Effect of Nucleoside Reverse Transcriptase Inhibitors on Mitochondrial DNA Synthesis in Rats and Humans
JAIDS Journal of Acquired Immune Deficiency Syndromes: Volume 37(1) 1 September 2004
Collins, Michelle L PhD; Sondel, Nicole BA; Cesar, Denise BS; Hellerstein, Marc K MD, PhD*
From Department of Nutritional Sciences and Toxicology, University of California at Berkeley, and *Department of Medicine, Division of Endocrinology and Metabolism, University of California at San Francisco, CA.
Nucleoside reverse transcriptase inhibitors (NRTIs) have been hypothesized to inhibit mitochondrial DNA polymerase γ, resulting in decreased mtDNA synthesis and mitochondrial insufficiency in HIV-1- infected patients. mtDNA synthesis was measured directly using a stable isotope mass spectrometric method following NRTI treatment in rodents. 3′-Azido-3′-deoxythymidine (AZT) was added to water (1 mg/mL) and administered ad libitum to female Sprague-Dawley rats for 1-8 weeks (n = 4 or 5 animals/timepoint).
Neither body weight nor food intake was affected by AZT intake. Untreated controls and AZT-treated rats were given 4% 2H2O as drinking water for 2 weeks.
AZT (approximately 100 mg/kg/d) produced a significant (P < 0.05) decrease in cardiac and hindlimb muscle mtDNA fractional synthesis compared with control groups (from 13.8 ± 4.2% to 7.0 ± 4.8% and from 7.6 ± 1.8% to 4.5 ± 0.4%, respectively) after 4 weeks. Cytochrome c oxidase content in hindlimb muscle was also decreased by 50% compared with controls after 4 weeks of AZT treatment (P < 0.07) and a calculated index of absolute mitochondrial biogenesis rate was significantly reduced by week 2 of AZT (P < 0.05) in hindlimb muscle. In preliminary studies, platelet mtDNA enrichments were compared to monocyte nDNA enrichments (used as a marker of a fully turned over tissue) in healthy human subjects. Fractional synthesis of mtDNA in platelets reached 98 ± 3% after 5 weeks of 2H2O labeling.
It is concluded that NRTIs decrease mtDNA synthesis and oxidative enzyme content and thus mitochondrial biogenesis in rodents and that the effects of NRTIs on mitochondrial biogenesis in tissues of HIV-1- infected humans can in principle be measured using this approach.
The enormous benefits of potent antiretroviral regimens for people with HIV-1 infection have been threatened by treatment-associated toxicities. A variety of changes in metabolism and body composition have been reported, in addition to other end-organ toxicities. Mitochondrial toxicity has been hypothesized to explain adverse effects related to nucleoside reverse transcriptase inhibitors (NRTIs).
The biochemical basis of this proposal is that NRTI agents such as 3_-azido-3_-deoxythymidine (AZT) inhibit, in vitro, the activity of DNA polymerase _, which is required for replication of mitochondrial DNA and, thus, for mitochondrial biogenesis. Clinical features of the syndrome associated with NRTI therapy, including lactic acidosis, myopathy, elevated liver function tests, and neuropathic pain, also resemble those of genetic mitochondrial diseases. To test this hypothesized mechanism properly, mitochondrial biogenesis must be measurable. Until recently, however, no simple, direct measurement technique for quantifying mitochondrial biogenesis, and mitochondrial DNA (mtDNA) synthesis in particular, had been available. We recently described a simple method for measuring mtDNA synthesis in vivo. This technique involves administration of 2H2O (heavy water) in drinking water and measurement of deuterium incorporation into the deoxyribose (dR) moiety of deoxyadenosine isolated from mtDNA. This technique allows sensitive detection of small, cumulative differences in mtDNA synthesis.
It should be noted that turnover rates of mitochondria are relevant to the time course of drug- or disease-induced mitochondrial disorders. The normal half-life of tissue mitochondria will influence the time required for organ dysfunction to emerge if a drug completely blocks mtDNA replication, for example. Because tissue mtDNA half-life is normally on the order of many weeks or months, even in rodents, the lag between inhibition of mitochondrial biogenesis and first phenotypic signs of reduced mitochondrial content may be long. Tissues may also exhibit differential susceptibility to toxicities based on intrinsic kinetic differences in mitochondrial turnover. Thus, a reliable method for measuring mtDNA synthesis could be useful in evaluating the role of NRTI-induced inhibition of mtDNA replication in the clinical syndrome.
Here, we apply the 2H2O labeling method to test the hypothesis that NRTI agents inhibit mtDNA synthesis in vivo in rats, and we provide preliminary evidence for applicability of this method in humans.
NRTIs such as AZT are important components of the antiretroviral combination therapies used in the treatment of patients infected with HIV-1. NRTIs have been suggested to contribute to mitchondrial dysfunction in rats and humans. As shown in the current study, AZT-water-treated animals had significant decreases in mtDNA fractional synthesis and in an index of absolute mitochondrial biogenesis rate. The enrichment of DNA in bone marrow cells was not altered. Although previous studies have shown a reduction of mtDNA levels in muscle biopsies from AZT-treated humans, no method has previously been available for measuring reduced mtDNA synthesis in response to NRTI therapies.
The dose of AZT used here in drinking water was 1 mg/mL, or about 100 mg/kg body weight. This dose is somewhat higher than human treatment regimens (8-12 mg/kg/d, or 50-75 mg/kg/d in rats, after correction for the >5- to 6.2-fold higher metabolic and drug disposal rate in rats compared with humans). The 100-mg/kg/d regimen is the dose and route of administration used by previous investigators, however. We also tried a lower dose in diet but observed no effects on mtDNA synthesis. This lack of effect may be due to the lower dose or, less likely, the mixture with food. It should be noted that AZT is not one of the more potent inhibitors of mtDNA synthesis, compared with NRTIs such as stavudine (d4T) or didanosine (ddI). These relative toxicity assessments are largely based on clinical associations and in vitro measurements, however. The availability of a sensitive assay of mtDNA replication in vivo should permit direct comparisons among different agents. The current study represents a proof-of-principle for this approach. Although other tissues were not studied here, it is now possible to measure mtDNA synthesis in liver and other tissues (Thomas and Hellerstein, unpublished observations). It will be of considerable interest to analyze these tissues in future studies.
NRTIs function by acting as competitive inhibitors of the RNA/DNA reverse transcriptase of HIV and cause chain termination in the growing viral DNA chain. Many of the important and treatment-limiting side effects of NRTIs may be related to the effect of these agents on human DNA polymerases, in particular, mitochondrial DNA polymerase _. Depletion of mtDNA during chronic NRTI therapy may lead to cellular respiratory dysfunction and generalized tissue- and drug-specific toxicities, including but not limited to myopathy and lactic acidosis. It has been difficult to prove this hypothesis directly in NRTI-treated patients, however. The present work was also intended as a proof-of-principle for application of this heavy water labeling technique to human mitochondrial biogenesis. Potential applications for the next set of studies might include comparisons among NRTIs, dose-response relationships, time course studies, tissue differences (such as liver, muscle, and adipocytes), drug interactions, interindividual variability, genetic or gender differences, and effects of clinical cofactors. The absence of toxicities for the 2H2O labeling method described here also allows repeat measures in an individual and monitoring of mtDNA synthesis over time.
Our index of the absolute rate of mitochondrial biogenesis might be criticized because it combines a kinetic measurement of mtDNA synthesis with a pool size measure of mitochondrial enzyme content. If mtDNA and enzyme content were to be dissociated, the calculation would not be valid. Previous research in the field of mitochondrial biogenesis has indicated that the ratio of mitochondrial proteins to DNA does not change when mitochondrial mass increases, however. With exercise, for example, the number of mtDNA molecules per mitochondrion increases when mitochondria expand in size and aerobic capacity, so that the ratio of DNA to protein remains unchanged. Moreover, oxidative enzyme content or activity (such as COX activity) has traditionally been used in this field as the measure of functional mitochondrial mass.38 Although it is possible that, under certain pathophysiologic conditions, mitochondrial enzyme and DNA content could be dissociated, we are aware of no evidence in the published literature for an effect of this type. Accordingly, we believe that this index of absolute mitochondrial biogenesis represents a valid calculation. This parameter showed a marked and early response to AZT treatment and may prove to be a sensitive metric of NRTI-induced mitochondrial toxicity.
The human results shown here from blood platelets were promising, although preliminary. The 2H2O protocol was well tolerated, as we observed previously and the mtDNA isolated was free of genomic DNA. Blood platelets are highly accessible, abundant, rich in mitochondria, and free of nuclei, so they may be an ideal tissue for human testing of drugs or genes with putative effects on mtDNA synthesis. Alternatively, direct measurements of muscle, liver, or adipose tissue mtDNA synthesis may be more informative (McComsey et al, unpublished observations).
Management of potential mitochondrial toxicity during NRTI therapy has remained a challenge. The results presented here in experimental animals suggest that early detection of NRTI-associated mitochondrial toxicities may be possible. An early detection method that was applicable in the clinical setting might be used to prevent mitochondrial dysfunction before symptomatic consequences develop.
Currently, the definitive diagnosis of mitochondrial toxicity due to nucleosides or any other etiology requires muscle or liver biopsy and measurement of enzyme activities or ultrastructure. A method that avoided tissue biopsy would be advantageous for routine screening and monitoring. Measurement of mtDNA synthesis in platelets could be a useful tool for monitoring and evaluating mitochondrial toxicity in HIV-infected patients receiving antiretroviral therapy, as well as in patients with other diseases, such as hepatitis and certain cancers, that are also treated with nucleoside analogues. Recently, Cote et al presented a method for quantifying total mtDNA content of peripheral blood cells by PCR as a possible marker of mitochondrial toxicity of NRTI therapy. This method measured only content, not synthesis rate, of mtDNA and therefore is likely to represent a later manifestation of mitochondrial toxicity (ie, a less sensitive test). In addition, the method of Cote et al was not applied to tissues, such as muscle.
In summary, a reliable method for measuring mtDNA synthesis has been developed and confirms that AZT reduces mtDNA synthesis and mitochondrial biogenesis in rats. The results described here are consistent with the hypothesized role of these agents in postulated clinical mitochondrial toxicity syndromes. Carefully designed clinical studies will be required to test this hypothesis definitively.
Fifty-four female 8-month-old Sprague-Dawley rats (210-280 g) from Simonsen (Santa Clara, CA) were studied. Housing was in wire cages, 3 per cage with a 12-hour light/dark cycle. Rats were randomly assigned to the AZT-diet, diet-control, AZT-water, or water-control groups. The standard therapeutic dose of AZT in humans with HIV-1 infection is 600 mg per day (or 8-12 mg/kg/d in 50- to 70-kg individuals). This would be roughly equivalent to 50-75 mg/kg/d in rats. We used a slightly higher dose in drinking water (100 mg/kg/d) because of the relatively short period of exposure (8 weeks) and because this is the dose and route previously used by investigators.8,11 We also tried a lower dose in diet (15 mg/kg/d). Control rats received no AZT and were studied concurrently with AZT-treated animals. Food intake and body weight were measured daily.
2H2O Labeling Technique for Measuring mtDNA Synthesis
The 2H2O labeling protocol was as we have described previously in rodents.10,12 DNA synthesis is measured based on the incorporation of deuterium into C-H bonds of dR in purine deoxyribonucleosides present in DNA. Deuterium label incorporation can only occur during replication of DNA (ie, new DNA synthesis from deoxyribonucleotide-triphosphates). Stable isotope labeled glucose13-16 or heavy water12,16 has been used for measuring DNA synthesis. In either case, the isotopic enrichments of dR from purine deoxyribonucleotides in DNA are determined by gas chromatography/mass spectrometry (GC/MS), after isolation, hydrolysis and derivatization of DNA. Because no radioactivity or genotoxic reagents are involved, this technique is safe for use in humans.
Labeling in rats consisted of an initial intraperitoneal priming bolus of 2H2O (100%) to 2.0-2.5% body water enrichment, based on an estimated 60% body weight as water. This was followed by administration of 4% 2H2O in the drinking water for 2 weeks. The 4% enrichment of 2H2O in drinking water was chosen as a convenient dose that produces sufficient enrichments in biosynthetic products of interest and has no known toxicities. 2H2O (70 and 100%) was purchased commercially from Cambridge Isotopes (Andover, MA). Drinking was ad libitum. Rats were killed by CO2 asphyxiation and tissues were immediately removed.
Isolation of Bone Marrow Cells and Cardiac and Hindlimb Muscle Mitochondria
Bone marrow cells and cardiac (0.5 g) and hindlimb muscle (0.3 g) samples from individual animals were removed immediately after sacrifice, as described previously.10,12 Muscle samples were homogenized as described elsewhere.10,12,17,18 Mitochondria from the homogenate were isolated by density gradient centrifugation.18 Nuclear DNA (nDNA) contamination is removed enzymatically by treatment with DNAse (Sigma, St. Louis, MO). Absence of nDNA contamination in muscle samples was confirmed by polymerase chain reaction (PCR) followed by gel electrophoresis (see below). More than sufficient mtDNA is obtained from 0.3-0.5 g of muscle tissue for GC/MS measurement of mtDNA kinetics in individual animals.
Volunteers were recruited by advertisement and gave written informed consent before enrolling in the study. All protocols were approved by the University of California at San Francisco Committee on Human Research and the University of California at Berkeley Committee for the Protection of Human Subjects. Subjects were healthy, HIV-seronegative subjects with no history of any medical diseases, metabolic disorders, or use of medications with known metabolic effects.
We initiated 2H2O administration to human subjects under observation in a metabolic ward setting, at the General Clinical Research Center of the San Francisco General Hospital, as described previously.10,12 This was done to avoid the possibility of transient vertigo or dizziness, which has been reported as a rare adverse effect of rapid changes in body water enrichment. The 2H2O administration protocol consisted of 50- to 70-mL doses of 70% 2H2O given every 3-4 hours, for 18-24 hours in the metabolic ward. No subject experienced any adverse symptoms, using this protocol. The study volunteers were subsequently maintained on 50-70 mL daily intake of 2H2O, with a goal of maintaining about 1.5-2% body water enrichment (assuming total body water turnover of roughly 3.5 L/d in healthy, ambulatory subjects).12 Subjects underwent blood draws at weeks 1-9 for collection of platelets and monocytes.
Isolation of mtDNA From Human Platelets
Whole blood (10 mL) from patients was collected in 2 separate anticoagulated (heparinized) tubes (one for platelets and a second 10-mL tube for monocyte isolation; see below). Blood from the first tube was centrifuged and the leukocyte upper layer was removed. This platelet rich layer was then centrifuged, resulting in a leukocyte upper layer and a platelet pellet. The platelet pellet was removed. Although platelets are anucleate and should contain no nDNA,19 to be certain, we incubated platelets with DNAse. The cell membrane was disrupted with a lysis buffer AL (Qiagen, Valencia, CA) prior to incubation, to allow DNAse access to any nDNA present. Absence of nDNA contamination in platelet mtDNA samples was evaluated quantitatively by use of real-time PCR as described previously.20 Electron microscopy of the platelet fraction confirmed pure platelets, uncontaminated by erythrocytes or other leukocytes.
Isolation of DNA From Blood Monocytes
Blood monocytes were isolated from Ficoll-Hypaque gradient isolated peripheral blood mononuclear cells, by use of anti-CD14 magnetic beads (Miltenyi Biotec, Inc., Auburn, CA). The monocyte fraction was resuspended in 200 _L phosphate-buffered saline and nDNA was isolated as described previously.12 Monocytes are used as a nearly completely turned over tissue, for calculation of fractional DNA synthesis in a tissue of interest.
Real-time quantitative PCR analysis was performed with an Applied Biosystems (Foster City, USA) 5700 Sequence Detector. The Alu TaqMan system consisted of the amplification primers, AluF 5_-GGAGGCTGAGGCAGGAGAA-3_ and AluR 5_- ATCTCGGCTCACTGCAACCT-3_, and a fluorescent probe, 5_-(FAM)CGCCTCCCGGGTTCAAGCG-3_.
TaqMan amplification reactions were set up in a reaction volume of 25 _L by use of the Applied Biosystems TaqMan Universal PCR Mastermix. PCR primers and TaqMan probes were synthesized by Applied Biosystems. Each reaction contained 1_ PCR Mastermix, 900 nmol of each primer, and 250 nmol for the Alu probe.21 Twenty ng of mtDNA was used for TaqMan reactions. Human Placenta DNA (Sigma) was used as a positive control for mtDNA purification. Quantitative analysis of nDNA contamination from mtDNA preparations was by comparison to serial dilutions of human placenta DNA standards.
Western Blotting of Cytochrome c Oxidase
Hindlimb muscle samples were collected from the week 4 groups and homogenized in 5 volumes of 0.1% Triton-X [polyoxyethylene (10) isoctylphenyl ether] in 0.1 M tromethamine hydrochloride (TRIS) buffer, ph 8.35. A small quantity of dye (5 _L) containing 0.5% methyl green in 30% sucrose was added to each sample to provide a visual indication of migration. Samples (10 _L per lane) were loaded on to sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Gels were processed at 200 V for 1 hour and transferred to 10% nitrocellulose membranes. Proteins were blocked in TBS (TRIS-buffered saline)-milk buffer (50 mM TRIS-tris(hydroxymethyl) aminomethane hydrochloride 5% nonfat dry milk, and 170 mM NaCl; pH 7.5) for 1 hour. Membranes were then incubated in fresh buffer containing primary monoclonal antibody against mouse cytochrome c oxidase subunit IV and actin (internal loading control) for 1 hour. Membranes were rinsed with TBS and incubated with the secondary antibody (goat antimouse Ig horseradish peroxidase) for 1 hour. Blots were developed and relative optical densities were determined with a Bio Image Advanced Quantifier (Bio Image Systems, Jackson, MI).
Kinetic Measurements and Calculations Isolation of DNA and Deoxyadenosine
Bone marrow nDNA was isolated by use of a Qiamp column (Qiagen) as described previously. mtDNA was also isolated from the mitochondrial fraction of cardiac muscle and hindlimb muscle and from lysed platelets using the Qiagen Qiamp Kit. mtDNA and nDNA were hydrolyzed enzymatically to free deoxyribonucleosides as described previously. In brief, an LC18 SPE column (Supelco, Bellefone, PA) was used to separate deoxyadenosine (dA) from the other deoxyribonucleosides. The column was washed with 100% methanol (2 mL) and water (2 mL). The hydrolyzed DNA sample was then added to the column and nucleosides other than dA were eluted with an H2O wash (5 mL). The dA was then eluted with 50% methanol (1 mL) as previously described.
Derivatization of dA and GC/MS Analysis
The dR moiety of dA was analyzed by GC/MS, after conversion to its pentane-tetra-acetate derivative, as described elsewhere. The isotopic enrichment of dR was determined by GC/MS (m/z 245 and 246, representing M0 and M1 masses, respectively). There is no exchange between solvent water or other sample matrix protons and hydrogen atoms in C-H bonds of dR in DNA or in free deoxyribonucleosides. Also, the derivative that was analyzed contains only the dR moiety, not the base portion, of purine deoxyribonucleosides,23 so label incorporation into the base moiety is not a confounding factor.
Bone marrow dA enrichments are used as a comparison (denominator) tissue, representing asymptotic labeling values12,13,23 in rat studies, whereas monocytes were used as the comparison value for human studies.12 Unlabeled (natural abundance) dA standards were analyzed concurrently in each run to establish the dependence of measured isotopic ratios on the amount of sample injected (abundance sensitivity). This dependence can be characterized by plotting the abundance of the parent M+0 ion (m/z 245) vs. the ratio of M+1 to M+0 plus M+1 ions [246/(245+246)]. A linear regression of the ratio vs. M0 abundance was calculated, as described previously. The regression line was then used to calculate the natural abundance ratio at any particular M0abundance, for calculations of excess abundances (enrichments) in samples.
Fractional synthesis rate of DNA was calculated by use of the precursor-product relationship. The central principle behind the mathematics of the precursor-product relationship is that the isotopic enrichment of a product derived exclusively from a precursor pool will approach the isotopic enrichment of the precursor pool, with the shape of an exponential curve. Under conditions of steady state in pool size, the fractional synthesis rate represents the fractional removal or replacement rate (ie, assuming that every molecule produced must be balanced by a molecule destroyed). The replacement constant (k) and half-life (t1⁄2) of mtDNA after 2H2O labeling could then be calculated as described previously.
An index of the absolute rate of mitochondrial biogenesis was generated by multiplying fractional mtDNA synthesis times an estimate of tissue mitochondrial pool size. Recovery efficiency of mitochondria by centrifugation from tissues for mtDNA isolation may vary and can not be easily assessed. Because the content of cytochrome c oxidase is measured directly on tissue extracts and is corrected for actin (as an internal standard), cytochrome c oxidase content was used as a measure of tissue mitochondrial mass. This calculation assumes that the ratio of mtDNA to oxidative enzyme content remains relatively constant within mitochondria in a tissue, so that the pool size of mtDNA is reflected by the pool size of cytochrome c oxidase (see below for discussion).