| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Biochemistry and Biophysics, Oregon State University, Corvallis, Oregon, USA
1Correspondence: Department of Biochemistry and Biophysics, Oregon State University, 2011 Agricultural and Life Sciences Bldg., Corvallis, OR 97331-7305, USA. E-mail: mathewsc{at}onid.orst.edu
 |
ABSTRACT |
|---|
Key Words: mitochondria • DNA precursors • deoxyribonucleotides • nucleotide pools
|
INTRODUCTION |
|---|
2
3)
testifies to the importance of maintaining mitochondrial genomic integrity and copy number. To understand how these functions are carried out, it is essential to know the metabolic sources of the precursor nucleotide pools that serve the mitochondrial genome, because the concentrations of deoxyribonucleoside triphosphates (dNTPs) at replication sites are important determinants of both the rate and fidelity of DNA replication (4)
.
The dNTP pools that serve the nuclear genome are synthesized primarily in the cytosol and presumably pass into the nucleus passively through the nuclear pore complex (5
, 6)
. However, the highly selective permeability of the mitochondrial membrane system suggests that the provision of dNTPs to the mitochondrial replisome is a more complex process. Related to this is the fact that bulk intracellular dNTP accumulation is regulated in concert with the cell cycle (4)
, concomitant with total metabolic demand for DNA precursors, while mitochondrial DNA replication proceeds throughout the cell cycle and in noncycling cells. Hence, precursor pools for mitochondrial DNA must be continuously present. The importance of maintaining these pools is underscored by the existence of about a half dozen human disease states characterized by mutations in enzymes of dNTP synthesis that affect either mtDNA maintenance or genomic stability (1
2
3
4)
.
At present we know little about the enzymes and transport systems responsible for synthesizing and maintaining intramitochondrial dNTP pools. In this article we will summarize what is known and identify major unanswered questions and suggestions for future research.
|
EARLY FINDINGS |
|---|
showed that in cultured mouse cells treated with antimetabolites, mtDNA replication proceeded almost undisturbed, but nuclear replication was strongly inhibited. This suggested that mtDNA is serviced by distinct pools that selectively resist the depletion caused by metabolic inhibitors. This hypothesis was tested directly in our laboratory, when Bestwick et al. (8)
measured dNTP pools in HeLa cells and found that they selectively resisted depletion when cells were treated with either methotrexate or 5-fluorodeoxyuridine, whereas whole-cell pools, which represent primarily nuclear and cytosolic components, were diminished. In a related study, Bestwick and Mathews (9)
observed that the mitochondrial dCTP pool was labeled more efficiently by exogenous [3H] uridine than was the whole-cell dCTP pool, suggesting that mitochondria use salvage pathways to dNTPs more efficiently than does the remaining cellular machinery.
The idea that deoxyribonucleotides synthesized in the cytosol might be transported into mitochondria received support as early as 1968, when Karol and Simpson (10)
used a density-labeling protocol to show that incubating rat liver mitochondria with exogenous dNTPs led to incorporation of nucleotides into mtDNA in a replicative process. In 1994, Enríquez et al. (11)
showed that isolated rat liver mitochondria could incorporate exogenous dNTPs for several hours, with a label from [
32P]dCTP being incorporated into a product having the same molecular weight as mtDNA and containing label in both strands. The process was shown to be dependent on exogenous ADP and an oxidizable substrate, indicating a requirement for mitochondrial ATP generation. Both of these studies suggested that dNTPs could be directly transported into mitochondria from the cytosol, although neither group of authors made that explicit suggestion. However, the data as presented do not rule out the possibility that dNTPs undergo dephosphorylation at the mitochondrial membrane and are transported inward as mono- or diphosphates, then rephosphorylated within the organelle.
The importance of understanding the nature and sources of mitochondrial DNA precursor pools assumed greater urgency beginning in the late 1990s with the discovery of several human disease states in which mitochondrial genomic instability or inability to maintain adequate levels of mtDNA were found to be correlated with mutations affecting enzymes or proteins involved in nucleotide metabolism (see below).
|
POSSIBLE METABOLIC ROUTES TO INTRAMITOCHONDRIAL dNTPs |
|---|
shows four possible routes by which dNTPs might arrive within mitochondria. The list of routes is not exhaustive, but includes all for which there exists at least some evidence to support the respective pathway shown.
|
Nucleotide transport
As noted above, the ability of extramitochondrial dNTPs to label DNA in isolated mitochondria suggested the existence of active transport systems capable of carrying out dNTP import. Such a system was described in 1999 by Bridges et al. (12)
, who reported on a dCTP transport system from human leukemia cells. A purified mitochondrial protein fraction from these cells was reconstituted into liposomes and shown to carry out dCTP transport with a KM of 3 µM, well within the physiological range. The process evidently was energy independent because transport occurred in the absence of added ATP or an energy-generating system. The nucleotide was shown to be transported at the triphosphate level because rates of radiolabeled dCTP uptake were identical whether a 32P label was placed on the or 
position. Activity was not affected by inhibitors of the well-known ATP/ADP carrier, ruling out an involvement of this system (although a role for the ATP/ADP carrier in deoxyribonucleotide metabolism has been proposed; discussed later). The system was weakly inhibited by other dNTPs, leaving open the important question of whether distinct transport systems exist for each of the dNTPs.
A different transport protein was described by Dolce et al. in 2001 (13)
and designated the deoxynucleotide carrier, or DNC. This protein was expressed from a cDNA that encodes a member of a family of known mitochondrial transport proteins. After expression in bacteria, the protein was reconstituted into phospholipid vesicles and found to transport all four deoxyribonucleoside tri- and diphosphates in exchange for dNDPs, ADP, or ATP, with diphosphates transferred more efficiently than triphosphates. Ribonucleoside di- and triphosphates were also found to be good substrates for this protein. Although the concentration dependence of the protein was not systematically analyzed in this study, data were reported with nucleotide substrate concentrations considerably higher than physiological values (14)
, suggesting that the actual biological role of the protein might be something other than deoxyribonucleotide transport. As discussed later, there is good evidence that this is so.
Recently Ferraro et al. (15)
reported that mouse liver mitochondria carry out transport of thymidine nucleotides at the monophosphate level. Mitochondria concentrated dTMP 
100-fold from the medium, with some of that nucleotide phosphorylated to dTDP and dTTP within the organelle. The activity reached saturation at low dTMP concentrations. There was no requirement for ATP or an oxidizable substrate, and the activity was insensitive to inhibitors of the ATP/ADP carrier. Transport activity was not inhibited by other deoxyribonucleoside monophosphates or triphosphates. The high affinity and evident high specificity of the system support the premise that this is a physiologically significant route to intramitochondrial dNTPs, and the lack of requirement for energy or another carrier suggests that the system exchanges extramitochondrial nucleotide anion for intramitochondrial OH–. The question of whether other deoxyribonucleotides might be transported at the monophosphate level was not addressed in this study.
Nucleoside salvage within the mitochondrion
Conclusive evidence exists for the importance of deoxyribonucleosides as mitochondrial dNTP precursors. In the first place, two of the four deoxyribonucleoside kinases known to exist in mammalian cells are localized to mitochondria: deoxyguanosine kinase (dGK) and TK2, an isoform of thymidine kinase (16)
. These two enzymes have sufficiently broad substrate specificity to account for phosphorylation at significant rates of all four canonical deoxyribonucleosides. Equally important, a genetic deficiency of either enzyme has serious clinical consequences in humans.
Although, as noted, mitochondrial dGK and TK2 can account for conversion of all four deoxyribonucleosides to their monophosphates within the organelle, it is not clear that this route can be significant for the synthesis of all four dNTPs. This is because little information is available about the activities of dNMP kinases within the mitochondrion. Once formed, dNDPs can readily be converted to triphosphates because mitochondria contain substantial amounts of an isoform of nucleoside diphosphate kinase, an enzyme of low specificity known to be capable of converting all eight canonical nucleoside diphosphates to their respective triphosphates (17)
. So a key objective in evaluating these salvage pathways to mitochondrial dNTPs is to analyze mitochondria for dNMP kinase activities.
Equally important as evidence supporting a role for mitochondrial deoxyribonucleoside kinases is the toxicity of nucleoside analogs used to treat HIV infections (18
19
20
21)
. These analogs, such as 3'-azidothymidine and 2',3'-dideoxycytidine, must be converted after cellular uptake to the corresponding dNTPs, which act as inhibitors of HIV reverse transcriptase probably by incorporation, followed by chain termination because of the absence of a 3'-hydroxyl for chain extension. Some of these nucleoside reverse transcriptase inhibitor (NRTI) drugs are substrates for one of the mitochondrial deoxyribonucleoside kinases, so that some or all of the conversion to dNTPs occurs within the mitochondrion; the dNTP analogs turn out to act against DNA polymerase , the mitochondrial DNA polymerase, as well as the intended target, HIV reverse transcriptase. A consequence is a series of toxic side effects, including myopathy, cardiotoxicity, or peripheral neuropathy, depending on the specific drug in question. Correlated with toxicity is the failure of cells to maintain normal levels of mitochondrial DNA.
To substantiate pol as the target for toxicity of NRTI drugs, Johnson and colleagues (22)
have carried out extensive pre-steady-state kinetic analysis of recombinant human pol with triphosphates of the several NRTI drugs that have been approved for antiretroviral therapy. One parameter, the toxicity index, is related to the rate of incorporation of an analog into DNA relative to its removal by the proofreading 3' exonuclease activity of the polymerase. A dNTP analog with a high toxicity index would show a high rate of incorporation by pol , accompanied by inefficient removal by proofreading. This parameter compares well with measurements of clinical toxicity, supporting the premise that mitochondrial polymerase is the target for NRTI toxicity. Another parameter, the theoretical therapeutic index, is compared with the clinical therapeutic index, which is a ratio of doses causing significant toxicity to those showing significant reverse transcriptase inhibition. The theoretical therapeutic index is a ratio of the ability of pol to discriminate against an analog dNTP compared with the ability of HIV reverse transcriptase to discriminate against the same analog. Effective analogs are those that are incorporated poorly by mitochondrial polymerase (high discrimination) and efficiently by reverse transcriptase (low discrimination). These and other parameters obtained through kinetic analysis are expected to be useful in evaluating new nucleoside analogs as anti-HIV drugs, because their effectiveness and toxicity can be predicted by simple kinetic analysis prior to more involved testing of clinical effectiveness and toxicity.
The toxicity and therapeutic indices just described (22)
were generally found to correlate well with clinical observations. An exception is AZT, which is predicted to be relatively effective but nontoxic, whereas AZT is known to be quite toxic. This apparent anomaly may be explained by the data of McKee et al. (23)
, who explored uptake and metabolism of thymidine and AZT in isolated rat heart mitochondria. Whereas thymidine was efficiently taken up and converted to dTTP, AZT accumulated within mitochondria as the monophosphate. Evidently thymidylate kinase or the enzyme responsible for converting dTMP to dTDP does not act on AZT-MP, the monophosphate of AZT. McKee et al. showed further that AZT is a fairly strong inhibitor of TK2, the mitochondrial thymidine kinase, and they postulated that the resultant inhibition of dTTP accumulation within the mitochondrion could be the basis for inhibition of mtDNA replication and, hence, for AZT toxicity. In subsequent work (24
, 25)
, the McKee laboratory has demonstrated that AZT is indeed a competitive inhibitor of TK2, as seen in both isolated rat heart and liver mitochondria with Ki values in the 10 to 15 µM range. Lynx and McKee (25)
suggested that AZT toxicity might be partially overcome by supplementation with thymidine. Of course, if thymidine treatment causes the dTTP pool to expand, that could lessen the therapeutic efficacy of AZT.
De novo synthesis within the mitochondrion
The data cited above indicate that mitochondria can take up DNA precursors at either the deoxyribonucleoside or deoxyribonucleotide levels and convert them to dNTPs for mtDNA synthesis. Since human mitochondria contain 1500 or more different proteins (26)
, it is worth asking whether these proteins include the machinery for de novo biosynthetic pathways to dNTPs. Human mitochondria contain a distinct isoform of dUTPase, which is the product of alternative splicing of the same gene that encodes the cytosolic form of the enzyme (27)
. Although this enzyme might be present to remove dUTP produced through the action of thymidine kinase on deoxyuridine (16)
, this and other considerations prompted our laboratory to ask whether mitochondria contain a complete de novo pathway, beginning with ribonucleotide reductase. In 1994, our laboratory published preliminary evidence for a distinctive form of ribonucleotide reductase in HeLa cell mitochondria (28)
. However, the high activity of the cytosolic enzyme in these rapidly proliferating cells made it impossible to rule out cytosolic contamination as the basis for the mitochondrion-associated activity. More recently, one of us (S. Song, Ph.D. thesis, Oregon State University, 2006) examined this question again using a nonproliferating tissue, namely, mammalian liver. As shown in Table 1
, preparations from rat, mouse, and pig liver all showed ribonucleotide reductase-specific activity at least twice as high in mitochondria as in cytosol, essentially ruling out cytosolic contamination as the basis for the mitochondrial activity. In the same experiment, HeLa cells showed cytosolic activity >3-fold higher than mitochondrial activity, a finding suggesting that some of the mitochondrial activity we observed earlier (28)
in HeLa cells may have represented cytosolic contamination. Unfortunately, the liver mitochondrial activity has resisted our attempts at purification and further characterization, so the data of Table 1
, obtained with crude mitochondrial extracts, must be regarded as preliminary.
|
A quite different involvement of ribonucleotide reductase in mtDNA maintenance was presented recently. In an oral meeting report, Shadel and Eaton (Abstracts Ann. Mtg. Am. Soc. Biochem. Mol. Biol., 2006) described preliminary evidence suggesting a role for a DNA damage-inducible form of ribonucleotide reductase in maintaining mitochondrial dNTP pools. The induction of a novel small subunit of ribonucleotide reductase, p53R2, by p53 expression has been known for several years (29)
and is considered to represent part of the metabolic response to DNA damage. Shadel and Eaton observed abnormally low mtDNA levels in cells unable to express p53, and they followed this up by showing that knockdown of p53R2 by RNAi caused mtDNA levels to drop in HeLa cells. These observations suggest an intriguing relationship between mitochondrial function and DNA damage.
|
HUMAN MUTATIONS AFFECTING MITOCHONDRIAL NUCLEOTIDE METABOLISM |
|---|
Mitochondrial neurogastrointestinal encephalomyopathy
This condition (MNGIE), a recessive nuclear gene disorder, affects mitochondrial function and is associated with point mutations and multiple long deletions in mitochondrial DNA of skeletal muscle. Nishino et al. (30)
showed that this condition results from a deficiency of thymidine phosphorylase, an enzyme that catalyzes the reversible breakdown of thymidine to thymine plus deoxyribose-1-phosphate. Patients with this condition display elevated levels of circulating thymidine (31)
, leading to the suggestion that uptake of excess thymidine by mitochondria stimulates salvage synthesis of dTTP, which in turn unbalances the other dNTP pools and stimulates mutagenesis. This hypothesis has been supported by work in our laboratory.
Progressive external ophthalmoplegia
This disease (also called PEO) is an autosomal dominant condition in which mtDNA undergoes multiple deletion mutations. An autosomal recessive form of the condition is also known. One form of the disease has been related to a mutation in the heart/skeletal muscle isoform of the ATP/ADP carrier (32)
. Since this carrier exchanges intramitochondrial ATP for extramitochondrial ADP, Kaukonen et al. (32)
speculated that because ADP is a ribonucleotide reductase substrate, the ATP/ADP carrier deficiency might cause a mutagenic dNTP pool imbalance. Another form of PEO has been shown to result from a mutation affecting DNA polymerase . Although dNTP pool abnormalities may be responsible for genomic instability in other forms of PEO, that probably is not a factor here because the mutant enzyme has been shown to have reduced replication fidelity in vitro (33)
. Recently, however, Baruffini et al. (34)
analyzed biochemical mechanisms involved in PEO by introducing into the yeast gene for DNA polymerase mutations corresponding to human pol mutations responsible for either the dominant or recessive form of PEO. Both yeast mutants displayed elevated frequencies of petite mutations. Both showed amelioration of the mutator phenotype when ribonucleotide reductase was overexpressed, presumably a result of increased dNTP pools.
Still another form of PEO results from mutations in the mitochondrial helicase called TWINKLE (35)
. Because helicases use nucleoside triphosphate energy to drive DNA unwinding, the authors speculated that a dNTP imbalance could arise as a consequence of abnormal nucleoside triphosphatase activities of the mutant TWINKLE protein. However, to date no analyses have been done of the effects of any of these mutations on intramitochondrial dNTP pools. Recently, however, Tyynismaa et al. (36)
described an animal model for the disease. A transgenic mouse strain was created with alanine substituted for threonine at TWINKLE position 360, the same mutation seen in humans with this condition. These mice displayed nearly all the clinical symptoms of humans with the condition, providing a model system for analyzing the effects of mutant TWINKLE on parameters of dNTP metabolism.
Mitochondrial DNA depletion myopathy
A number of conditions involve defective mitochondrial function associated with abnormally low levels of mitochondrial DNA. In fact, as noted earlier, mtDNA depletion is probably one of the main features of AZT toxicity. With respect to enzymes of dNTP metabolism, two forms of this disease are known. One form results from mutations affecting TK2, the mitochondrial isoform of thymidine kinase (37
, 38)
, and another from deficiency of deoxyguanosine kinase (39
, 40)
. The TK2 deficiency syndrome includes muscle weakness and liver or kidney failure, while the dGK deficiency also involves liver failure but in addition involves severe neurological abnormalities. Saada et al. (41)
measured mitochondrial dNTP pools in fibroblasts from two patients with the TK2 deficiency. Although mtDNA levels were normal in these cultured cells, both cell lines displayed pool abnormalities, with dTTP pools, as expected, being lower than those in control cells by one-half in one line and two-thirds in the other. On the other hand, Desler et al. (42)
reported that mitochondrial dNTP levels were relatively unaffected by a TK2 deficiency. However, in their study the deficiency was induced in HeLa cells by an antisense RNA, which lowered the TK2 activity by less than one-half, making this a less rigorous analysis of the functional significance of TK2.
With respect to the dGK deficiency syndrome, Wang and Eriksson (40)
introduced two clinically observed dGK mutations into the human dGK gene and purified the two mutant enzymes as recombinant proteins. One mutant enzyme lost all activity with deoxyadenosine, a physiological substrate, while the other showed greatly reduced Vmax values for all substrates. These studies, in addition to helping define the significance of salvage pathways to dNTPs in mitochondria, may help in revealing ways to treat mitochondrial DNA depletions.
Amish lethal microcephaly
This condition, in which the head is small and the brain malformed, has been observed only in Old Order Amish families whose ancestors lived in Lancaster County, Pennsylvania. Through whole-genome screening and fine-scale genetic mapping, Rosenberg et al. (43)
correlated this condition with a mutation in the deoxynucleotide carrier, DNC (13)
. The mutation found in patients (Gly177Ala) was introduced into the wild-type protein and the purified mutant protein was incorporated into proteoliposomes. Liposomes containing the wild-type protein efficiently catalyzed an ATP/ADP exchange, whereas the mutant protein was inactive in this assay. The authors speculated that the mutational defect in deoxyribonucleotide transport led to mitochondrial pools too low to support normal mtDNA replication. However, studies of mitochondrial nucleotide pools failed to support this prediction (44)
. As part of a collaborative study with the laboratories of L. Biesecker and F. Palmieri, we found mitochondrial dNTP and rNTP pools to be normal, both in lymphoblasts from humans with this condition and in embryos of mice engineered to have the disease-related mutation introduced into their DNC protein. Further analysis of the DNC gene sequence by our collaborators (44)
revealed a similarity to a protein in yeast known to participate in mitochondrial uptake of thiamine pyrophosphate. This led in turn to the finding that cells carrying the Amish lethal microcephaly gene, whether in patient lymphoblasts or transgenic mouse embryo fibroblasts, have deficient intramitochondrial levels of thiamine pyrophosphate. Although this finding suggests that Amish lethal microcephaly does not result from defective nucleotide metabolism, it does help to explain why patients with this condition excrete large quantities of -ketoglutarate, a substrate for the thiamine pyrophosphate-requiring -ketoglutarate dehydrogenase complex.
The above conclusion, that the primary function of the DNC is not deoxyribonucleotide transport, was anticipated by results of Lam et al. (45)
, who reported that cultured cells overexpressing DNC were not sensitized to mtDNA damage caused by antiviral nucleoside analogs, nor did down-regulation of DNC affect either mtDNA depletion or the rate of dTTP uptake into isolated mitochondria. On the other hand, Lewis et al. (46)
reported that transgenic expression of the human DNC gene in mice caused mitochondrial damage in the animals and further damage by nucleoside reverse transcriptase inhibitors. The ensembles of nucleoside analogs include one that was used by both groups: 2',3'-didehydro-2',3'-dideoxythymidine (d4T). So there is a discrepancy between the results of these two laboratories, which might have its basis in the fact that one study was done in cell culture and one with transgenic animals.
|
MITOCHONDRIAL dNTP POOL ASYMMETRIES |
|---|
reported that MNGIE patients have elevated plasma concentrations of both thymidine and deoxyuridine, a finding consistent with the deficiency of an enzyme involved in catabolism of both nucleosides. The authors predicted that the excess nucleoside could be taken up into cells and mitochondria, where it might generate a mutagenic pool imbalance in mitochondria. In our laboratory, HeLa cells, used as a model system, were cultured in the presence of thymidine at 50 µM (47)
. After 4 h incubation, the mitochondrial dTTP pool had expanded by 2.5-fold relative to the untreated control; dGTP accumulated in a similar manner. The most dramatic effect was a depletion of dCTP, already the least abundant dNTP in HeLa cell mitochondria, to almost immeasurably low levels. These pool effects are consistent with the possibility that ribonucleotide reductase is active within mitochondria because eukaryotic forms of this enzyme show that dTTP activates GDP reduction and inhibits CDP reduction, in accord with our observations of dGTP accumulation and dCTP depletion. However, this result could also signify that the cytosolic ribonucleotide reductase is a major contributor to mitochondrial dNTP pools.
How do these observations align with the genetic abnormalities of MNGIE? Mitochondrial DNA from MNGIE patients shows both point mutations and deletions. Sequence analysis of MNGIE patient DNA shows the most common mutation to be an AT
GC transition, where T in the template strand is followed by two A’s (48)
. Accumulated dGTP presumably competes with dATP for incorporation opposite T, and this mismatch is sealed in place by a next-nucleotide effect in which the expanded dTTP pool pairs efficiently with the two template A’s.
Because MNGIE also involves deletion mutations, we wanted to see whether the thymidine-induced pool imbalance could cause mutations in our model system; indeed, prolonged incubation of HeLa cells with thymidine led to the accumulation of large deletions in mtDNA of treated cells, but not controls. Hirano et al. (49)
have speculated that deletions could arise if polymerase stalling led to parent-template strand unwinding at the replication fork, with the parental strand ultimately pairing with a downstream region of homology. Such a process might be facilitated by a replication mechanism like that of mtDNA, in which the two parental strands replicate independently of one another. So the important question at this stage is whether the dCTP depletion we have observed is sufficient to cause polymerase to stall in vivo.
The laboratory of Bianchi and Reichard has carried out comparable model studies of MNGIE (50
, 51)
. Their approach, which is complementary to ours, involves administration of radiolabeled thymidine to cultured cells, thereby permitting analysis of thymine nucleotide pool dynamics and turnover in addition to static measurements of the pool sizes. Their data indicate that the primary source of mitochondrial dNTPs in cycling cells is de novo synthesis beginning with cytosolic ribonucleotide reductase. Salvage pathways are presumed to become more prominent in quiescent cells, where ribonucleotide reductase is far less active. From the tissues affected in MNGIE, Bianchi and Reichard and co-workers make the reasonable assumption that the pathological consequences of MNGIE arise primarily in quiescent cells. They found that incubation of quiescent cells with thymidine or deoxyuridine at concentrations found in MNGIE patient plasma leads to mitochondrial dTTP pool expansions of up to 8-fold. Mitochondrial dCTP pools, on the other hand, were affected only slightly and purine dNTP pools were not reported (51)
. During prolonged incubation under these conditions, mtDNA levels were depleted, but possible accumulation of mutations was below the limits of detection.
Although the experimental approaches differ between our laboratory (47)
and the studies by Ferraro et al. (50)
and Pontarin et al. (51)
, the main conclusions from both laboratories support the premise (31)
that thymidine and/or deoxyuridine accumulation in the plasma is chiefly responsible for the pathological consequences of MNGIE. It is interesting that Lara et al. (52)
are experimenting with platelet infusion as a therapeutic intervention. Infusion of platelets from healthy donors to MNGIE patients restored circulating thymidine phosphorylase (TP) activity and reduced plasma thymidine and deoxyuridine levels. Such intervention, as the authors point out, would need to occur very early in the course of the disease, since the genetic damage to mitochondria is irreversible.
On the other hand, a recent report from a Japanese group raises the question of whether thymidine phosphorylase deficiency is really the prime cause of MNGIE (53)
. These authors identified a point mutation in the TP gene that was present in both an MNGIE patient and her unaffected mother. That particular mutation was then found in 2.8% of 145 normal individuals analyzed, all of whom had significant TP activity. It would be of interest to know how plasma thymidine/deoxyuridine levels correlate with this particular polymorphism and whether extragenic mutations exist that affect the synthesis of TP and which might be the true causative factor in this disease.
Mitochondrial dNTP pools in animal tissues
Until recently, the few laboratories measuring mitochondrial dNTP pools directly had done so with cultured cells. For several reasons it was desirable to investigate mitochondria from animal tissues partly to evaluate mitochondrial pools in nonproliferating cells. Another interest in our laboratory was related to aging-associated accumulation of mitochondrial muations (54)
and to the much higher spontaneous mutation rate for the mitochondrial than for the nuclear genome (55)
. Since dNTP concentrations at replication sites are determinants of replication fidelity, it was of interest to learn whether pool abnormalities contributed to high mutation rates and, if so, whether pools changed with age in a way that would help explain the accumulation of mitochondrial mutations in aging animals.
To that end, we measured dNTP pools in mitochondria isolated from rat liver, heart, skeletal muscle, and brain (56)
. We found no significant changes between young and aged animals. However, we did note significant differences among the different organs analyzed. All four tissues showed dGTP as a much larger component of the total dNTP pool than had been seen in nuclear or whole-cell pools (Fig. 2
). Whereas dGTP is almost always the least abundant dNTP in whole-cell extracts, usually comprising just 5–10% of total dNTP, that figure was as high as 85 to 90% in mitochondria from skeletal muscle or heart. In vitro analysis of replication errors, carried out in collaboration with the laboratories of Copeland and Kunkel (56)
, indicated that the pool asymmetries probably contribute to the high mitochondrial mutation rate. Not yet explored is the metabolic basis for the large asymmetries seen and the interorgan differences.
|
Ferraro et al. (15)
have emphasized the importance of rapid chilling of cells and organs during mitochondrial isolation to prevent cells from going anaerobic, which would convert ATP to ADP and AMP, with downstream effects on pool sizes of other nucleoside triphosphates. They reported data on mouse liver dNTP pools, which show less asymmetry than we had reported for rat tissues. Like us, they saw that mitochondrial dGTP levels are higher than in cytosol or whole cells, but that it is not the most abundant nucleotide. In mouse liver they reported dCTP>dATP>dTTP>dGTP, whereas our analyses in rat liver yielded dCTP>dGTP>dATP>dTTP. The quantitative differences may result from our use of different animal species. The extreme pool asymmetries we saw were in heart and skeletal muscle mitochondria, not liver.
The point about anaerobic metabolism is well taken, and the precautions they suggested are being incorporated into our protocols. However, in all of our experiments to date, organs were dissected out within 1 min after death and dropped into isolation buffer at ice temperature—precautions emphasized by Ferraro et al. Moreover, we had reasonably good agreement among different animals in each group, suggesting that variability resulting from anaerobic metabolism was not a factor. In fact, a comparison of standard deviations in our study with that of Ferraro et al. indicates quite similar levels of uncertainty in the pool data presented. The methods used by both laboratories for mitochondrial extraction and dNTP analysis are virtually identical: 60% methanol extraction followed by 3 min in boiling water, evaporation, and dissolution of the residue in water. Both laboratories use the same DNA polymerase-based assay for dNTPs, which is more sensitive than HPLC and essential because of the small amounts of dNTPs in mitochondrial extracts. As noted by Ferraro et al., the enzymatic assay is fraught with potential pitfalls, but these can be dealt with by routine precautions such as spiking extracts with standards to correct for losses during extraction, running polymerase reactions to completion, and running reactions at several different dilutions of the sample.
In considering whether transient anaerobiosis is responsible for our observed pool asymmetries, it is difficult to see how the ATP depletion caused by anaerobiosis could specifically deplete, for example, the dTTP pool and not the dGTP pool, yielding our observations in muscle or heart. Resolution of the disagreement between our two laboratories should involve measurements on the same organs from the same animals plus the precautions taken by Ferraro et al. to prevent transient anaerobiosis.
|
Cytosolic and mitochondrial dNTP dynamics |
|---|
 TOPABSTRACTINTRODUCTIONEARLY FINDINGSPOSSIBLE METABOLIC ROUTES TO...HUMAN MUTATIONS AFFECTING...MITOCHONDRIAL dNTP POOL... Cytosolic and mitochondrial dNTP... PERSPECTIVE AND CONCLUSIONSREFERENCES |
|---|
, and they have presented evidence that it participates in regulation of the dTTP pool by participating in a substrate cycle. The enzyme acts specifically at the monophosphate level, converting dTMP to thymidine. The three phosphates of thymidine are expected to be in rapid equilibrium, so that depletion of dTMP would cause the rapid breakdown of dTTP to dTDP and then to dTMP. Also significant are the discovery of a transport system operating at the dNMP level and evidence that the deoxyribonucleoside salvage pathway to mitochondrial dNTPs is the chief process operating in quiescent cells, while the deoxyribonucleoside-dependent pathway is primarily active in proliferating cells. Both findings were mentioned earlier in this article.
Finally, evidence from the Bianchi-Reichard laboratory indicates closer communication between mitochondrial and cytosolic dNTP pools (58
, 59)
than might have been predicted from our early analyses of mitochondrial dNTPs in HeLa cells (8)
. A key experiment was carried out with cells lacking TK1, the cytosolic thymidine kinase. These cells showed a rapid labeling of the mitochondrial dTTP pool by exogenous thymidine, followed by a slower increase in specific activity of the cytosolic dTTP pool, an observation indicating that TK2 in the mitochondrion is essential in these cells for dTTP synthesis and that dTTP formed in the mitochondrion can then exit the organelle and mix with dTTP formed by endogenous metabolism in the cytosol. The phosphorylation level at which mitochondrial efflux occurs is not known, but the existence of a monophosphate-dependent uptake system might suggest that efflux occurs also at the monophosphate level. Taken together, these results suggest a pattern of thymine nucleotide metabolism, as shown in Fig. 3
.
|
|
PERSPECTIVE AND CONCLUSIONS |
|---|
and 2) the recent demonstration that deoxyribonucleotides, at least dTMP, can be transported at the monophosphate level (15)
. The description in 1999 by Bridges et al. of a dCTP transport system points up the importance of determining whether triphosphate-dependent systems exist for transporting the other canonical dNTPs. Our preliminary evidence for an intramitochondrial ribonucleotide reductase activity suggests the need to evaluate ribonucleotide transport processes in addition to the well-known ATP/ADP carrier. Because ribonucleotides are needed for mitochondrial gene transcription, this is an important question, whether or not rNDPs turn out to be significant precursors to mitochondrial dNTPs. Whatever the roles of cytosolic or mitochondrial ribonucleotide reductases, the activities of deoxyribonucleoside monophosphate kinases in mitochondria have received little or no attention, and these should be investigated. Because multiple isoforms of the ATP/ADP carrier have been described and because one (32)
has been shown to play a role in dNTP metabolism, the actions of these proteins in different tissues should be explored as possibly contributing to observed tissue-specific variations in mitochondrial dNTP pools. Our finding of variable and asymmetric dNTP pools in mitochondria from different animal tissues suggests that dNTP metabolism varies considerably in mitochondria from different organs, and could complicate the issue of devising treatment strategies for mitochondrial diseases or of minimizing toxicities of antiviral nucleoside analogs. For example, could a high deoxyguanosine kinase level in muscle and heart contribute to the high mitochondrial dGTP pools in these organs? If so, that could make these tissues particularly resistant to dGK deficiency. Here we must consider dNTP turnover as a contributor to dNTP pool sizes, as espoused particularly by Bianchi and Reichard and their colleagues. Could the vanishingly low dTTP levels that we see in brain and skeletal muscle mitochondria result from high activity of the dNT-2 nucleotidase in these mitochondria? The specificity of this enzyme could cause preferential depletion of dTTP. If present at high levels, we would need to consider the possibility that activity of the enzyme might be continuing during nucleotide extraction, which could make our values for dTTP levels artifactually low.
With regard to dNTP turnover and known enzymes, we might consider the role of MTH-1, the mammalian homologue of the bacterial MutT protein. This enzyme, which is found in mammalian mitochondria (60)
, is thought to play its chief role in "sanitizing" the nucleotide pool by removing the oxidized guanine nucleotide 8-oxo-dGTP, whose incorporation into DNA, if not prevented, would be highly mutagenic. However, the enzyme also cleaves dGTP to dGMP, albeit less efficiently. Although dGTP is not known to be a physiological substrate, it is worth asking whether low mitochondrial activities of this enzyme could contribute to the extremely high dGTP pools in heart and skeletal muscle.
In addition, results from the Bianchi-Reichard laboratory on pool turnover and equilibration between compartments need to be extended to the other three canonical deoxyribonucleotides. Do specific transport systems exist for all four dNMPs? Is rapid equilibration between cytosolic and mitochondrial pools a general feature of deoxyribonucleotide metabolism? The ease with which one can follow thymine nucleotide kinetics by use of radiolabeled thymidine means that experiments with the other nucleotides will be more difficult, but they are still important.
Parallel efforts to understand mitochondrial DNA precursor metabolism are well under way at the level of bioinformatics and metabolic modeling (61
, 62)
. Given the complexity of the issues involved, the numbers of mitochondrial enzymes whose activities have not yet been assessed and the likelihood that the metabolic patterns vary significantly in different tissues, it is crucial that additional experimental data be collected to help make the simulations biologically plausible.
Although laborious enzymology lies ahead, the prospects of helping to treat or minimize the effects of mitochondrial diseases, of learning what controls mutagenesis in the mitochondrial genome, and of learning how to more effectively use antiviral nucleoside analogs richly justify the efforts that will be required.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication January 16, 2007. Accepted for publication February 22, 2007.
|
REFERENCES |
|---|
. J. Biol. Chem. 276,23616-23623 account for mitochondrial toxicity from nucleoside analogs. J. Mol. Biol. 329,45-57[CrossRef][Medline] associated with progressive external ophthalmoplegia causes error-prone DNA synthesis. J. Biol. Chem. 277,15225-15228This article has been cited by other articles:
|
|
 |
|
  S. Bulst, A. Abicht, E. Holinski-Feder, S. Muller-Ziermann, U. Koehler, C. Thirion, M. C. Walter, J. D. Stewart, P. F. Chinnery, H. Lochmuller, et al. In vitro supplementation with dAMP/dGMP leads to partial restoration of mtDNA levels in mitochondrial depletion syndromes Hum. Mol. Genet., May 1, 2009; 18(9): 1590 - 1599. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. K. Mathews Academic Life: The Whole Package J. Biol. Chem., July 25, 2008; 283(30): 20613 - 20620. [Full Text] [PDF]
|
|
|
|
 |
|
 G. S. Shadel Expression and Maintenance of Mitochondrial DNA: New Insights into Human Disease Pathology Am. J. Pathol., June 1, 2008; 172(6): 1445 - 1456. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z. F. Pursell, J. T. McDonald, C. K. Mathews, and T. A. Kunkel Trace amounts of 8-oxo-dGTP in mitochondrial dNTP pools reduce DNA polymerase {gamma} replication fidelity Nucleic Acids Res., April 1, 2008; 36(7): 2174 - 2181. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Solaroli, X. Zheng, M. Johansson, J. Balzarini, and A. Karlsson Mitochondrial Expression of the Drosophila melanogaster Multisubstrate Deoxyribonucleoside Kinase Mol. Pharmacol., December 1, 2007; 72(6): 1593 - 1598. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
