Potential conflict of interest: Dr. Walker is a consultant for Pharma Nord.
Zalcitabine is an antiretroviral nucleoside analogue that exhibits long-term toxicity to hepatocytes by interfering with the replication of mitochondrial DNA (mtDNA). Uridine antagonizes this effect in vitro. In the present study we investigate the mechanisms of zalcitabine-induced hepatotoxicity in mice and explore therapeutic outcomes with oral uridine supplementation. BalbC mice (7 weeks of age, 9 mice in each group) were fed 0.36 mg/kg/d of zalcitabine (corresponding to human dosing adapted for body surface), or 13 mg/kg/d of zalcitabine. Both zalcitabine groups were treated with or without Mitocnol (0.34 g/kg/d), a dietary supplement with high bioavailability of uridine. Liver histology and mitochondrial functions were assessed after 15 weeks. One mouse exposed to high dose zalcitabine died at 19 weeks of age. Zalcitabine induced a dose dependent microvesicular steatohepatitis with abundant mitochondria. The organelles were enlarged and contained disrupted cristae. Terminal transferase dUTP nick end labeling (TUNEL) assays showed frequent hepatocyte apoptosis. mtDNA was depleted in liver tissue, cytochrome c-oxidase but not succinate dehydrogenase activities were decreased, superoxide and malondialdehyde were elevated. The expression of COX I, an mtDNA-encoded respiratory chain subunit was reduced, whereas COX IV, a nucleus-encoded subunit was preserved. Uridine supplementation normalized or attenuated all toxic abnormalities in both zalcitabine groups, but had no effects when given without zalcitabine. Uridine supplementation was without apparent side effects. Conclusion: Zalcitabine induces mtDNA-depletion in murine liver with consequent respiratory chain dysfunction, up-regulated synthesis of reactive oxygen species and microvesicular steatohepatitis. Uridine supplementation attenuates this mitochondrial hepatotoxicity without apparent intrinsic effects. (HEPATOLOGY 2007;45:72–79.)
Mitochondrial hepatotoxicity is a complication in HIV patients taking nucleoside analogue reverse transcriptase inhibitors (NRTI) as part of their antiretroviral treatment. Individuals may experience hepatomegaly, steatosis, steatohepatitis and even acute liver failure with life-threatening lactic acidosis.1–6 NRTI undergo intracellular triphosphorylation and then inhibit polymerase-gamma, the intramitochondrial enzyme which is responsible for the replication of mitochondrial DNA (mtDNA).7 Preclinical studies have demonstrated that NRTI exposure to hepatocytes cause mtDNA depletion and intracellular steatosis.8, 9 More recently, low hepatic mtDNA-levels due to prolonged treatment with dideoxynucleoside NRTI (didanosine, stavudine and zalcitabine) were confirmed in HIV/HCV coinfected humans.10 Acute catastrophic hepatotoxicity was also observed in humans treated with fialuridine, an investigational nucleoside analogue for chronic hepatitis B and potent inhibitor of mtDNA synthesis.11, 12 Treatment options for these forms of mitochondrial hepatotoxicity are limited. It is commonly recommended to discontinue the responsible agent and to switch to less toxic antiretrovirals. The recovery from mitochondrial toxicity however may have a slow offset. Switching HIV drugs may also encounter compliance problems and carry risks with regard to the potential development of HIV-resistance or side effects.
We have recently discovered that uridine prevents and even reverses mitochondrial toxicity in hepatocytes exposed to pyrimidine NRTI.13 The beneficial effects of uridine were dose-dependent and observed in concentrations between 50-200 μM.
The mechanism of the protective effects of uridine has not been fully delineated.14 Mitochondrial toxicity and related respiratory chain dysfunction are thought to cause a deficiency of intracellular pyrimidines, because a normal electron flux through the respiratory chain is required for the activity of dihydroorotate dehydrogenase (DHODH), an enzyme which is essential for pyrimidine de novo synthesis.15 There is evidence that low intracellular pyrimidine levels can trigger cell cycle arrest and apoptosis.16, 17 Intramitochondrial pyrimidine deficiency may also aggravate mtDNA-depletion by allowing the triphosphorylated pyrimidine analogues to compete more efficiently with their natural pyrimidine counterparts at polymerase-gamma. Uridine can replenish pyrimidine pools by being salvaged into pyrimidines distal from DHODH.14
The unknown in vivo effects of uridine on mitochondrial hepatotoxicity were the aim of our study. We established a new mouse model in which we used zalcitabine as the strongest polymerase-gamma inhibitor among the pyrimidine NRTI.7, 8 As a source of uridine, we used Mitocnol (NucleomaxX), a dietary supplement with a high bioavailability of uridine.18
After approval by the animal ethics board, female BALB/C mice were purchased at Charles River, Germany. The rodents received humane care according to the NIH guidelines (http://grants.nih.gov/grants/olaw/olaw.htm), were housed at a normal night-day rhythm, and were fed a normal mouse chow ad libitum (SSniff R/M-H, Spezialdiäten, Germany). At 7 weeks of age, the mice were divided into six groups of nine animals each. Controls consisted of mice without any treatment (group A). Group B mice received 340 mg/kg/d of Mitocnol (Pharma Nord, Vojens, Denmark) in the drinking water. Groups C and D received zalcitabine (0.36 mg/kg/d) in the drinking water. This daily dose corresponds to human dosage adjusted to body area and was calculated on the basis of a daily liquid consumption of 5 ml.19 Further mice (groups E and F) received a higher dose of zalcitabine (13 mg/kg/d). Groups D and F were co-treated with 340 mg/kg/d of Mitocnol in the drinking water, whereas groups C and E received no Mitocnol.
Observations for fluid consumption, clinical signs and mortality were carried out daily; body weights were recorded weekly. All animals were killed by cervical dislocation at 22 weeks of age, immediately prior to organ collection and postmortem examination. Livers were weighted, snap-frozen and cryopreserved in liquid nitrogen until subsequent analysis. Aliquots were fixed in glutaraldehyde (3%) for subsequent electron microscopy.
Histopathology, Mitochondrial Ultrastructure and Apoptosis.
The degree of liver fibrosis, steatosis and necroinflammatory activity was scored with haematoxylin and eosin using the modified histological activity index (HAI) and the nonalcoholic steatohepatitis (NASH) scores.20, 21 The percentage of hepatocytes displaying microvesicular steatosis was also assessed with oil red O. The evaluating person was blinded to the group status. Two randomly selected liver samples from each group were examined with electron microscopy.22
For the detection of apoptosis, terminal transferase dUTP nick end labeling (TUNEL) assays were applied on two randomly selected liver sections (6 μm) from each group. Apoptotic nuclei were identified with the ChromaTidelexaFluor 488-5-dUTP (Molecular Probes, Eugene, OR), and the Terminal Transferase Kit (Roche, Mannheim, Germany). Slides were counterstained with Hoechst 33342 for 10 minutes (37°C) and embedded with Vectashield (Vector Laboratories, Burlingame, CA) mounting medium according to the manufacturer's instructions. Slides were examined in a blinded fashion with a confocal microscope (Zeiss, Oberkochen, Germany) at 488 nm and 350 nm.
Amount of Hepatic Lipids.
Lipids were freshly extracted from the livers using methanol/chloroform/water according to the Bligh-Dyer method and quantified spectrophotometrically using a sulfo-phospho-vanillin reaction on lipid standards (Sigma, Taufkirchen, Germany) as described.23
The enzymatic activity of cytochrome c-oxidase (COX), succinate dehydrogenase (SDH) and citrate synthase (CS) were measured spectrophotometrically in freshly prepared tissue extracts.22 COX is a multisubunit respiratory chain complex which is encoded by both nuclear DNA (nDNA) and mtDNA, whereas SDH is a respiratory chain enzyme which is encoded entirely by nDNA. CS is a nDNA-encoded component of the Krebs-cycle and located in the mitochondrial matrix.
MtDNA-Encoded Respiratory Chain Protein.
The mtDNA encoded subunit I of cytochrome c-oxidase (COX I) was quantified by immunoblot.8 The COX I signal was normalized to the expression of the subunit IV of cytochrome c-oxidase (COX IV), which is encoded by nDNA. Blots were also probed with a third antibody (Research Diagnostics Inc., Flanders, NJ) against glycerol aldehyde phosphate dehydrogenase (GAPDH), an enzyme which is entirely encoded in the nucleus.
MtDNA Copy Number.
Total DNA was extracted with the QIAamp DNA isolation kit (Qiagen, Hilden, Germany). MtDNA and nDNA copy numbers were determined by quantitative PCR.24 Briefly, mtDNA was amplified between nucleotide positions 2469 and 2542 and was quantified with a FAM-fluorophore labeled probe. For the detection of nDNA we selected GAPDH between nucleotide positions 494 and 671 and used a HEX-fluorophore labeled probe.
Each 25 μl reaction contained 20 ng of genomic DNA, 100 nM probe, 200 nM primers and Taq-man absolute Master Mix (Abgene, Hamburg, Germany). Amplifications of mitochondrial and nuclear products were performed as triplicates. Absolute mtDNA and nDNA copy numbers were calculated using serial dilutions of plasmids with known copy numbers.
Lipid Peroxidation and Superoxide Production.
Malondialdehyde (MDA) is one of the end products of lipid peroxidation and an indicator of free radical production and oxidative stress. MDA was spectrophotometrically quantified in tissues with an assay for thiobarbituric acid reactive material.25 Superoxide production was measured in situ on transverse tissue sections with the oxidative fluorescent dye dihydroethidium (Sigma, Taufkirchen, Germany).22 The intensity of the fluorescence was quantified using Scion Image (Scion Corp).
Group means were compared ANOVA and Kruskal-Wallis ANOVA on the ranks, followed by unpaired t tests or Wilcoxon Mann-Whitney tests, as appropriate. Correlations were computed by nonlinear exponential regression analysis. Graphics and calculations were performed using the Sigma Plot 2000 version 8.0 (SPSS Inc.) and the Sigma Stat version 3.1 (Jandel Inc.) packages.
Macroscopic and Microscopic Pathology.
The daily fluid consumption of the mice was unaffected by any treatment. One of the nine mice receiving high dose zalcitabine without uridine (group E) died in week 21. This animal was excluded from the analysis because we were unable to control for postmortem time. The maximal medium body weight of group E was reached at week 19 and at that time point was lower compared to all other groups. From week 19 until week 22 group E mice lost about 2% of body weight, whereas there was a 11% gain in the controls (P = 0.004) and a 3% gain in group F (P = 0.04). The liver weight of group E animals was elevated by 19% compared to controls (P < 0.001). The liver weight of all other groups did not differ from controls (Table 1).
NOTE. Significance (P < 0.05) vs. controls * and vs. no uridine †). Highly significant (P < 0.001) vs. control ** and vs. no uridine ††. Units: §, % of control mean; #, μmol/min−1 g protein−1; ‡, copies/hepatocyte; ∥, μmol/g tissue; ∞, mg/g tissue.
Liver weight (g)
0.68 ± 0.07
0.67 ± 0.11
0.73 ± 0.11
0.68 ± 0.07
0.81 ± 0.06**
0.70 ± 0.10†
100 ± 17
104 ± 15
45 ± 15**
73 ± 24*†
21 ± 9**
52 ± 9**††
Citrate synthase activity#
1783 ± 261
1945 ± 236
2249 ± 782
1796 ± 258
1729 ± 331
2311 ± 828
100 ± 15
96 ± 21
86 ± 31
95 ± 30
74 ± 20*
97 ± 13†
100 ± 32
107 ± 28
108 ± 19
119 ± 25
105 ± 18
97 ± 20
658 ± 101
626 ± 104
522 ± 105*
607 ± 91
413 ± 48**
543 ± 78*††
207 ± 78
184 ± 37
335 ± 114*
215 ± 103†
446 ± 117**
253 ± 59††
100 ± 49
101 ± 36
207 ± 53**
116 ± 54†
420 ± 162**
178 ± 36*††
The degree of hepatotoxicity as histologically assessed with the necroinflammatory score was substantially elevated in all groups treated with zalcitabine, dose dependent and most prominent in group E (Fig. 1A). Coadministration of uridine reduced the necroinflammatory score in both the high dose (P < 0.001) and the low dose (P = 0.02) zalcitabine groups.
Zalcitabine also caused a dose dependent liver steatosis as evidenced by elevated steatosis scores (Fig. 1B). On oil red staining, the steatosis was microvesicular and panacinar (Fig. 2B). In group E, there was also mild mononuclear infiltration with predominantly perisinusoidal and just in a few cases periportal involvement. The cytoplasm contained structures consistent with megamitochondria. Lipogranulomas were not observed. Uridine alone had no intrinsic effect on liver steatosis, but when given with zalcitabine, reduced steatosis score in both dose groups (group C vs. D, P = 0.02; group E vs. group F, P < 0.001, Fig. 1B and Fig. 2C). Among all animals, the steatosis score correlated positively with the necroinflammatory score (r = 0.86, P < 0.001).
On electron microscopy the hepatocytes of the zalcitabine high dose group contained increased numbers of mitochondria (Fig. 2E). The organelles were enlarged and had disrupted cristae. There were small intracytoplasmic vacuoles, consistent with fat storage. Uridine supplementation abrogated this ultrastructural damage; only a slight increase in mitochondrial number was observed (Fig. 2F). The mitochondria of the low dose zalcitabine group appeared to also have a slight increase in number; the organelles in all other groups did not show ultrastructural abnormalities (not shown).
Apoptotic nuclei were abundant in the zalcitabine high dose group (Fig. 2K) and to a lesser amount in the low dose group (not shown). Uridine supplementation prevented the zalcitabine induced apoptosis (Fig. 2L) but when given without zalcitabine had no effects on apoptosis (not shown).
As expected, the steatosis score was positively correlated with the amount of intrahepatic lipids (r = 0.73, P < 0.001) among all mice. Zalcitabine increased the amount of hepatic lipids in a dose dependent fashion (Fig. 1B). Uridine alone had no effect on hepatic lipids but when co administered with zalcitabine reduced the elevated hepatic lipid content in the low and the high dose zalcitabine groups. Hepatic lipids in the low dose zalcitabine plus uridine group did not statistically differ from controls (P = 0.07). Hepatic lipids correlated positively with the necroinflammatory score among all animals (r = 0.76; P < 0.001; Fig. 3A).
The mean ratio of hepatic COX and SDH-activity is an index of mtDNA encoded respiratory chain function and was reduced in all mice treated with zalcitabine (Table 1). In the low dose zalcitabine group the mean COX/SDH-ratio was 45% of control values and 21% in the high dose zalcitabine group. Uridine supplementation per se did not alter the COX/SDH-ratio, but when given in conjunction with zalcitabine attenuated although not fully normalized respiratory dysfunction. The enzymatic activity of CS did not differ among all groups. Among all mice, the COX/SDH-ratio correlated negatively with the steatosis score (r = −0.79, P < 0.001), the necroinflammatory score (r = −0.86; P < 0.001) and the amount of hepatic lipids (r = −0.78, P < 0.001, Fig 3B).
MtDNA-Encoded Respiratory Chain Protein.
The expression of the mtDNA-encoded cytochrome c-oxidase subunit was normalized for the expression of the nDNA-encoded subunit by calculating the COX I/COX IV-ratio. The mean COX I/COX IV-ratio was reduced in the high dose zalcitabine group (Table 1) compared to controls (P = 0.007). Co-treatment with uridine significantly improved and normalized the COX I/COX IV-ratio. In the low dose zalcitabine group, the COX I/COX IV was 14% lower but not significantly different from controls. The COX I/COX IV ratio was inversely correlated with the steatosis score (r = −0.34, P = 0.04), the necroinflammatory score (r = −0.35; P = 0.01) and the lipid content (r = −0.4; P = 0.01). The COX IV/GAPDH ratio did not statistically differ between all groups (Table 1). Taken together, these results indicate that the respiratory chain defect is restricted to the mtDNA-encoded respiratory chain subunit, whereas the nDNA-encoded subunit is preserved.
The mean wild mtDNA copy number was reduced in mice treated with low and high dose zalcitabine (79% ± 16%, P = 0.02 and 63% ± 7%, P < 0.001 of control values, respectively). Uridine had no intrinsic effect on mtDNA biogenesis, but when administered with zalcitabine attenuated and in the case of low dose zalcitabine fully abrogated mtDNA depletion (Table 1). Among all groups, liver mtDNA copy numbers were inversely correlated with the steatosis score (r = −0.69, P < 0.001), the necroinflammatory score (r = −0.63; P < 0.001) and the amount of hepatic lipids (r = −0.54, P < 0.001, Fig 3C), and positively correlated with the COX/SDH-ratio (r = 0.6, P < 0.001).
Intrahepatic Reactive Oxygen Species.
Mean MDA levels are an indirect indicator of the formation of reactive oxygen species (ROS) and were elevated in the livers of the low and high dose zalcitabine group (161% ± 55%, P = 0.01 and 215% ± 57%, P < 0.001 of control values, respectively). Mitocnol alone did not induce an elevation in mean MDA levels (Table 1) but normalized the elevated MDA content in mice exposed to both zalcitabine doses. MDA levels were positively correlated with the steatosis score (r = 0.63, P < 0.001) and the amount of intrahepatic lipids (r = 0.69, P < 0.001). MDA levels were negatively correlated with the COX/SDH ratio (r = −0.67, P < 0.001), the COX I/COX IV ratio (r = −0.67, P < 0.001) and mtDNA copy numbers (r = −0.62, P < 0.001).
Among the livers of all mice, superoxide content was highly correlated with MDA levels (r = 0.9, P < 0.001). Similar to MDA levels, superoxide was increased by a factor of 2.1 and 4.2 in the low and high dose zalcitabine groups when compared to controls. In the zalcitabine low dose group; superoxide content was normalized by uridine coadministration although uridine did not appear to affect ROS production when given without zalcitabine. In the zalcitabine high dose group, uridine reduced superoxide content. Superoxide levels correlated positively with the steatosis score (r = 0.77, P < 0.001) and the amount of intrahepatic lipids (r = 0.69, P < 0.001, Fig. 3D), whereas inverse correlations were identified between superoxide content and the COX/SDH-activity ratio (r = −0.73, P < 0.001), the COXI/COXIV ratio (r = −0.48, P = 0.002), and hepatic mtDNA copies (r = −0.59, P < 0.001).
Our investigations establish a new model of nucleoside analogue mediated mitochondrial hepatotoxicity, provide insight into its pathogenesis and suggest beneficial effects of uridine supplementation.
In this model, zalcitabine induced a dose dependent mtDNA-depletion and specifically impaired mtDNA-encoded respiratory chain functions, in keeping with its inhibitory activity on polymerase-gamma. Impairment of oxidative phosphorylation inhibits beta oxidation and leads to an intracellular increase of triglycerides and non-esterified fatty acids, the cause of microvesicular steatosis.26In vitro investigations have suggested that residual wild-type mtDNA-levels in the order of 20% protect from respiratory chain dysfunction.27, 28 We have however not found support for this mtDNA threshold effect in vivo. The in vivo and in vitro discrepancy of the mtDNA genotype phenotype correlation may be related to the possibility that hepatocytes maintain ATP-synthesis more efficiently in vitro than in vivo, because they can more efficiently compensate for their defect in oxidative phosphorylation by increasing glycolysis due to the ample supply of glucose in the culture medium.
Although mtDNA-depletion is probably sufficient to explain the onset of liver pathology in humans,29–31 additional and possibly interconnected effects are likely to contribute to mitochondrial injury.32 We have observed an increase in hepatic ROS production. It is likely that the increased ROS production originates at least in part from the respiratory chain, because the decreased activity of respiratory chain complexes partially blocks the flow of electrons, allowing them to react with oxygen.32 The enhanced mitochondrial ROS formation can then directly attack mtDNA and respiratory chain polypeptides, but also mitochondrial cardiolipin, the latter being important for COX function and also releasing reactive lipid peroxidation products.32, 33 Polymerase-gamma is also a target of oxidative damage.34 Through these effects, ROS may close several vicious circles that contribute to the respiratory decline.
Our study suggests that the dietary supplementation of uridine attenuates the mitochondrial hepatotoxicity of zalcitabine in accord with prior in vitro data.13 The mechanism of the beneficial effect of uridine is not fully delineated but it is conceivable that uridine itself or its metabolites disinhibit mtDNA replication by competing with antiviral pyrimidine analogues at polymerase-gamma or other steps of intracellular NRTI transport or phosphorylation.13, 14 Uridine may alternatively correct an intracellular pyrimidine deficit which in itself induces cell cycle arrest and apoptosis, as demonstrated by experimental work with direct DHODH inhibitors.16, 17 The TUNEL results suggest that apoptosis can principally accompany hepatic mtDNA-depletion, but further quantitative conclusions from the observed extent of apoptosis should not be made. The importance of the intracellular pyrimidine pools for the survival of cells with a dysfunctional respiratory chain is also supported by the fact that cells without a single molecule of mtDNA are rescued from cell death and grow virtually normal, if the intracellular pyrimidine pools are replenished by uridine.27
Extensive pharmacokinetic studies suggest that uridine concentrations that are protective in vitro can be safely achieved with oral and parenteral dosing.14 Oral uridine supplementation (150 mg/kg/d) is also recommended and safely used long-term in patients with hereditary orotic aciduria, an inborn error of pyrimidine de novo synthesis, in which uridine reverses megaloblastic anemia and other symptoms.35
There is a theoretical concern, that high uridine levels are not only able to compete with NRTIs at the level of polymerase-gamma but may also do so at the level of HIV-reverse transcriptase. Phenotypic HIV-resistance assays however did not find a negative effect of uridine on viral suppression.36, 37 Clinical trials performed for other indications did not detect a negative effect of uridine supplementation on the antiretroviral efficacy of NRTI, although the number of patients were few and follow-up was short.38, 39
Our pilot study has several limitations. Serum pharmacokinetics of zalcitabine was not performed because there is no reliable assay. The optimal dose of uridine is also unknown because intramitochondrial concentrations of NRTIs and of uridine metabolites cannot be measured. It is also not possible to extend the observations from mice to humans and from zalcitabine to the other more frequently used NRTI. In vitro however, uridine does also antagonize the mitochondrial toxicity of stavudine and zidovudine which like zalcitabine are pyrimidine analogues, whereas the mitochondrial toxicity of didanosine, a purine analogue was not abrogated.13, 24 This observation underlines the hypothesis of a competitive effect of uridine with pyrimidines.
HCV-infection (and HCV genotype I in particular) has also been found to modestly decrease hepatic mtDNA.40, 41 An increased formation of ROS is thought to represent an important contributor for the HCV-mediated mtDNA decline. The effects of uridine on HCV related mitochondrial dysfunction have however not been examined.
The clinical experience of uridine supplementation for hepatotoxicity is very limited but promising. In one HIV patient uridine supplementation was described to reverse mitochondrial steatohepatitis despite continued long-term antiretroviral treatment with stavudine.42 Using a non-invasive 13C-methionine breath test in HIV-positive individuals under treatment with stavudine or zidovudine, a three day course of Mitocnol was recently found to reproducibly enhance the function of hepatic mitochondria over a period of 4 weeks.43
We conclude that uridine has beneficial effects in this murine model of pyrimidine analogue induced hepatotoxicity. Uridine supplementation should be examined in a carefully monitored clinical trial in HIV-infected patients who have pyrimidine analogue-induced mitochondrial steatohepatitis and who cannot be switched to antiretrovirals with a lower potential of hepatotoxicity.
We thank Pharma Nord (Vojens, Denmark) for providing Mitocnol and Kerstina Melkaoui, Karin Sutter and Carmen Kopp for expert technical assistance.