in control heart muscle, and their appearance was highly dependent on template DNA concentration. Had we not ensured that template DNA concentrations were as accurately mea-

sured and as uniform as possible in the amplification reactions, we might easily have concluded, erroneously, that ACM is associated with cardiac mtDNA deletions. LX-PCR can undoubtedly reveal diseaseassociated, deleted mtDNA molecules, when present in amounts detectable by Southern blotting, for example in the muscle of Kearns-Sayre syndrome patients (39, 40), or in various tissues from Pearson syndrome patients (40). The fact that it can also reveal deleted molecules in control samples, where they have never been detected by Southern blotting, hence must be present at extremely low levels, argues against the use of this highly sensitive method as a diagnostic test for the meaningful presence of such species in pathological states. Using similar LX-PCR procedures, we have amplified mtDNA deletion products from template DNAs obtained from a variety of sources, including fetal and adult human liver and cultured human cell-lines (O.A. Kajander, unpublished data). Such products are therefore neither age- nor cell-type specific, and probably represent aberrant molecules present at a low level in all cells, in some ways akin to the sublimons of plant mtDNA (41). Rather small variations in template DNA concentration in LX-PCR reactions can clearly make the difference between failure to synthesize any of the predicted, full-length mtDNA product at all, or else failure to synthesize anything but the full-length product. Clearly the efficiency of synthesis on different template molecules in the starting mixture is enormously influenced either by DNA concentration itself, or by the presence of contaminants which are carried over even in a deproteinized DNA preparation such as we have used. Because of this, we would advise that the use of LX-PCR for diagnostic purposes should be avoided, except where the results can be corroborated using an independent, non-PCR method. Even if the presence of such molecules can be demonstrated by another method, comparisons of the deletion profiles of different samples using this technique are quantitatively unreliable, and should be abandoned in favour of other approaches, even if less sensitive or more laborious. The existing literature on disease- and age-association of multiple mtDNA deletions, where based largely or exclusively on LX-PCR findings, should be critically re-evaluated in the light of our results. ACKNOWLEDGMENTS This work was funded by the Finnish Academy, the Medical Research Fund of Tampere University Hospital, the Yrjo¨ Jahnsson Foundation, the Finnish Foundation of Alcohol Research, and the Pirkanmaa Region Fund of the Finnish Cultural Foundation. We thank Hans Spelbrink for critical reading of the manuscript and Anja Rovio for technical assistance.

513

Vol. 254, No. 2, 1999

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

REFERENCES 1. Holt, I. J., Harding, A. E., Cooper, J. M., Schapira, A. H. V., Toscano, A., Clark, J. B., and Morgan-Hughes, J. A. (1989) Ann. Neurol., 26, 699 –708. 2. Zeviani, M., Servidei, S., Gellera, C., Bertini, E., Di Mauro, S., and Di Donato, S. (1989) Nature, 239, 309 –311. 3. Casademont, J., Barrientos, A., Cardellach, F., Rotig, A., Grau, J. M., Montoya, J., Beltran, B., Cervantes, F., Rozman, C., Estivill, X. et al. (1994) Hum. Mol. Genet., 3, 1945–1949. 4. Bohlega, S., Tanji, K., Santorelli, F. M., Hirano, M., al Jishi, A., and Di Mauro, S. (1996) Neurol., 46, 1329 –1334. 5. Santorelli, F. M., Sciacco, M., Tanji, K., Shanske, S., Vu, T. H., Golzi, V., Griggs, R. C., Mendell, J. R., Hays, A. P., Bertorini, T. E. et al. (1996) Ann. Neurol., 39, 789 –795. 6. Moslemi, A. R., Lindberg, C., and Oldfors, A. (1997) Hum. Mutation, 10, 381–386. 7. Horvath, R., Fu, K., Johns, T., Genge, A., Karpati, G., Shoubridge, E. A. (1998) J. Neuropathol. Exp. Neurol., 57, 396 – 403. 8. Melov, S., Shoffner, J. M., Kaufman, A., and Wallace, D. C. (1995) Nucl. Acids Res., 23, 4122– 4126. 9. Liu, V. W. S., Zhang, C. F., and Nagley, P. (1998) Nucl. Acids Res., 26, 1268 –1275. Neuropathol. Exp. Neurol., 57, 396 – 403. 10. Melov, S., Minerfeld, D., Esposito, L., and Wallace, D. C. (1997) Nucl. Acids Res., 25, 974 –982. 11. Melov, S., Lithgow, G. J., Fischer, D. R., Tedesco, P. M., and Johnson, T. E. (1995) Nucl. Acids Res., 23, 1419 –1425. 12. Ozawa, T., Tanaka, M., Sugiyama, S., Hattori, K., Ito, T., Ohno, K., Takahashi, A., Sato, W., Takada, G., Mayumi, B., et al. (1990) Biochem. Biophys. Res. Commun., 170, 830 – 836. 13. Suomalainen, A., Paetau, A., Leinonen, H., Majander, A., Peltonen, L., Somer, H. (1992) Lancet, 340, 1319 –1320. 14. Li, Y. Y., Hengstenberg, C., and Maisch, B. (1995) Biochem. Biophys. Res. Commun., 210, 211–218. 15. Bobba, A., Giannattasio, S., Pucci, A., Lippolis, R., Camaschella, C., and Marra, E. (1995) Clin. Chim. Acta, 243, 181–189. 16. Marin-Garcia, J., Goldenthal, M. J., Ananthakrishnan, R., Pierpont, M. E. M., Fricker, F. J., Lipshultz, S. E., Perez-Atayde, A. (1996) Cardiovasc. Res., 31, 306 –313. 17. Remes, A. M., Hassinen, I. E., Ikaheimo, M. J., Herva, R., Hirvonen, J., and Peukhurinen, K. J. (1994) J. Am. Coll. Cardiol., 23, 935–942. 18. Casali, C., Santorelli, F. M., Damati, G., Bernucci, P., Debiase, L., and Di Mauro, S. (1995) Biochem. Biophys. Res. Commun., 213, 588 –593. 19. Zeviani, M., Mariotti, C., Antozzi, C., Fratta, G. M., Rustin, P., and Prelle, A. (1995) Muscle & Nerve, S3, 170 –S174. 20. Li, Y. Y., Maisch, B., Rose, M. L., and Hengstenberg, C. (1997) J. Mol. Cell. Cardiol., 29, 699 –2709.

21. Reynier, P., and Malthiery, Y. (1995) Biochem. Biophys. Res. Commun., 217, 59 – 67. 22. Sudarikova, Y. V., Bakeeva, L. E., Tsiplenkova, V. G. (1997) Biochemistry (Moscow), 62, 989 –1002. 23. Preedy, V. R., Siddiq, T., Why, H., and Richardson, P. J. (1994) Alcoholism, 29, 141–147. 24. Cunningham, C. C., and Spach, P. I. (1994) Alcoholism Clin. Exp. Res., 18, 132–137. 25. Siddiq, T., Salisbury, J. R., Richardson, P. J., and Preedy, V. R. (1993) Alcoholism Clin. Exp. Res., 17, 894 – 899. 26. Mikami, K., Sato, S., and Watanabe, T. (1990) Alcohol & Alcoholism, 25, 651– 660. 27. Das, A. M., and Harris, D. A. (1993) Biochim. Biophys. Acta., 1181, 295–299. 28. Marin-Garcia, J., Ananthakrishnan, R., and Goldenthal, M. J. (1995) Alcoholism Clin. Exp. Res., 19, 1463–1466. 29. Kou, S. Y., and Cohen, N. S. (1998) Int. J. Biochem. Cell Biol., 30, 475– 485. 30. Jaatinen, P., Saukko, P., Sarviharju, M., Kiianmaa, K., and Hervonen, A. (1994) Alcohol & Alcoholism, 29, 269 –282. 31. Mansouri, A., Fromenty, B., Berson, A., Robin, M. A., Grimbert, S., Beaugrand, M., Erlinger, S., and Pessayre, D. (1997) J. Hepatol., 27, 96 –102. 32. El Meziane, A., Lehtinen, S. K., Hance, N., Nijtmans, L. G. J., Dunbar, D., Holt, I. J., and Jacobs, H. T. (1998) Nat. Genet. 18, 350 –353. 33. Jacobs, H. T., Posakony, J. W., Grula, J. W., Roberts, J. W., Xin, J. H., Britten, R. J., and Davidson, E. H. (1983) J. Mol. Biol., 165, 609 – 632. 34. Gellissen, G., Bradfield, J. Y., White, B. N., and Wyatt, G. R. (1983) Nature, 301, 631– 634. 35. Fukuda, M., Wakasugi, S., Tsuzuki, T., Nomiyama, H., Shimada, K., and Miyata, T. (1985) J. Mol. Biol. 186, 257–286. 36. Parfait, B., Rustin, P., Munnich, A., and Rotig, A. (1998) Biochem. Biophys. Res. Commun., 247, 57–59. 37. Davis, J. N., and Parker, W. D. (1998) Biochem. Biophys. Res. Commun., 244, 877– 883. 38. Wallace, D. C., Stugard, C., Murdock, D., Schurr, T., and Brown, M. D. (1997) Proc. Natl. Acad. Sci. USA, 94, 14900 –14905. 39. De Coo, I. F. M., Gussinklo, T., Arts, P. J. W., Van Oost, B. A., and Smeets, H. J. M. (1997) J. Neurol. Sci., 149, 37– 40. 40. Kleinle, S., Wiesmann, U., Superti-Furga, A., Krahenbuhl, S., Boltshauser, E., Reichen, J., Liechti-Gallati, S. (1997) Hum. Genet., 100, 643– 650. 41. Small, I. D., Isaac, P. G., and Leaver, C. J. (1987) EMBO J., 6, 865– 869. 42. Anderson, S., Bankier, A. T., Barrell, B. G., de Bruijn, M. H. L., Coulson, A. R., Drouin, J., Eperon, I. C., Nierlich, D. P., Roe, B. A., Sanger, F. et al. (1981) Nature, 290, 457– 465.

514