| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Published online before print May 5, 2008
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Amgen Award Lecture |
From the Departments of Pathology and Genetics, Yale University School of Medicine, New Haven, Connecticut
Abstract
Mitochondria are central players in cellular energy metabolism and, consequently, defects in their function result in many characterized metabolic diseases. Critical for their function is mitochondrial DNA (mtDNA), which encodes subunits of the oxidative phosphorylation complexes essential for cellular respiration and ATP production. Expression, replication, and maintenance of mtDNA require factors encoded by nuclear genes. These include not only the primary machinery involved (eg, transcription and replication components) but also those in signaling pathways that mediate or sense alterations in mitochondrial function in accord with changing cellular needs or environmental conditions. Mutations in these contribute to human disease pathology by mechanisms that are being revealed at an unprecedented rate. As I will discuss herein, the basic protein machinery required for transcription initiation in human mitochondria has been elucidated after the discovery of two multifunctional mitochondrial transcription factors, h-mtTFB1 and h-mtTFB2, that are also rRNA methyltransferases. In addition, involvement of the ataxia-telangiectasia mutated (ATM) and target of rapamycin (TOR) signaling pathways in regulating mitochondrial homeostasis and gene expression has also recently been uncovered. These advancements embody the current mitochondrial research landscape, which can be described as exploding with discoveries of previously unanticipated roles for mitochondria in human disease and aging.
Mitochondria are complex and dynamic organelles that are absolutely required for the development, function, and longevity of virtually all eukaryotic cells and organisms. Mitochondria are central players in metabolism by virtue of harboring dozens of enzymes within their double-membrane structure, including, for example, the TCA cycle and those involved in the catabolism or biosynthesis of fatty acids, amino acids, heme, and steroids. Genetic defects that result in altered expression or activity of such enzymes cause many classic inborn errors of metabolism. An important subset of these involves the oxidative phosphorylation (OXPHOS) complexes I to V in the inner mitochondrial membrane that produce ATP, the most often touted primary function of mitochondria. In humans, these complexes comprise 
80 protein subunits, 13 of which are encoded by maternally inherited mtDNA.1
Thus, unlike most other mitochondrial metabolic diseases, which are inherited in a Mendelian manner, OXPHOS disorders can also be strictly maternally inherited via mutations in mtDNA. Somatic mutations in mtDNA also accumulate in tissues with age and are thought to contribute to age-related disease pathology and the aging process itself.2-7
Although defective energy metabolism resulting from loss of ATP production is certainly a common attribute of most OXPHOS diseases, it is not usually a stand-alone feature. In fact, there are a number of additional salient downstream consequences that stem from the nature of the electron transfer processes that underlie the OXPHOS mechanism, as well as the many additional functions of the organelles, which reach well beyond metabolism per se. These may include the enhanced production of damaging reactive oxygen species (ROS) by the electron transport chain (usually complex I or III),2 aberrant apoptosis,8 and altered ion homeostasis.9 Finally, it has become evident that cellular signaling pathways sense and control mitochondrial function,10 providing yet another potential pathogenic consequence of OXPHOS disruption, defective signal transduction, that is only beginning to be understood.
Expression and Maintenance of Human mtDNA: Basic Principles
In humans, mtDNA is a double-stranded circle of 16,569 bp that encodes 37 genes: 13 mRNAs, 2 rRNAs, and 22 tRNAs.1 The mRNAs specify essential integral membrane components of the OXPHOS complexes, the rRNAs are subunits of mitochondrial ribosomes, and the tRNAs mediate translation of the 13 mRNAs by these dedicated mitochondrial ribosomes. Genes are located on both mtDNA strands, which are called the heavy (H) strand and light (L) strand based on their relative buoyant densities in denaturing CsCl gradients. Transcription initiates at one of two promoters on the H-strand (called HSP1 and HSP2) and a single L-strand promoter (the LSP). The LSP and HSP1 are located in the major noncoding region of the molecule called the displacement loop (D-loop) regulatory region, whereas HSP2 is located downstream of HSP1 within the tRNAPhe gene. Transcripts derived from HSP2 and LSP can be near genome-length polycistronic products, whereas those from HSP1 are terminated at a specific site (in the tRNALeu gene) downstream of the 16S rRNA and produce mainly the two rRNAs (12S and 16S). Interestingly, this termination event is linked physically to initiation at HSP1 mediated by simultaneous binding of the termination factor, mTERF, to the termination site and the promoter, thus forming a regulatory loop.11 Because of the unique gene arrangement in mtDNA (ie, the rRNAs and most mRNAs are immediately flanked by tRNAs), tRNA processing is believed to be the major mechanism that liberates the majority of the 37 mature RNA molecules from the polycistronic primary transcripts. In mammals, the mature mRNAs are devoid of significant 5'-untranslated sequences, and thus the mechanism of ribosome binding and mitochondrial translation initiation remains obscure.
In mammalian cells, mtDNA is maintained at a high copy number. Generally, there are thousands of copies per cell with the precise number apparently regulated by tissue-specific factors. Within the mitochondrial matrix complexes of 2 to 10 mtDNA molecules are packaged into nucleoprotein complexes called nucleoids that are primarily inner-membrane associated.12-14 Thus, each nucleoid-containing mitochondrion by definition has multiple mtDNA copies. However, the notion of specifying the number of mtDNA molecules/mitochondrion is a vague one given that mitochondria usually form a dynamic, branched network that can fuse, divide, and intermix components, including nucleoids.15
There are additional hallmark features of mammalian mtDNA that are worthy of discussion. First, in addition to the three promoters there are several other cis-acting sequences that are of regulatory significance. For example, in the D-loop regulatory region there are four sequence elements (conserved sequence blocks: CSB I, CSB II, CSB III; and origin of H-strand synthesis: OH) that are postulated to be important for initiation of transcription-primed, leading-strand DNA synthesis according to the asymmetric model of mtDNA replication.16
These elements are downstream of the LSP and are involved in configuring and processing the LSP transcript to form RNA primers for initiation by the mtDNA polymerase, Pol 
.17
Therefore, the RNA primers for leading-strand mtDNA replication are generated by POLRMT (the mitochondrial RNA polymerase). Second, approximately two-thirds the distance around the mtDNA molecule from OH is OL, which is a primary site of initiation of lagging-strand mtDNA synthesis, according to the asymmetric replication model. This site lies in an unusual region of the genome where there is a clustering of five adjacent tRNA genes. Although this site is a major region of initiation of lagging-strand synthesis, it appears that other sites on the molecule may serve as alternative initiation sites.18
Third, one of the first features of mtDNA to be recognized is the D-loop itself. This is a stable three-stranded DNA structure of 570 to 665 nucleotides in length (in humans) that begins at OH and extends downstream where it ends at a few distinct sites.19
This structure has all of the features of a stalled (or terminated) leading-strand replication intermediate; however, whether this is actually the case has not been strictly determined. Since its discovery, the significance of the need to maintain this structure in a subset (a large subset in some cells) of mtDNA molecules has eluded the field, but it is logical to assume it is of regulatory importance with regard to expression, replication, and/or inheritance of mtDNA. Lastly, in addition to the asymmetric model of mammalian mtDNA replication (on which I have expounded above) other models have been proposed.20-22
These remain controversial and have been debated in the literature.23-25
Clearly, additional molecular and mechanistic studies are merited to test all of these models critically and to address the possibility that multiple mechanisms may be involved that operate under different conditions or in a tissue-specific manner.
With regard to regulation of mtDNA expression and maintenance, a key point to re-emphasize is that, except for the mtDNA-encoded rRNAs and tRNAs, all of the factors required for transcription, RNA processing, translation, replication, and repair of mtDNA are encoded by nuclear genes, translated by cytoplasmic ribosomes, and imported into mitochondria to their sites of action. In other words, there is an important and relatively large subset of the 1500 nucleus-encoded proteins in the mitochondrion that is devoted to mitochondrial gene expression and mtDNA maintenance. From this situation it follows that signaling pathways must exist to coordinate the activities of these distinct genetic compartments (the nucleus and mitochondria) to maintain and modulate mitochondrial gene expression and OXPHOS activity, not to mention the many other functions of these amazing organelles. These mitochondrial regulatory factors and pathways are yet another potential underlying cause of human disease. In fact, pathogenic mutations in nuclear genes encoding proteins required for mitochondrial translation26,27
and mtDNA maintenance28
are well documented. One goal of this review is to summarize published evidence that the human mitochondrial transcription machinery and signaling pathways that regulate mitochondrial gene expression and DNA maintenance contribute to human disease pathology and aging in unprecedented ways, as described below.
The Core Human Mitochondrial Transcription Machinery Is a Mixed Three-Component System
Transcription of human mtDNA is directed by a dedicated mitochondrial RNA polymerase, POLRMT, that is a member of the bacteriophage T3/T7 family of single-subunit RNA polymerases.29
Although the prokaryotic origin of mitochondrial RNA polymerases was perhaps expected given the bacterial ancestry of the organelle, it was nonetheless surprising when the first mitochondrial RNA polymerase to be cloned and sequenced (that of budding yeast) was found to be homologous to these bacteriophage enzymes as opposed to the multisubunit bacteria RNA polymerases themselves. The carboxy-terminal portion of the 140-kDa POLRMT contains the conserved bacteriophage-related sequence motifs that compose the catalytic domain based on structure-function studies of the phage enzymes.29
Unlike the bacteriophage RNA polymerases, however, POLRMT has a large amino-terminal extension that contains two pentatricopeptide repeat (PPR) domains, which are conserved only in vertebrate mtRNA polymerases.30
The function of the PPR domains and the rest of the amino terminal domain of POLRMT remains to be determined, but likely involves coupling other processes (eg, RNA processing or translation) to transcription as we have shown for the amino terminal domain of yeast mitochondrial RNA polymerase.30-33
Such a function is also logical to propose based on the fact that other PPR domain proteins have documented roles involved in RNA regulation.34-36
Also unlike bacteriophage RNA polymerases, which do not require additional protein factors to proceed through the various stages of transcription, most, if not all, mitochondrial RNA polymerases require associated transcription factors for function. The first mammalian mitochondrial transcription factor to be identified was human mitochondrial transcription factor A (h-mtTFA or TFAM), which was identified by its ability to facilitate specific promoter-driven transcription by partially purified POLRMT that alone had only nonspecific RNA polymerase activity.37,38 This function of h-mtTFA is mediated through its binding to DNA elements directly upstream of the LSP and HSP1 initiation sites.39 As a member of the high-mobility-group (HMG) box family of proteins,40 h-mtTFA binds DNA in a relatively sequence-nonspecific manner and bends and wraps DNA via the namesake HMG-box DNA binding domains of this class of proteins.41,42 In h-mtTFA there is a linker between the two HMG-boxes and a 25-amino acid C-terminal tail that is critical for specific DNA binding upstream of the promoters and its ability to stimulate specific transcription initiation.43
Studies of the mitochondrial transcription machinery in the budding yeast, Saccharomyces cerevisiae (referred to from here on as yeast), revealed the involvement of a second type of mitochondrial transcription factor that is unrelated to h-mtTFA, called sc-mtTFB (or Mtf1p by standard yeast nomenclature), which led to speculation about the existence of orthologs in higher eukaryotes.44
Although biochemical evidence for an mtTFB-like transcription factor activity in Xenopus was reported,45
we provided the first direct proof of a vertebrate ortholog of yeast mtTFB with the isolation of a cDNA encoding this transcription factor, now called h-mtTFB1 or TFB1M.46
Our subsequent studies revealed that h-mtTFB1 binds DNA nonspecifically and stimulates transcription initiation in vitro by binding to the C-terminal tail of h-mtTFA.46,47
Very soon after we identified h-mtTFB1, Falkenberg and colleagues48
reported the isolation of the same protein as well as a second human paralog called h-mtTFB2 (or TFB2M), which also interacts with the C-terminal tail of h-mtTFA.47
They also reconstituted specific mitochondrial transcription in vitro using h-mtTFA and a complex of POLRMT with either h-mtTFB1 or h-mtTFB2, with all proteins produced from recombinant sources.48
Altogether, these seminal studies defined the core human mitochondrial transcription system as a mixed three-component system (consisting of POLRMT, h-mtTFA, and either h-mtTFB1 or h-mtTFB2), which is absolutely required to achieve specific initiation at HSP1 and LSP (Figure 1)
. The transcription factor requirements for initiation at HSP2 have not been determined.
|
. In the case of h-mtTFB1, this enzymatic activity is not required for its transcription factor function in vitro,47
indicating these functions are separable. The biological significance of the rRNA methyltransferase activity of these transcription factors will be discussed in a subsequent section. The Mechanism of Human Transcription Initiation
Because we1
and others52
have recently described in detail the experimental results that form the basis for models describing the mechanism of transcription initiation at human mtDNA promoters, I will only summarize these here. The essence of the current model, for which there appears to be a general consensus in the field, is that promoter recognition and open complex formation is mechanistically similar to that of bacteriophage T7. However, instead of the polymerase itself being sufficient to recognize and melt the promoter, as is the case for T7, POLRMT requires the simultaneous presence and coordinated activities of h-mtTFA and either h-mtTFB1 or h-mtTFB2 (Figure 1)
. That is, consistent with its bacteriophage ancestry, POLRMT binds directly to critical promoter nucleotides and hence contributes significantly to the sequence specificity of initiation at mtDNA promoters.53
However, its ability to do so is dependent of several additional conditions being met, including it being in a 1:1 complex with either h-mtTFB1 or h-mtTFB2, the presence of promoter-bound h-mtTFA, and physical distortion of the promoter region into an initiation-competent configuration (eg, a particular bend or kink) (Figure 1)
. Currently, it is this last requirement that remains obscure. Specifically, whether the bending and unwinding of the promoter template occurs before or after promoter recognition by POLRMT is unknown. In the former case, the promoter could be presented to POLRMT in a unique configuration that allows it to be discerned, which could be provided by the unique DNA properties of the HMG boxes of h-mtTFA.41
In the latter case, POLRMT might be capable of binding the promoter unaided (like T7 RNA polymerase) whereas other factors assist subsequent initiation or stabilization of an open promoter complex. Finally, which of the three core components provides the promoter-melting function is unclear. Again, h-mtTFA could be the principle promoter modulator in this regard via its aforementioned unique DNA-distorting capabilities. However, a second possibility that is equally likely (and not mutually exclusive) is that h-mtTFB1 or h-mtTFB2 stabilizes the open complex formation by binding to the unwound template or nontemplate promoter DNA strands. Such a single-stranded DNA binding function could in principle be a functional manifestation of their rRNA methyltransferase ancestry, in that this activity likely requires the ability of h-mtTFB1 and h-mtTFB2 to bind single-stranded nucleic acid to a significant extent.46
Based on this synopsis, one can ascertain that many important mechanistic details have been elucidated with regard to the mechanism of human mitochondrial transcription. However, many important questions remain that are worthy of future study. The relatively simple nature of this system provides a facile model system to learn not only more about fundamental aspects of transcription but also which of the core transcription components are regulatory with regard to mitochondrial gene expression and function.
The h-mtTFB1 and h-mtTFB2 Mitochondrial Transcription Factors Retain Ancestral rRNA Methyltransferase Activity and Coordinately Regulate Mitochondrial Gene Expression, Activity, and Biogenesis
That mammals have two paralogs in the mtTFB family of mitochondrial transcription factors that are homologous to KsgA-type rRNA methyltransferases led to several immediate questions surrounding whether they actually had rRNA methyltransferase activity and how their dual functions were parsed with regard to specific effects on mitochondrial gene expression. In our original cloning of h-mtTFB1, we demonstrated that it binds the methyl-donating co-factor S-adenosylmethionine,46 providing the first clue that it might have enzymatic function. We subsequently showed that both h-mtTFB1 and h-mtTFB2 have this enzymatic activity as evidenced by their ability to methylate the homologous bacterial 16S rRNA substrate when expressed in a ksgA-null E. coli strain.49,51 Interestingly, h-mtTFB2 has significantly less activity on this heterologous substrate than h-mtTFB1,49 which is opposite of their relative transcription factor activity profiles.48 This poses the interesting possibility that both factors possess both functions but vary in their relative activities of each. Based on this premise, we proposed49 that differential expression of the two factors in tissues could be regulatory by allowing differential modulation of transcription/replication (via their different transcription factor activity) and translation (via their different rRNA methyltransferase activity). In this regard it is interesting to note that the relative abundance of h-mtTFB1 and h-mtTFB2 protein in HeLa cell mitochondria is not equal, but that their expression is coordinately regulated to some degree.54
To date only two studies have been published regarding the in vivo roles of the mtTFB1 and mtTFB2 proteins. In human HeLa cells or Drosophila Schneider cells, overexpression of mtTFB2 results in increased mtDNA copy number and steady-state levels of mitochondrial transcripts,54,55
whereas knockdown of Drosophila mtTFB2 by RNAi had the opposite effect on these parameters.55
Similar experiments with human and Drosophila mtTFB1 point to a primary role in mitochondrial translation and not transcription.54,56
These results are consistent with the simple interpretation that mtTFB1 is the primary rRNA methyltransferase (Figure 1)
and mtTFB2 is the primary transcription factor, as one might predict from their relative rRNA methyltransferase and transcription factor activity profiles. However, these studies do not formally eliminate the possibility of partially overlapping methyltransferase and transcription factor functions or that there is crosstalk between the two factors. With regard to the latter possibility, we found that overexpression of h- mtTFB1 in HeLa cells also resulted in an increase in mitochondrial mass, suggesting a novel role for h-mtTFB1 in inducing mitochondrial biogenesis in addition to its function in translation.54
Furthermore, overexpression of h-mtTFB2 results in a concomitant increase in the steady-state level of h-mtTFB1 and an increase in mitochondrial membrane potential along with an increase in mitochondrial mass.54
Altogether, these results hint at the exciting possibility that rRNA methylation (Figure 1)
is a previously unrecognized signal for mitochondrial biogenesis and that a retrograde pathway operates to ensure coordinate regulation of h-mtTFB1 and h-mtTFB2 to properly modulate mitochondrial biogenesis and function. The presence of such an intricate pathway might begin to explain why maintenance of two mtTFB paralogs has been selected for during evolution in most metazoans, including humans. Finally, in addition to the linkage of h-mtTFB1 and h-mtTFB2 to rRNA methyltransferases, another connection between mitochondrial transcription and translation is the binding of mitochondrial ribosomal protein L12 (MRPL12) directly to POLRMT.57
This interaction may couple transcription and translation directly or provide a means to coordinate transcription of mtDNA-encoded rRNAs with the import of nucleus-encoded ribosomal mitochondrial ribosomal proteins. In either case, this poses the interesting possibility that key aspects of mitochondrial ribosome assembly are monitored by cells, perhaps as a signal to determine the rate of overall mitochondrial biogenesis or as a metric of mitochondrial functional capacity.
The h-mtTFB1 Gene Is a Nuclear Modifier of Maternally Inherited Deafness
Several mutations in mtDNA cause maternally inherited deafness.58,59 The most extensively characterized of these is the relatively common A1555G mutation in the 12S rRNA gene associated primarily with nonsyndromic and/or aminoglycoside antibiotic-induced deafness.60 Early pedigree analysis and cell culture studies with mutant patient-derived cells revealed that the associated deafness and mitochondrial phenotypes are influenced strongly by the nuclear genetic background.58,61-63 In fact, several nuclear modification loci of the A1555G mutation have been described.64-67 These nuclear genetic background influences likely explain some of the variability in individuals within and between A1555G deafness pedigrees in the age of onset of hearing loss and resistance to the ototoxicity of aminoglycosides.
After our isolation of h-mtTFB1 and demonstration that, in addition to acting as a transcription factor, it also methylates a conserved stem-loop in mitochondrial 12S rRNA46,51
(Figure 1)
, Bykhovskaya and colleagues64
reported that a polymorphism near the h-mtTFB1 gene (TFB1M) provides a protective effect in individuals with the deafness-associated A1555G mutation. Although the mechanism through which h-mtTFB1 modifies the A1555G deafness phenotype remains unknown, it is likely that it is through its impact on mitochondrial translation either indirectly, via its transcription factor function, or directly, via its methylation of the 12S rRNA.68
At present, the latter seems most likely. This is because h-mtTFB1 methylates two adjacent adenine residues in a highly conserved stem-loop structure near the 3'-end of the 12S rRNA that, perhaps not coincidentally, is immediately downstream of where the A1555G point mutation occurs (Figure 1)
. The potential relevance of this relationship and h-mtTFB1 in mitochondrial deafness syndromes was originally noted by us51
based on the fact that expression of h-mtTFB1 in E. coli restored methylation of the 16S stem-loop and sensitivity to the antibiotic kasugamycin in a strain lacking KsgA (the bacterial ortholog h-mtTFB1). Given that the A1555G mutation also predisposes individuals to aminoglycoside antibiotic-induced deafness,58,60
we hypothesized (based on the bacterial analogy) that the methylation status of the 12S rRNA in human mitochondria could modify the phenotype of the A1555G mutation. This idea was promoted further by Bykhovskaya and colleagues64
who proposed that the chromosome 6 marker near the TFB1M gene provides its protective effect by altering h-mtTFB1 activity. For example, they suggest that decreased expression of h-mtTFB1 could cause a corresponding decrease in 12S rRNA methylation that suppresses the loss of ribosomal function resulting from the A1555G mutation. This seems plausible because both the A1555G mutation and the methylation status of this stem-loop alter the structure of the rRNA in the same domain of the ribosome.60,69
Thus, an altered and malfunctioning ribosome conformation imposed by the A1555G mutation could, in principle, be restored by lack (or alteration) of methylation of the nearby stem-loop. In this manner, lower or altered h-mtTFB1-driven rRNA methylation would reduce the penetrance of the deafness phenotype. That the A1555G mutation increases or decreases rRNA methylation and hence plays a role in the deafness phenotype also remains a distinct possibility. Testing molecular models such as these is key to understanding how mtDNA mutations cause deafness in the first place and how nuclear modifiers such as h-mtTFB1 operate to modulate their pathogenic consequences.
Regulation of Mitochondrial Gene Expression and Respiration by the TOR Pathway Limits Life Span
Mitochondria have long been implicated in aging, and many reviews on the mitochondrial theory of aging have been published. The mitochondrial theory2,70 is often married to the so-called free radical theory of aging,71 because of the fact that ROS, an important source of biological free radicals, are produced by mitochondrial respiration and are postulated to damage cellular components and ultimately to lead to loss of normal cell and tissue function that underlies the aging process. In fact, a vicious cycle of mitochondria-driven oxidative stress, whereby mitochondrial ROS damage components of the resident respiratory chain, which, in turn, leads to even more mitochondrial ROS production, is often postulated as a key feature of aging and age-related pathology.3,7,72 Thus the acquisition of mitochondria might represent an example of antagonistic pleiotropy, where the obvious early growth benefits of harnessing the energy in nutrients via oxygen-mediated catabolism are ultimately off set by the deleterious consequences of oxygen-mediated damage throughout the course of an individual’s life time. Although weighing the pros and cons of this theory is beyond the scope of this review, I will expound on recent results from aging studies in yeast that point to a key role for respiration in regulating life span and some of the first direct evidence for the operation of the vicious cycle of mitochondrial ROS production in its regulation. Interestingly, these studies point to oxidative damage of factors other than mtDNA (eg, the OXPHOS system itself) as a major contributor to aging in yeast, which is perhaps relevant to finding that increased ROS and oxidative damage are not observed in mice engineered to accumulate mtDNA mutations at a high rate.3,6,8
Because of its conditional ability to survive in the absence of mitochondrial respiration (ie, if grown on a fermentable carbon source), yeast has remained an invaluable model organism for understanding mitochondrial function.73,74 Yeast has also come to the forefront of aging research, by providing investigators two experimental approaches to measure life span and characterize genes and pathways involved.75 The first is replicative life span, which is defined as the number of times that a mother cell can divide (ie, bud) before senescence, and the second is chronological life span (CLS), which represents the amount of time that cells remain viable under nondividing conditions (eg, in stationary phase cultures). There is now substantial evidence that oxidative stress and mitochondrial respiration are involved in both forms of yeast aging. For example, increased expression of the ROS-detoxifying enzymes superoxide dismutase (encoded by the genes SOD1 and SOD2) results in extended CLS, whereas deletion of these genes has the opposite effect.76,77 In addition, increased mitochondrial respiration is required for the longevity-promoting effects of caloric restriction on replicative lifespan78 and CLS,79 whereas defective respiration can increase80 or decrease81 CLS depending on the precise nature of the mitochondrial defect. Similar effects of respiration and oxidative-stress resistance on life span have been observed in other model eukaryotes.82-87 Despite these correlations, what remains unclear is the precise relationship between respiration rate, mitochondrial ROS production, and life span.3 That is, higher rates of respiration can result in higher or lower rates of mitochondrial ROS production, and depending on the model system and conditions studied, life span can be increased or decreased. This dispels the commonly stated view that higher rates of respiration necessarily increase ROS production and curtail life span (promote aging phenotypes). Our recent work on the role of the TOR pathway in regulating yeast CLS is a salient case in point.88
It has been shown in multiple model systems that down-regulation of the pro-growth TOR signaling pathway extends life span.89-91 To examine the mechanism underlying this conserved response, we examined yeast strains in which TOR signaling was reduced via deletion of the TOR1 gene (encoding one of the two TOR kinases in this organism) and showed that these strains have an approximately threefold increase in median CLS compared to their wild-type counterparts.88 However, the more significant finding from this study is that the ability of reduced TOR signaling to increase CLS absolutely required mitochondrial respiration. Specifically, deletion of TOR1 results in increased translation and steady-state levels of mtDNA-encoded respiratory chain enzyme subunits and elevated rates of mitochondrial respiration. Furthermore, the increased respiration and the extended lifespan of tor1-null cells are glucose- and oxygen-dependent. Our conclusion from these results is that TOR normally inhibits respiration and that ROS-mediated damage (or some other oxygen-dependent phenomenon) during glucose-dependent growth limits stationary phase survival (ie, CLS). We went on to speculate that the salient life span-limiting parameter was ROS-mediated damage to the mitochondrial components themselves, setting into motion a vicious cycle of oxidative stress.3,88 This, however, awaits further experimental confirmation. Finally, it was recently reported that the mammalian TOR (mTOR) pathway regulates respiration in cultured human Jurkat T cells;92 thus, that mTOR signaling limits life span or impacts aging phenotypes in mammals via its effects on respiration and ROS production remains an intriguing possibility. The fact that mTOR and one of its negative regulators is physically associated with the mitochondrial outer membrane is engaging in this regard.92,93
Beyond Salvage: dNTPs Derived from the de Novo Ribonucleotide Reductase Pathway Contribute to mtDNA Maintenance
It is obvious that mitochondria require a constant source of deoxynucleoside triphosphates (dNTPs) to allow continuous replication and repair of their mtDNA. In contrast to nuclear DNA replication, which occurs exclusively in S phase, mtDNA replication is not strictly cell cycle-dependent. In fact, mtDNA replication occurs in all stages of the cell cycle and persists even in nondividing cells.19,94 For these reasons, it has been generally accepted that deoxynucleotide salvage pathways provide the dNTPs for mtDNA replication, a notion that is supported by the presence of salvage pathway enzymes (eg, deoxynucleoside and deoxynucleotide kinases) in mitochondria.95,96 Furthermore, that salvage pathways are critical for mtDNA maintenance is evidenced convincingly by inherited human mtDNA depletion syndromes, a subset of which is caused by mutations in genes encoding mitochondrial and cytoplasmic deoxynucleotide salvage enzymes.28,95,96 Nonetheless, early observations pointed to the potential involvement of the de novo pathway of dNTP synthesis in mtDNA replication and maintenance,96 that is, dNTPs derived by ribonucleotide reduction by the enzyme ribonucleotide reductase (RNR). Although a report of a mitochondria-localized form of RNR surfaced,97 it has not been verified or generally accepted in the field. In fact, the presence of transporter proteins that facilitate import of deoxynucleosides (for salvage phosphorylation) and presumably also deoxynucleotides (from cytoplasmic RNR reactions) would seem to obviate the need for bona fide mitochondrial ribonucleotide reduction.95
Substantial evidence from budding yeast has amassed showing that alterations in the activity or abundance of large R1 subunit of RNR have corresponding effects on mtDNA copy number and stability. For example, overexpression of RNR1 rescues mtDNA instability caused by point mutations in or haploinsufficiency of Mip1p, the yeast mtDNA polymerase.98 In addition, we have shown that strains that overexpress or harbor an activating point mutation in RNR1 or that recapitulate increased flux through the Mec1-Rad53 signaling pathway (which increases RNR expression and activity) have increased amounts and stability of mtDNA.99-101 Conversely, others have shown that defective Mec1-Rad53 pathway signaling decreases mtDNA stability (ie, increases petite mutant formation).102,103 Taken together these results clearly point to RNR pathway-driven alterations in the cytoplasmic deoxynucleotide pool as a significant parameter that can influence mtDNA replication and/or stability. Whether these effects are levied at the level of replication, repair, or stability of mtDNA remains an open question. At the time, the relevance of these results to mammalian cells was also questionable given that yeast does not have deoxynucleotide salvage pathways104 and is forced to obtain dNTPs for mtDNA replication and repair from the de novo pathway. However, that yeast Mec1p and Rad53p have orthologous signaling kinases in mammalian cells, ATM/ATR and CHK1/CHK2,105 respectively, led us to propose that this pathway may represent a conserved mechanism through which mtDNA replication and stability are modulated.101
Recent reports have shed significant new insight into the role of the de novo RNR pathways in mammalian mtDNA maintenance. For example, there is good evidence for ribonucleotide reduction in resting cultured human cells,106 conditions previously thought to be under the purview of the salvage pathways alone. In addition, even in noncycling cells, the R1 and p53R2 subunits of RNR make a small, but significant contribution to the dNTP pool that, in principle, could contribute to mtDNA replication and repair (as well as basal nuclear repair).107 This hypothesis was unequivocally confirmed by the recent report by Bourdon and colleagues,108 who found linkage of mutations in the p53R2 subunit of RNR in mitochondrial disease patients with severe mtDNA depletion in muscle. Finally, we recently reported that pharmacological inhibition of RNR in wild-type human fibroblasts or RNAi-mediated inhibition of the R1 or R2 subunits of RNR in HeLa cells results in mtDNA depletion.109 Thus, it has become quite evident that both salvage and de novo pathways contribute significantly to mtDNA replication and maintenance in mammalian cells. Furthermore, it appears that RNR-derived dNTPs are used for mtDNA replication in both cycling and noncycling cells. This poses the exciting possibility that different cell and tissue types have evolved to depend differentially on these two pathways for replication and repair of mtDNA. Understanding the underlying dynamics involved may finally shed some light onto how tissues maintain characteristic amounts of mtDNA and why mtDNA depletion syndromes display such exquisite tissue-specific pathology. Finally, based on these new concepts, it is logical to predict that the genes encoding the R1 and R2 subunits of RNR, like p53R2,108 will represent disease loci for mtDNA depletion syndromes or other mtDNA-based diseases.
A Newly Uncovered Role for ATM in RNR and Mitochondrial Homeostasis May Contribute to the Complex Pathology of Ataxia-Telangiectasia (A-T)
ATM is a serine/threonine protein kinase (related to the phosphoinositide 3-kinases) with a well documented role in sensing nuclear DNA damage.110,111
In particular, it responds to double-strand breaks in DNA and initiates a signaling cascade that leads to cell cycle arrest and DNA repair or, under some circumstances, apoptosis. However, ATM and the related kinase ATR are orthologs of yeast Mec1p, which, as discussed in the previous section, signals to RNR and modulates mtDNA copy number and stability.99,101
This led us to investigate whether such a function for ATM was conserved and likewise involved in mtDNA maintenance and dynamics in mammals. To this end we first analyzed A-T patient fibroblasts that have a null mutation in ATM for disruption in mtDNA metabolism. Here, we found that A-T cells exhibit conditional mtDNA depletion even in the absence of DNA damage and failed to promote the RNR-dependent increase in mtDNA when DNA damage was induced by ionizing radiation.109
These cells also have significant disruptions in the steady-state levels of all three RNR subunits in the presence or absence of DNA damage as well as other disruptions in mitochondrial homeostasis. Tissue-specific alterations in mtDNA copy number were also observed in ATM-null mouse tissues, as was a general reduction of the R1 subunit of RNR in all tissues examined.109
Additional evidence for ATM’s involvement in mitochondrial homeostasis was reported previously by Stern and colleagues112
and also recently by Ambrose and colleagues.113
Thus, ATM clearly has a role in regulating RNR and mitochondrial function under normal growth conditions as well as in response to DNA damage (Figure 2)
. Some of the downstream consequences of ATM action are mediated by p53,114,115
including the induction of p53R2 expression in response to DNA damage.116,117
Given that p53 is also implicated in the regulation of mitochondrial respiration118-121
and mtDNA maintenance,122-124
it is likely that some of the effects of loss of ATM on mitochondria we and others have reported are attributable to disruptions of p53 function (Figure 2)
.
|
, a disease with a wide range of clinical symptoms.125
In addition to the hallmark ataxia, because of degeneration of cerebellar neurons, A-T patients are prone to lymphoma, sterility, diabetes, and premature aging phenotypes. With the exception of lymphoma, all of these symptoms are commonly observed in patients with mitochondrial diseases,125-128
supporting the notion that A-T may at some level be a mitochondrial disorder. Perhaps even more compelling, however, is that oxidative stress and aberrant apoptosis are key features of the disease,129,130
which are commonly invoked mechanisms of mitochondrial-driven pathology as already discussed. Thus, one model that is worthy of future investigation is that the underlying mitochondrial dysfunction in A-T patient cells and tissues leads to enhanced mitochondrial ROS production, setting into motion a vicious cycle of oxidative stress that causes tissue damage and dysfunction in A-T patients (Figure 2)
. Such a scenario would presumably be exacerbated by the loss of the normal nuclear DNA damage response in the absence of ATM. Finally, the reported disruptions in RNR in the absence of ATM109
may contribute to the observed pathology as well, either by causing the mitochondrial dysfunction (attributable to mtDNA copy number and stability defects) or by negatively impacting the nuclear DNA damage response (through inappropriate dNTP pool management). Although, at this point, these ideas should be considered in the realm of speculation, they are certainly deserving of serious consideration given they hold the promise for novel therapeutic strategies for this devastating disease. New Insights into Mitochondrial Disease Pathology and Aging
Although mitochondrial diseases resulting from mtDNA mutations are relatively rare, when nuclear mutations that cause mitochondrial dysfunction are also considered, the incidence of primary mitochondrial diseases is estimated to be at least 1:8000 births.131 However, undoubtedly, even this is a major underestimate of the role of mitochondria in human disease. Mitochondria are being implicated in a multitude of cellular processes at an unprecedented rate, which brings into view a corresponding increase in new mechanisms through which mitochondrial dysfunction can contribute to human disease pathology. Our studies directed toward deciphering the basic mechanism of mitochondrial gene expression and mtDNA maintenance have almost invariably led us to novel disease connections and insights into aging and the regulation of life span. This indicates to me that the contribution of mitochondria to human health and aging is grossly underestimated and that the mechanisms involved extend far beyond the stereotypical metabolic maladies typically associated with these organelles. As we continue to learn their basic properties, diverse roles in cellular homeostasis, and complex pathogenic mechanisms in the coming years, mitochondria will likely represent a rational therapeutic target not only for primary mitochondrial diseases but also for more common health problems such as diabetes, heart disease, common neurodegenerative diseases, cancer, and age-related pathology.7,132
Acknowledgements
First, I thank Amgen for continuing to support the Outstanding Investigator Award and the American Society for Investigative Pathology awards committee for their recognition of my achievements as worthy of this accolade. Second, I am greatly indebted to many people in the scientific half of my life who have provided overwhelming positive influences on my thinking, career development, and general state of well-being. Unfortunately, this list of names is too long to state in its entirety here, but certainly includes my primary scientific mentors, David Clayton and Thomas Baldwin, whose training and friendship have been invaluable, Paul Doetsch, Laurie Kaguni, Dave Lambeth, Jon Morrow, Mark Schmitt, contemporaries in the Baldwin and Clayton laboratories, the "yeast" club, and the Stanford Crowd. Third, I must acknowledge the hard work, dedication, and creativity of the past and present members of my own laboratory and our collaborators. Finally, I have been very fortunate to have the sincere and loving support of my parents and family, without which no success is possible or meaningful. In this regard, I extend a special acknowledgment to my wife Susan Kaech who has provided me constant love, support, and encouragement while, at the same time, being a superb scientist and colleague who has been a major positive influence on my science.
Footnotes
Address reprint requests to Gerald S. Shadel, Departments of Pathology and Genetics, Yale University School of Medicine, 310 Cedar St., P.O. Box 208023, New Haven, CT 06520-8023. E-mail: gerald.shadel{at}yale.edu
This review summarizes work on multiple projects in my laboratory throughout the last 10 years that were supported by grants from the National Institutes of Health (HL-059655, ES-011163, and NS-056206), the Army Research Office (DAAD19-00-1-0560), the Glenn/AFAR BIG award, the A-T Children’s Project, the Robert Leet and Clara Guthrie Patterson Trust, and the National Organization for Hearing Research Foundation.
The ASIP-Amgen Outstanding Investigator Award is given by the American Society for Investigative Pathology to recognize excellence in experimental pathology research. Gerald S. Shadel, a recipient of the 2007 Amgen Outstanding Investigator Award, delivered a lecture entitled "Expression and Maintenance of Mitochondrial DNA: New Insights into Human Disease Pathology," on April 30, 2007 at the annual meeting of the American Society for Investigative Pathology in Washington, DC.
Accepted for publication February 5, 2008.
References
This article has been cited by other articles:
 |
|
 |
|
  J. Cotney, S. E. McKay, and G. S. Shadel Elucidation of separate, but collaborative functions of the rRNA methyltransferase-related human mitochondrial transcription factors B1 and B2 in mitochondrial biogenesis reveals new insight into maternally inherited deafness Hum. Mol. Genet., July 15, 2009; 18(14): 2670 - 2682. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. A. Gangelhoff, P. S. Mungalachetty, J. C. Nix, and M. E. A. Churchill Structural analysis and DNA binding of the HMG domains of the human mitochondrial transcription factor A Nucleic Acids Res., June 1, 2009; 37(10): 3153 - 3164. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Mitsuishi, K. Miyashita, A. Muraki, and H. Itoh Angiotensin II Reduces Mitochondrial Content in Skeletal Muscle and Affects Glycemic Control Diabetes, March 1, 2009; 58(3): 710 - 717. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
