Sensitive Assay for Mitochondrial DNA Polymerase - CiteSeerX

Clinical Chemistry 45:10 1725–1733 (1999)

Molecular Diagnostics and Genetics

Sensitive Assay for Mitochondrial DNA Polymerase g Robert K. Naviaux,1* David Markusic,2 Bruce A. Barshop,3 William L. Nyhan,3 and Richard H. Haas2,3 Background: The mitochondrial DNA polymerase g is the principal polymerase required for mitochondrial DNA replication. Primary or secondary deficiencies in the activity of DNA polymerase g may lead to mitochondrial DNA depletion. We describe a sensitive and robust clinical assay for this enzyme. Methods: The assay was performed on mitochondria isolated from skeletal muscle biopsies. High-molecular weight polynucleotide reaction products were captured on ion-exchange paper, examined qualitatively by autoradiography, and quantified by scintillation counting. Results: Kinetic analysis of DNA polymerase g by this method showed a Km for dTTP of 1.43 mmol/L and a Ki for azidothymidine triphosphate of 0.861 mmol/L. The assay was linear from 0.1 to 2 mg of mitochondrial protein. The detection limit was 30 units (30 fmol dTMP incorporated in 30 min). The linear dynamic range was three orders of magnitude; 30 –30 000 units. Imprecision (CV) was 6.4% within day and 12% between days. Application of this assay to a mixed population of 38 patients referred for evaluation of mitochondrial disease revealed a distribution with a range of 0 –2506 U/mg, reflecting extensive biologic variation among patients with neuromuscular disease. Conclusion: This assay provides a useful adjunct to current laboratory methods for the evaluation of patients with suspected mitochondrial DNA depletion syndromes. Mitochondrial DNA (mtDNA)4 replication is a complex process involving .20 proteins organized along the inner mitochondrial membrane as a multienzyme complex called the mtDNA replisome, or replication factory (1–3 ). A principal component of the mtDNA replisome is the mtDNA polymerase g. This enzyme is found as an a,b

heterodimer and as an a monomer associated with four other unidentified cellular proteins (4 ). Both a and b subunits have been cloned (5–7 ). The a subunit is catalytic and contains both the polymerase and the 39-to-59 proofreading exonuclease activities. It is 140 000 daltons in size. The b subunit is 43 000 daltons and of unknown function. The substrate virtuosity displayed by DNA polymerase g is unique among eukaryotic polymerases. The mitochondrial enzyme is an effective reverse transcriptase (RNA-directed DNA polymerase) when measured on homopolymeric RNA templates. It is also an effective DNA polymerase when directed by single-stranded homopolymeric DNA, single-stranded heteropolymeric DNA, or activated double-stranded DNA templates (8 ). The experimental choice of template radically influences both the measured kinetic performance of the enzyme and its response to specific inhibitors, such as azidothymidine (AZT) (9 ). The most sensitive assays developed to date have exploited the RNA-directed DNA polymerase activity of the enzyme. These assays are typically ;10- to 50-fold more sensitive than conventional assays, which are performed with activated double-stranded DNA templates (9 ). Mitochondrial disorders are among the most recently recognized groups of diseases known to medicine. They affect both children and adults and are particularly difficult to diagnose because of a broad spectrum of phenotypic presentations (10, 11 ). Muscle biopsy is often obtained in the course of the diagnostic evaluation of patients with suspected mitochondrial disease. Several inherited forms of mtDNA depletion have been described (12–14 ). Depletion of mtDNA may also result from AZT (15 ) or fialuridine (16 ) toxicity. Deficiency in any one of the 20 or so components of the mtDNA replisome could theoretically lead to mtDNA depletion, and we have

The Mitochondrial and Metabolic Disease Center, University of California, San Diego, School of Medicine, Departments of 1 Medicine, 2 Neurosciences, and 3 Pediatrics, 200 West Arbor Dr., San Diego, CA 92103-8467. *Author for correspondence. Fax 619-543-7868; e-mail [email protected]. Received June 1, 1999; accepted July 19, 1999.

4 Nonstandard abbreviations: mtDNA, mitochondrial DNA; AZT, azidothymidine; CS, citrate synthase; DTT, dithiothreitol; cpm, counts per minute; AZT-TP, azidothymidine triphosphate; and ddTTP, dideoxythymidine triphosphate.

© 1999 American Association for Clinical Chemistry

1725

1726

Naviaux et al.: Mitochondrial DNA Polymerase Assay

reported a patient in whom mtDNA depletion was associated with deficient activity of mtDNA polymerase g (17 ). The development of a sensitive and reliable clinical assay for DNA polymerase g may be useful in clinical practice and in the elucidation of mechanisms underlying the mtDNA depletion syndromes.

Materials and Methods isolation of mitochondria Mitochondria were isolated from 200 – 800 mg of fresh (unfrozen) human anterior quadriceps (vastus intermedius) skeletal muscle by a modification of the method of Hatefi et al. (18 ). Briefly, this method utilized differential centrifugation of muscle homogenate at 800g to sediment cellular nuclei, and at 10 000g to sediment mitochondria and permit the removal of nonmitochondrial cytoplasmic proteins, RNAs, nucleotides, and small organic molecules by serial washes. After isolation, mitochondria were resuspended to an approximate concentration of 1–2 g/L in 0.25 mol/L sucrose and frozen in liquid nitrogen or at 280 °C until assayed. A large-scale preparation of beef heart mitochondria or human muscle mitochondria was prepared and frozen at 2 g/L in multiple 20-mL aliquots for use as an internal control.

protein and citrate synthase determinations Mitochondrial protein was determined from thawed samples by a small-scale modification of the Lowry method (Bio-Rad kit no. 500-0112). This required ;20 mg of mitochondrial protein. Mitochondrial yield was quantified by measurement of the mitochondrial matrix protein citrate synthase (CS) by the method of Sheperd and Garland (19 ), which required ;5 mg of mitochondrial protein. One unit of CS was defined as 1 pmol/min of acetyl CoA consumed.

dna polymerase assays Approximately 5 mg of thawed mitochondria was lysed and stabilized in an equal volume of glycerol lysis buffer (25 mmol/L Tris, pH 8.0, 50 mmol/L KCl, 20 mL/L Triton X-100, 500 mL/L glycerol), mixed by trituration in a 10- to 20-mL micropipette, brought to a concentration of at most 1 g/L with additional glycerol lysis buffer, and placed on ice. The RNA-directed DNA polymerase assay was a modification of methods described by Goff et al. (20 ). Briefly, four twofold serial dilutions of lysed mitochondria in glycerol dilution buffer (25 mmol/L Tris pH 8.0, 50 mmol/L KCl, 500 mL/L glycerol) containing ;1.0, 0.5, 0.25, and 0.125 g/L of mitochondrial protein were prepared for each sample. One microliter of each dilution was pipetted into the bottom of 25-mL GeNunc module (48-well, 4 3 12 tray; Nunc cat. no. 2-32549) reaction trays at room temperature. An equal volume of glycerol lysis buffer was pipetted into four separate wells and used as the blank. One tube of beef heart or human muscle mitochondria was thawed and used as an internal standard that was included in every assay as a control for

day-to-day variation. Nine microliters of 1.113 reaction buffer was added to each well, and the full 10-mL reaction volume was mixed by rapid micropipetting six times. The GeNunc reaction trays were sealed with an adhesive top (supplied by the manufacturer) and floated in a 37 °C water bath for 30 min. Reactions were stopped by the addition of 5 mL of each reaction onto 1.3-cm squares drawn and numbered with a pencil (conveniently done during the 30-min incubation period) on DE81 anionexchange paper (Whatman; cat. no. 3658 915). The DE81 paper was washed three times for 5 min each in 250 mL of 23 standard saline citrate (0.3 mol/L NaCl, 0.03 mol/L sodium citrate, pH 7.0), once in 250 mL of 950-1000 mL/L ethanol, air-dried, and then analyzed by autoradiography for 1 h at room temperature using BioMax MSTM film and a matched intensifying screen (Kodak). One microliter of 1.113 reaction buffer was spotted in quadruplicate on DE81 paper and used directly, without washing, as an internal control for radioisotopic specific activity and decay. The DE81 paper was then cut into squares containing the high-molecular weight polymerase reaction products and quantified by scintillation counting in 5 mL of Ecolume (ICN). The mean background (determined from quadruplicate blanks) was subtracted from each reaction, and the total dTMP incorporation into high-molecular weight products (radioactive and nonradioactive) was calculated. Optimized conditions were as follows: 50 mmol/L Tris-HCl, pH 8.0, 100 mmol/L NaCl, 2.5 mmol/L KCl, 5 mmol/L MgCl2 and 0.5 mmol/L MnCl2, 10 mmol/L dithiothreitol (DTT), 1 mL/L Triton X-100, 25 mL/L glycerol, 5 mmol/L dTTP, 10 mg/L poly(rA) (Pharmacia; cat. no. 27-4110-02; mean length, 500 nucleotides; Mr 5 1.85 3 105; 54 nmol/L, yielding a nucleotide concentration of 27 mmol/L), 5 mg/L oligo(dT)12–18 (Pharmacia; cat. no. 27-7858-02; Mr 5 4905; 1.0 mmol/L, yielding a nucleotide concentration of 15 mmol/L and a primer-to-template molar ratio of 18.5:1), 500 mCi/mL 32P-a-dTTP (3000 Ci/mmol), 37 °C for 30 min. The final specific activity of the radionucleotide was typically 150 –220 counts/min (cpm) per fmol. In practice, 0.45 mL of 1.113 reaction buffer (enough for 45 reactions) was prepared by combining 100 mL of 53 buffer stock (0.25 mol/L Tris, pH 8.0, 25 mmol/L MgCl2, 5 mL/L Triton X-100, 50 mmol/L DTT, and 25 mmol/L KCl), 100 mL of 53 template stock [50 mg/L poly(rA), 25 mg/L oligo(dT12–18), and 25 mmol/L dTTP], 5 mL of 50 mmol/L MnCl2, 25 mL of 32P-a-dTTP (10 mCi/mL, 3000 Ci/mmol, 0.167 mmol/L final radionuclide concentration), and 220 mL of diethylpyrocarbonate-treated water. The 53 buffer and template stocks could be stored for up to 6 months at 220 °C. MnCl2 could not be added until the day of the assay because of precipitation that occurred in the concentrated stocks. Single-stranded DNA-directed assays were identical to the RNA-directed assays except that 10 mg/L of poly(dA) was substituted for poly(rA). Activated DNA-directed

1727

Clinical Chemistry 45, No. 10, 1999

Fig. 1. DNA polymerase g reaction progress curves. (A), low protein range. Samples of isolated human skeletal muscle mitochondria, containing 0.1 and 0.3 mg of protein, were assayed for 32P-dTMP incorporation at 10 - to 60-min intervals. These reactions contained 10 mmol/L dTTP and 50 mCi/mL 32P-a-dTTP. (B), high protein range. Samples containing 1– 6 mg of mitochondrial protein were assayed as above. The assay was linear between 0 and 30 min.

assays were performed with 2 mg of mitochondrial protein, 100 mg/L activated calf thymus DNA (Worthington), 50 mmol/L Tris-HCl, pH 8.0, 50 mmol/L KCl, 5 mmol/L MgCl2, 0.5 mmol/L MnCl2, 10 mmol/L DTT, 1 mL/L Triton X-100, 5 mmol/L dTTP, 50 mmol/L each of the three dNTPs (dGTP, dATP, and dCTP), and 500 mCi/mL 32P-a-dTTP (3000 Ci/mmol), and incubation at 37 °C for 30 min. Specific activities were expressed in units per microgram of mitochondrial protein. One unit was defined as 1 fmol of nucleotides (total) incorporated into high-molecular weight products in a 30 min reaction at 37 °C. Moloney murine leukemia virus reverse transcriptase was obtained from Life Technologies. The specific activity was 350 000 BRL units/mg (1 BRL unit 5 1 nmol/10 min). This was equivalent to 1.05 3 109 fmol/mg incorporated in a 30 min reaction. AZT-triphosphate (AZT-TP) was obtained from Sierra Bioresearch. Dideoxythymidine triphosphate (ddTTP) was supplied as a 5 mmol/L stabilized solution from Pharmacia (cat. no. 27-2085-01).

kinetic analysis Michaelis-Menton kinetics were analyzed graphically using the methods of Eisenthal-Cornish-Bowden direct and Hanes-Wolf ([S]/v vs [S]) plots (21 ). AZT-TP and ddTTP inhibition kinetics were analyzed graphically using double reciprocal and secondary Kapp plots (22 ).

temperature and pH were 40 °C and 8.0, respectively. Thirty-minute reactions at 37 °C and a pH of 8.0 were adopted as standard conditions for the assay. To verify that the only DNA polymerase activity detectable in the mitochondrial isolates was DNA polymerase g and not DNA polymerase a, b, d, or e from the nucleus, we tested preparations with a variety of inhibitors. Table 1 shows the results of mtDNA polymerase activity measured on three different templates in the presence of several inhibitors. Aphidicolin (0.1 mmol/L) did not inhibit the mitochondrial polymerase on DNA templates but produced a 22% inhibition on RNA templates. Ethidium bromide was a potent inhibitor on both RNA and DNA templates. Actinomycin D inhibited activity only on double-stranded DNA templates. ddTTP and phosphonoformate (foscarnet) were potent inhibitors on all substrates, whereas AZT-TP was a significantly more potent inhibitor when measured on RNA templates. The effects of divalent cations are summarized in Table 2. Equivalent catalytic activity was observed on poly(rA) and poly(dA) templates. The mitochondrial enzyme was eightfold more efficient on these single-stranded temTable 1. Effect of inhibitors on mtDNA polymerase g activity. Templates Inhibitor

statistics Statistical analyses were performed with Stata (Stata Corporation) and Excel 98 (Microsoft) software packages.

Results Reaction progress curves showed that the assay was linear for up to 40 min (Fig. 1) in the range of 0.1–2 mg of mitochondrial protein (data not shown). The optimum

None AZT-TP, 0.1 mmol/L ddTTP, 0.1 mmol/L Aphidicolin, 0.1 mmol/L Foscarnet, 20 mmol/L Actinomycin D, 100 mg/L Ethidium bromide, 100 mg/L EDTA, 10 mmol/L

Poly(rA), %

Activated DNA, %

100 4.9 1.9 73 2.0 107 4.8 0.1

100 48 16 100 9.3 5.7 3.1 1.8

1728


Table 2. Cation preferences of mtDNA polymerase g. Templates Poly(rA) Cation conditions

Mn21 1 Mg21a Mn21 only Mg21 only Mn21 1 Mg21 Mn21 only Mg21 only

dTTP, mmol/L

5 5 5 1 1 1

Poly(dA)

Activated DNA

U

%

U

%

U

%

1105 939 730 596 864 530

100 85 66 100 145 89

942 301 725

100 32 77

138 62 127

100 45 92

a Standard reaction conditions: 50 mmol/L Tris, pH 8.0, 100 mmol/L NaCl, 2.5 mmol/L KCl, 0.5 mmol/L MnCl2, 5 mmol/L MgCl2, 10 mmol/L DTT, 1 mL/L Triton X-100, 25 mL/L glycerol, 5 mmol/L dTTP, 10 mg/L poly(rA), 5 mg/L oligo(dT12–18), 500 mCi/mL 32P-dTTP (3000 Ci/mmol); 37 °C for 30 min.

plates than on activated double-stranded DNA (1105 vs 138 U). Under conditions of limiting dTTP (1 mmol/L), polymerase reaction velocity on an RNA template was greatest with MnCl2 alone (145% 5 864 U). The addition of MgCl2 with MnCl2, or MgCl2 alone, reduced net polymerization under these conditions. However, when dTTP was raised to 5 mmol/L or higher, the highest activities were observed when both MnCl2 (0.5 mmol/L) and MgCl2 (5 mmol/L) were present. The effect of variations in other reaction conditions is summarized in Table 3. NaCl at 100 mmol/L produced maximum activity. Oligo(dT) primer length also affected the reaction rates. Primer sizes less than 9-mer were clearly inferior, whereas 12- to 18-mers and 30-mers were equivalent. We adopted oligo(dT12–18) in the routine assay. We typically isolated 250 mg of mitochondrial protein for every 100 mg of skeletal muscle over 250 mg processed (n 5 10). The efficiency of mitochondrial isolation by the method of Hatefi et al. (18 ) dropped with decreasing biopsy size and could not be used for specimens much smaller than ;250 mg. A typical open muscle biopsy obtained from a child with suspected mitochondrial disease was 600 mg and yielded 1000 mg of mitochondrial protein. The reproducibility of mitochondrial enzyme assays was limited initially by incomplete solubilization of organelles after a freeze-thaw cycle. This led to particulate behavior that arose from mitochondrial fragments, aggregates, and clumps that were not distributed homogeneously in solution. To address this problem, we solubilized the thawed mitochondrial preparations in an equal volume of glycerol lysis buffer consisting of 25 mmol/L Tris, pH 8.0, 500 mL/L glycerol, 50 mmol/L KCl, and 20 mL/L Triton X-100. Once solubilized, 66% of the polymerase activity was present in the soluble fraction of the mitochondrial lysates. This activity remained in solution and was not pelleted by centrifugation for 10 min at 10 000g (Table 3). This activity was stable in glycerol lysis buffer for at least 6 h on ice (Table 3). The effect of multiple freeze-thaw cycles was also tested. There was no significant effect on polymerase activity frozen in 0.25 mol/L sucrose and subjected to three freeze-thaw cycles (data not shown).

Total 32P-dTMP incorporation (as cpm) increased with decreasing total dTTP down to 0.167 mmol/L (100% of dTTP supplied as 32P-a-dTTP, 500 mCi/mL, 3000 Ci/ mmol; Fig. 2A). This produced more intense spots during autoradiography and more counts during scintillation counting, thus increasing assay sensitivity, but also led to an overall decrease in the calculated polymerase reaction velocity, measured as the total (radioactive plus nonradioactive) dTMP incorporated per unit of time. Increasing the concentration of 32P-a-dTTP beyond 500 mCi/mL (1.1 3 107 dpm and 1.67 pmol per 10-mL reaction, 0.167 mmol/L) was costly and unnecessary because assay sensitivity was not a limiting factor. Increasing the total dTTP to 10 mmol/L (100 pmol per 10-mL reaction) decreased radioactive incorporation by competition with unlabeled nucleotide (molar ratio of dTTP to 32P-dTTP was 10 mmol/L to 0.167 mmol/L, or 60:1) but produced an overall increase in polymerase reaction velocity. Formal kinetic analysis of the mitochondrial polymerase in isolates from human skeletal muscle revealed conventional hyperbolic initial velocity curves (Fig. 2A). The Km for dTTP was determined graphically by two methods. A Hanes-Wolf

Table 3. Effect of reaction conditions on mtDNA polymerase g activity. Condition

Percentage

Standard 2NaCl 2NaCl, 1KCl (100 mmol/L) 2Tris 2DTT 2Triton X-100 120 mL/L Triton X-100 120 mmol/L KPO4, pH 8.0 1 mmol/L dTTP 10 mmol/L dTTP pdT6 pdT9 pdT12–18 pdT30 Not sedimented by 10 min at 10 000g After 6 h on ice in lysis buffer

100 18 64 100 89 28 71 72 54 107 7 15 100 91 66 108


1729

Fig. 2. Michaelis-Menten kinetics. DNA polymerase g activity measured under optimized reaction conditions. Reactions contained 500 mCi/mL 32P-a-dTTP (0.167 mmol/L) and increasing concentrations of total dTTP, as indicated. (A), radionucleotide incorporation (cpm) vs reaction velocity (U/mg). (B), Hanes-Wolf plot (substrate/velocity vs substrate) of kinetic data. Samples containing 1 mg of mitochondrial protein were assayed under standard conditions (see Materials and Methods) and increasing concentrations of dTTP. [S], substrate concentration; v, velocity. (C), Eisenthal-Cornish-Bowden direct plot. Data illustrated in (B) were analyzed graphically according to the Eisenthal-Cornish-Bowden method (21 ).

plot produced a result of 1.48 mmol/L for the Km (Fig. 2B). An Eisenthal-Cornish-Bowden direct plot produced a result of 1.43 mmol/L (Fig. 2C). We adopted a dTTP concentration of 5 mmol/L. This was 3.5-fold higher than the Km for dTTP in the assay and produced a final radionucleotide specific activity of 150 –220 cpm/fmol. Under these conditions, the assay had a detection limit of 15–30 units (1 unit equals 1 fmol of dTMP incorporated in 30 min). The intrinsic, linear dynamic range was three orders of magnitude, permitting quantification of polymerase activities up to 30 000 U (2.2 3 106 incorporated cpm, 20% of the total radioactivity) per reaction. These higher limits were tested using purified Moloney murine leukemia virus reverse transcriptase (data not shown). Specific activities of mtDNA polymerase g in muscle isolates did not exceed 1800 U/mg. In practice, the linear dynamic range for mitochondrial isolates from muscle was 20-fold (0.1–2 mg) because of deviations from linearity that occurred as mitochondrial protein input exceeded 2 mg (data not shown). The mitochondrial polymerase was exquisitely sensitive to inhibition by AZT-TP (Ki 5 0.861 mmol/L; Fig. 3). The Ki for ddTTP was determined in a similar fashion and was found to be 0.0456 mmol/L (data not shown). In the lower ranges of inhibitor concentrations, both AZT-TP and ddTTP were competitive inhibitors. However, at

higher concentrations of AZT-TP (.10 mmol/L), the inhibition was mixed and included both competitive and noncompetitive components (data not shown). We routinely performed four serial twofold dilutions

Fig. 3. DNA polymerase g inhibition by AZT-TP. Samples containing 1 mg of mitochondrial protein and the indicated concentrations of dTTP were assayed in the presence of 0, 1, 5, and 10 mmol/L AZT-TP and analyzed by double reciprocal and secondary plots (inset) as described previously (22 ).

1730


Fig. 4. Comparison of regression and quadruplicate methods of analysis. (A), reproducibility of DNA polymerase g activity determined by regression analysis. A single sample of muscle mitochondria was assayed five times by dilution series. The high-molecular weight, polynucleotide reaction products were captured on DE81 anion-exchange paper and viewed qualitatively by autoradiography and then quantitatively by scintillation counting and regression analysis. (B), reproducibility of DNA polymerase g activity determined by quadruplicate measurement. The same sample assayed in panel A by regression was assayed five times by quadruplicate determinations. Specific activity (U/mg) was calculated by dividing the incorporated radioactivity (cpm 2 blank) by the specific activity of the radionucleotide and by the amount of mitochondrial protein (mg) added, as above. (C), the intercept problem. Results of the mitochondrial polymerase assay for a single sample of muscle mitochondria were calculated by regression in the graph on the left and for four different inputs of mitochondrial protein, determined in quadruplicate, on the right.

for each sample and analyzed the results by both autoradiography and scintillation counting (Fig. 4A). These data were used to calculate a best fit, four-point regression line for each sample. The within-day CV was 6.4% (n 5 5; Fig. 4A). The between-day CV was 12% (n 5 11) using replicate samples frozen in small aliquots, thawed and assayed on different days over a period of 6 months. When quadruplicate determinations were used to measure DNA polymerase g activity instead of a regression, the within-day imprecision (CV) was 2.5% (n 5 5; Fig. 4B). However, we found that the accuracy of this method was inferior to regression. The magnitude of this error was greater for more dilute samples of mitochondrial protein. When quadruplicate determinations of samples containing 0.1– 0.4 mg of protein were analyzed, this method systematically overestimated the true activity in some mitochondrial isolates because of the more significant effect that a non-zero y-intercept had on samples contain-

ing a lower absolute amount of polymerase activity. The consequences of the intercept problem are illustrated in Fig. 4C. In this experiment, a single sample of muscle mitochondria was analyzed both by serial, twofold, fourpoint regression and by quadruplicate determinations of four different inputs of mitochondrial protein. Fig. 4C illustrates the same data, analyzed by two different methods. Graphical analysis of the results produced a regression line that had a positive y-intercept that was not removed by subtraction of the experimental (glycerol lysis buffer) blanks. For example, in the sample illustrated on the left in Fig. 4C, the y-intercept was 40 117 cpm. The best fit regression line was y (cpm) 5 116 904x 1 40 117. The polymerase activity for this sample was calculated by dividing the slope by the specific activity of 32P-dTTP in the reaction, which was 188 cpm/fmol in the experiment illustrated. The result was 622 U/mg (116 904 cpm/mg divided by 188 cpm/fmol). The apparent specific activity


Fig. 5. Frequency distribution of DNA polymerase g activities normalized for CS. Mitochondrial polymerase activities were normalized for mitochondrial content by expressing them as a ratio of DNA polymerase g activity (U/mg) to CS activity (U/mg). The range of normalized activities in this population was 0 –2506.

of the sample was also calculated on the basis of quadruplicate determinations (shown on the right in Fig. 4C). These data converged on the true value of 622 U/mg as the absolute activity in the sample increased, but each of the quadruplicate values systematically overestimated the true activity because the intercept was not taken into account. We adopted regression analysis for the standard assay. We applied the assay to the measurement of mtDNA polymerase g in 38 patients (age range, 3 months to 38 years) with neuromuscular disease in whom muscle biopsies were carried out for the evaluation of a suspected mitochondrial cause. This population was chosen because consent for anterior quadriceps muscle biopsies is not easily obtained from age-matched, healthy controls. Twenty-four patients (63%) had confirmed mitochondrial disease. Fourteen patients (37%) had encephalomyopathies or myopathies of unknown cause. The results were normalized for mitochondrial content by expressing them as the ratio of polymerase (in U/mg) and CS (in U/mg) specific activities These were then arrayed as frequency distribution in Fig. 5. The mean, median, mode, and SD were 668, 551, and 490, and 506, respectively. The range was 0 –2 506. The patient with no detectable activity had Alpers syndrome and mtDNA depletion. The case report describing this patient has been published (17 ). The patient with the highest mitochondrial polymerase activity normalized to CS was an 11-year-old boy with profound lactic acidemia of abrupt onset at puberty.

Discussion We report a useful assay for measurement of mtDNA polymerase g activity in mitochondrial isolates obtained from human skeletal muscle. This assay had excellent sensitivity (30 U/mg mitochondrial protein), specificity (absence of nuclear DNA polymerase contamination), and reproducibility (within- and between-day CVs, 6.4% and 12%, respectively). It could be performed on isolated

1731

mitochondria that had been frozen and stored in liquid nitrogen or at 280 °C until sufficient numbers of samples had been accumulated for convenient assay. Typically, ,30 mg of total mitochondrial protein was sufficient for all three required assays: protein concentration (20 mg); CS activity (5 mg); and mitochondrial polymerase activity (5 mg). The effects of 16 factors on measured mtDNA polymerase g activity in skeletal muscle were evaluated: reaction time, total protein, temperature, pH, detergent, Na1, K1, Mg21, Mn21, PO42-, DTT, RNA template, single-stranded DNA template, activated double-stranded DNA template, nucleotide (dTTP) concentration, and primer length. Although it was not possible to explicitly test all combinations (.216) of these factors, a few important interactions were noted. For example, the presence of Mg21 under limiting dTTP concentrations, at or below the Km (1 mmol/L dTTP; Table 2), produced a strong reduction in net polymerase activity. However, when substrate dTTP was increased to 5 mmol/L, the addition of Mg21 stimulated polymerase activity 15% (85% with Mn21 alone, 100% with Mn21 and Mg21; 5 mmol/L dTTP; Table 2). The measured activity in this assay represents a sum of the forward polymerase and “backward” 39-to-59 proofreading exonuclease activities of DNA polymerase g. The exonuclease activity of DNA polymerase g is known to be Mg21 dependent (4 ), whereas the polymerase activity may use either Mn21 or Mg21 (with a preference for Mn21 on RNA templates and Mg21 on DNA templates; Table 2). This combination of catalytic characteristics leads to kinetic stalling of the enzyme under conditions of limiting dTTP (,2–5 mmol/L), during which the backward exonuclease activity predominated. The details of divalent cation preference were also influenced by monovalent cation concentration and ionic strength. Lower ionic strengths (0 –50 mmol/L NaCl) favor exonuclease activity, whereas higher ionic strengths (100 –150 mmol/L NaCl) favor polymerase and inhibit the exonuclease (4, 23 ). The assay was highly specific for mtDNA polymerase g. Cellular nuclei were removed by differential centrifugation before assay, and an RNA template was used to measure polymerase activity. Because cellular nuclei were removed, little or no detectable nuclear DNA polymerase activity was present. This was evidenced by the ethidium bromide-, AZT-TP-, and ddTTP-sensitive activity observed with the assay (Table 1), and the classical Michaelis-Menten kinetics of the enzyme (Fig. 2). The Km for dTTP was 1.43 mmol/L, as determined by the EisenthalCornish-Bowden method (21 ). This value, obtained with relatively crude mitochondrial isolates, corresponded well to published values of 0.4 –2 mmol/L using enzyme preparations that were purified as much as 10 000-fold (24 –26 ). In addition, RNA was used as the template. Under these conditions, even if nuclear polymerase contamination did occur, virtually no DNA polymerase a, d,

1732


or e activity would be detected because these nuclear polymerases are unable to use RNA templates (9, 24, 27 ). DNA polymerase b has good RNA-directed activity, roughly equal to its DNA-directed activity, but it is insensitive to AZT-TP [Ki .1000 mmol/L, (28 )]. The enzyme measured in this assay was exquisitely sensitive to AZT-TP (Ki 5 0.861 mmol/L; Fig. 3) and therefore was not DNA polymerase b. The constellation of catalytic properties exhibited by the enzyme measured in this assay showed that our methodology is highly specific for mtDNA polymerase g and is devoid of detectable contamination with DNA polymerases a, b, d, or e. The physical removal of cellular nuclei and cytoplasm (containing soluble proteins, nucleotides, and mRNA, rRNA, and tRNA) before assay increased the sensitivity, specificity, and reliability of the assay. However, it also led to the unavoidable loss of an unknown fraction of DNA polymerase g that occurs naturally outside of mitochondria in postmitotic cells, such as skeletal muscle. Bona fide DNA polymerase g occurs both in the nucleus (29 ) and in the cytoplasm outside of mitochondria as well as in mitochondria in nonproliferating cells (30 ). In contrast, dividing cells contain DNA polymerase g only in mitochondria (31 ). The physical isolation of mitochondria before assay also reduced the concentration of cellular nucleotides introduced nonspecifically into the reactions, permitting tighter control of dTTP concentrations for kinetic analysis (Fig. 2). The capture of high-molecular weight polynucleotide reaction products on anion-exchange paper (DE81) in this assay helped to avoid the tedious steps of trichloroacetic acid precipitation, collection, and washing of individual spun glass filters that are used routinely in most other polymerase assays (9, 24, 28 ). The simplicity of DE81 paper contributed to improved replicate precision and dramatically reduced the time and labor required to assay four-point dilutions and quadruplicate patient samples. We found that measurement of polymerase activity by regression analysis was significantly more accurate than quadruplicate determinations (Fig. 4). The accuracy of quadruplicate determinations was reduced because of the problem of assay background (non-zero y-intercept) in some samples that was not eliminated by subtraction of the sample buffer blank (Fig. 4C). When results were calculated by regression, the effects of a non-zero yintercept were easily identified and accounted for. The slope of the regression line was the true function of the polymerase activity in the sample. The y-intercept was a function of non-polymerase components in samples that produced a fixed background of radioactivity in the assay. Not all samples displayed non-zero y-intercepts, and the magnitude of this effect varied from sample to sample. It was small in certain mitochondrial isolates (data not shown) but large in others (Fig. 4C). Because it was not possible to predict which samples would show a significant non-zero y-intercept without performing regression

analysis, we routinely adopted regression as a standard part of this assay. Analysis of the frequency histogram shown in Fig. 5 showed that activities of mitochondrial polymerase in this heterogeneous population did not follow a gaussian distribution. The distribution was skewed to the right and revealed gaps that suggested that the distributions were sampled from two or more distinct populations. This was biologically plausible because the patients we studied represented a complex mixture of mitochondrial (63%) and nonmitochondrial (37%) encephalomyopathies and myopathies. The standard deviation was calculated (SD 5 506) but was not particularly meaningful because of the strong departure from a gaussian distribution. The large biologic range revealed by this assay was reflected by the large ratios of population standard deviations to the population means (s/m 5 506/668 5 76%). This biologic variation was 6- to 12-fold greater than the experimental variation of the assay (CV 5 6 –12%), and therefore was real and not an artifact of the assay methodology. When the clinical presentations were correlated with DNA polymerase g activity represented at the extremes of the frequency histograms, we found that patients on both tails of the distribution had very strong, high-mortality phenotypes. The patient with no detectable DNA polymerase g had Alpers syndrome and mtDNA depletion (17 ). The patient with the highest polymerase/CS activity was a well-nourished 11-year-old boy with precocious puberty and abrupt onset of profound lactic acidemia (17 mmol/L), diabetes insipidus, thiamine-resistant wet beriberi (encephalopathy, polyneuropathy, cardiomyopathy), seizure disorder, myopathy, temporal hair graying, and treatment-resistant megakaryocytic leukemia. Both patients died. The assay that we describe combines features from classical methods in virology that are used to measure reverse transcriptase in retroviruses (20 ) with conventional RNA- and DNA-directed DNA polymerase assays used to measure purified fractions of mtDNA polymerase g (4, 23 ). The combination of these methods enabled us to produce a more sensitive and reliable assay for the measurement of mtDNA polymerase g. This was done in clinically relevant samples (0.3– 0.6 g) of human skeletal muscle biopsies submitted as part of the systematic histochemical, ultrastructural, molecular, and biochemical evaluation of patients with suspected mtDNA depletion.

This work was supported by grants from the Lennox Foundation and generous donations from the Tyler Riff Ferguson Medical Research Fund, the Christini Fund, and the People of Cumberland, MD (to R.K.N.), US Public Health Service Grants DK07318 (to W.L.N.) and NS29504 (to R.H.H.), and NIH Grant MO1 RR00827 to the University of California, San Diego, General Clinical Research Center.


References 1. Reddy GPV, Pardee AB. Multienzyme complex for metabolic channeling in mammalian DNA replication. Proc Natl Acad Sci U S A 1980;77:3312– 6. 2. Shadel GS, Clayton DA. Mitochondrial DNA maintenance in vertebrates. Annu Rev Biochem 1997;66:409 –35. 3. Lemon DP, Grossman AD. Localization of bacterial DNA polymerase: evidence for a factory model of replication. Science 1998; 282:1516 –9. 4. Longley MJ, Ropp PA, Lim SE, Copeland WC. Characterization of the native and recombinant catalytic subunit of human DNA polymerase g: identification of residues critical for exonuclease activity and dideoxynucleotide sensitivity. Biochemistry 1998;37: 10529 –39. 5. Ropp PA, Copeland WC. Cloning and characterization of the human mitochondrial DNA polymerase, DNA polymerase g. Genomics 1996;36:449 –58. 6. Lecrenier N, Van Der Bruggen P, Foury F. Mitochondrial DNA polymerases from yeast to man: a new family of polymerases Gene 1997;185:147–52. 7. Wang Y, Farr CL, Kaguni LS. Accessory subunit of mitochondrial DNA polymerase from Drosophila embryos: cloning, molecular analysis, and association in the native enzyme. J Biol Chem 1997;272:13640 – 6. 8. Fry M, Loeb LA. Animal cell DNA polymerases. Boca Raton, FL: CRC Press, 1986:91–101. 9. Lewis W, Meyer RR, Simpson JF, Colacino JM, Perrino FW. Mammalian DNA polymerases a, b, g, d, e incorporate fialuridine (FIAU) monophosphate into DNA and are inhibited competitively by FIAU triphosphate. Biochemistry 1994;33:14620 – 4. 10. De Vivo DC. The expanding clinical spectrum of mitochondrial diseases. Brain Develop 1993;15:1–22. 11. Kerr DS. Protean manifestations of mitochondrial diseases: a minireview. J Pediatr Hematol Oncol 1997;19:279 – 86. 12. Moraes CT, Shanske S, Tritschler H-J, Aprille JR, Andreetta F, Bonilla E, et al. mtDNA depletion with variable tissue expression: a novel genetic abnormality in mitochondrial diseases. Am J Hum Genet 1991;48:492–501. 13. Ricci E, Moraes CT, Servidei S, Tonali P, Bonilla E, DiMauro S. Disorders associated with depletion of mitochondrial DNA. Brain Pathol 1992;2:141–7. 14. Poulton J, Sewry C, Potter CG, Bougeron T, Chretien D, Wijburg FA, et al. Variation in mitochondrial DNA levels in muscle from normal controls. Is depletion of mtDNA in patients with mitochondrial myopathy a distinct clinical syndrome? J Inherit Metab Dis 1995; 18:4 –20. 15. Dalakas MC, Illa I, Pezeshkpour GH, Laukaitis JP, Cohen B, Griffin JL. Mitochondrial myopathy caused by long-term zidovudine therapy. New Engl J Med 1990;322:1098 –105. 16. Mckenzie R, Fried MW, Sallie R, Conjeevaram H, Di Bisceglie AM, Park Y, et al. Hepatic failure and lactic acidosis due to fialuridine

17.

18.

19. 20.

21. 22.

23.

24.

25.

26.

27.

28.

29.

30. 31.

1733

(FIAU), an investigational nucleoside analogue for chronic hepatitis B. New Engl J Med 1995;333:1099 –105. Naviaux RK, Nyhan WL, Barshop BA, Poulton J, Markusic D, Karpinski NC, Haas RH. Mitochondrial DNA polymerase g deficiency and mitochondrial DNA depletion in a child with Alpers syndrome. Ann Neurol 1999;45:54 – 8. Hatefi Y, Jurtshuk P, Haavik AG. Studies on the electron transport system. 32. Respiratory control in beef heart mitochondria. Arch Biochem Biophys 1961;94:148 –55. Sheperd D, Garland PB. Citrate synthetase from rat liver. Methods Enzymol 1969;13:11– 6. Goff S, Traktman P, Baltimore D. Isolation and properties of Moloney murine leukemia virus mutants: use of rapid assay for release of virion reverse transcriptase. J Virol 1981;38:239 – 48. Cornish-Bowden A. Fundamentals of enzyme kinetics, revised ed. London, UK: Portland Press, 1995:30 –7. Lewis W, Levine ES, Griniuviene B, Tankersley KO, Colacino JM, Sommadossi JP, et al. Fialuridine and its metabolites inhibit DNA polymerase g at sites of multiple adjacent analog incorporation, decrease mtDNA abundance, and cause mitochondrial structural defects in cultured hepatoblasts. Proc Natl Acad Sci U S A 1996;93:3592–7. Gray H, Wong TW. Purification and identification of subunit structure of the human mitochondrial DNA polymerase. J Biol Chem 1992;267:5835– 41. Mosbaugh DW. Purification and characterization of porcine liver DNA polymerase g: utilization of dUTP and dTTP during in vitro DNA synthesis. Nucleic Acids Res 1988;16:5645–59. Lewis W, Simpson JF, Meyer RR. Cardiac mitochondrial DNA polymerase g is inhibited competitively and noncompetitively by phosphorylated zidovudine. Circ Res 1994;74:344 – 8. Izuta S, Saneyoshi M, Sakurai T, Suzuki M, Kojima K, Yoshida S. The 59-triphosphates of 39-azido-39deoxythymidine, 29,39dideoxynucleosides inhibit DNA polymerase g by different mechanisms. Biochem Biophys Res Commun 1991;179:776 – 83. Goulian M, Grimm SL. Three cytoplasmic DNA polymerases that utilize poly(rA)-oligo(dT). Biochem Biophys Res Commun 1990; 170:627–34. Copeland WC, Chen MS, Wang TSF. Human DNA polymerases a, b are able to incorporate anti-HIV deoxynucleotides into DNA. J Biol Chem 1992;267:21459 – 64. Hubscher U, Kuenzle CC, Spadari S. Identity of DNA polymerase g from synaptosomal mitochondria and rat-brain nuclei. Eur J Biochem 1977;81:249 –58. Bolden A, Noy GP, Weissbach A. DNA polymerase of mitochondria is a g-polymerase. J Biol Chem 1977;252:3351– 6. Davis AF, Ropp PA, Clayton DA, Copland WC. Mitochondrial DNA polymerase g is expressed and translated in the absence of mitochondrial DNA maintenance and replication. Nucleic Acids Res 1996;24:2753–9.