A Mitochondrial Paradigm of Metabolic and Degenerative Diseases, Aging, and Cancer: A Dawn for Evolutionary Medicine Douglas C. Wallace Center for Molecular and Mitochondrial Medicine and Genetics, Departments of Ecology and Evolutionary Biology, Biological Chemistry, and Pediatrics, University of California, Irvine 92697-3940 email:dwallace@uci.edu Abstract Life is the interplay between structure and energy, yet the role of energy deficiency in human disease has been poorly explored by modern medicine. Since the mitochondria use oxidative phosphorylation (OXPHOS) to convert dietary calories into usable energy, generating reactive oxygen species (ROS) as a toxic by-product, I hypothesize that mitochondrial dysfunction plays a central role in a wide range of age-related disorders and various forms of cancer. Because mitochondrial DNA (mtDNA) is present in thousands of copies per cell and encodes essential genes for energy production, I propose that the delayed-onset and progressive course of the age-related diseases results from the accumulation of somatic mutations in the mtDNAs of post-mitotic tissues. The tissue-specific manifestations of these diseases may result from the varying energetic roles and needs of the different tissues. The variation in the individual and regional predisposition to degenerative diseases and cancer may result from the interaction of modern dietary caloric intake and ancient mitochondrial genetic polymorphisms. Therefore the mitochondria provide a direct link between our environment and our genes and the mtDNA variants that permitted our forbears to energetically adapt to their ancestral homes are influencing our health today. Keywords mitochondria reactive oxygen species human origins diabetes neurodegenerative diseases aging mtDNA: mitochondrial DNA Mitochondrion (s), mitochondria (pl): cellular organelle of endosymbiotic origin that resides in the cytosol of most nucleated (eukaryotic) cells and which produces energy by oxidizing organic acids and fats with oxygen by the process oxidative phosphorylation (OXPHOS) and generates oxygen radicals (reactive oxygen species, ROS) as a toxic by-product ROS: reactive oxygen species, oxygen radicals OXPHOS: oxidative phosphorylation Mitochondrial DNA (mtDNA): the portion of the mitochondrial genome that currently resides in the matrix of the mitochondrion, as a circular DNA molecule containing the mitochondrial rRNA genes, tRNA genes, and 13 subunits of the mitochondrial oxidative phosphorylation (OXPHOS) enzyme complexes CR: mtDNA control region TCA: mitochondrial tricarboxylic acid cycle SDH: succinate dehydrogenase COX: cytochrome c oxidase, complex IV ETC: mitochondrial electron transport chain, a part of the OXPHOS system ANT: adenine nucleotide translocator Unc 1,2,3: uncoupling proteins 1,2,3 Reactive oxygen species (ROS): primarily superoxide anion (O2��� ���), hydrogen peroxide (H2O2), and hydroxyl radical (��� OH), commonly referred to as oxygen radicals generated as a toxic by-product of oxidative energy production by OXPHOS damage the mitochondrial and cellular DNA, proteins, lipids, and other molecules causing oxidative stress Oxidative phosphorylation (OXPHOS): the process by which the mitochondrion generates energy through oxidation of organic acids and fats with oxygen to create a capacitor [electron chemical gradient (��P = ���� + ��pH)] across the mitochondrial inner membrane. This ��P is used as a source of potential energy to generate adenosine triphosphate (ATP), transport substrates or ions, or produce heat. OXPHOS encompasses five multipolypepetide complexes I, II, III, IV and V. Complex I is NIH Public Access Author Manuscript Annu Rev Genet. Author manuscript available in PMC 2010 February 12. Published in final edited form as: Annu Rev Genet. 2005 39: 359. doi:10.1146/annurev.genet.39.110304.095751. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

NADH dehydrogenase or NADH:ubiquinone oxidoreductase, complex II is succinate dehydrogenase (SDH) of succinate:ubiquinone oxidoreductase, complex III is the bc1 complex or ubiquinole: cytochrome c oxidoreductase,complex IV is cytochrome c oxidase (COX) or reduced cytochrome c: oxygen oxidoreductase, and complex V is the ATP synthase or proton-translocating ATP synthase. Complexes I, III, IV, and V encompass both nDNA- and mtDNA-encoded subunits MnSOD (Sod2): mitochondrial matrix superoxide dismutase Cu/ZnSOD (Sod1): mitochondrial inner membrane space and cytosolic superoxide dismutase mtPTP: mitochondrial permeability transition pore Apoptosis: a process of programmed cell death resulting in the activation of caspase enzymes and intracellular nucleases that degrade the cellular proteins and nDNA. Apoptosis can be initiated via the mitochondrion through the activation of the mitochondrial permeability transition pore (mtPTP) in response to energy deficiency, increased oxidative stress, excessive Ca2+, and, or other factors CPEO/KSS: chronic progressive external ophthalmoplegia, Kearn-Sayre syndrome NARP: neurogenic muscle weaknes, ataxia, and retinitis pigmentosa LHON: Leber���s hereditary optic neuropathy MERRF: myoclonic epilepsy and ragged red fiber disease MELAS: mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes PGC-1: PPAR�� (peroxisome-proliferating-activated receptor ��) coactivator 1 FOXO: mammalian forkhead transcription factor APP: amyloid precursor protein Mitochondrial medicine: the new medical discipline that pertains to all clinical problems that involve the mitochondria, Evolutionary medicine: a clinical perspective that posits that many of the common clinical problems of today are rooted in adaptive genetic programs that permitted our human ancestors to survive in the environments which they confronted in the past LEAN DIETS, LONG LIFE, AND THE MITOCHONDRIA Age-related metabolic and degenerative diseases are increasing to epidemic proportions in all industrialized countries. Diabetes has increased sharply over the past century, with a worldwide incidence in 2000 of 151 million cases and the projected incidence for 2010 of 221 million cases (for the United States, 14.2 million in 2000 versus 17.5 million by 2010) (73,257). Today, Alzheimer disease (AD) affects about 4.5 million Americans, projected to increase to 11 to 16 million by 2050 (6,70). Parkinson���s disease (PD) afflicts approximately one million persons in the United States today 60,000 new cases are diagnosed every year, and the incidence is projected to quadruple by 2040 (160,164). The risk of cancer also increases with age. Prostate cancer is the most common malignancy among men in industrialized countries. In the United States in 2004, approximately 230,000 new cases were reported, with an annual death toll of approximately 30,000 men (31). Moreover, the incidence of prostate cancer has been increasing steadily in the United States and Canada over the past 30 years (132). As many of these age-related diseases show some familial association, a genetic basis has been assumed. Yet, the nature and frequency of genetic variants in the human population has not changed significantly in the past 50 years, even though the incidence of these diseases has climbed continuously. Therefore, it isn���t our genes that have changed it is our environment, and the biggest environmental change that we have experienced is in our diet. For the first time in human history, individuals can live their entire lives free from hunger. Yet, it has been known for over 70 years that laboratory rodents, if maintained on a restricted calorie diet, will be healthier, more active, more intelligent, and will live longer and have fewer cancers (69,126,127,130,204). So what is the link between genetics and diet for these diseases? Prodigious effort has been expended to identify chromosomal genetic loci that are responsible for these problems, but with disappointingly little success. One hint as to why nuclear DNA (nDNA) analysis has been so unfruitful is that all of these diseases have a delayed onset and a progressive course. Thus while the most important risk factor for contracting all of these diseases is age, nothing in Wallace Page 2 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Mendelian genetics can provide the needed aging clock. The aging clock must involve the gradual accumulation of sequential changes in thousands of copies of the same gene or genes, yet each Mendelian gene is present in only two copies. Furthermore, certain age-related diseases are preferentially found in distinct populations, suggesting that they are influenced by regional genetic differences. Yet most ancient nDNA polymorphisms are common to all global populations. Only one human genetic system is known to be present in thousands of copies per cell, to exhibit striking regional genetic variation, and to be directly involved in calorie utilization: the mitochondrial genetic system. The mitochondria are ancient bacterial symbionts with their own mitochondrial DNA (mtDNA), RNA, and protein synthesizing systems. Each human cell contains hundreds of mitochondria and thousands of mtD-NAs. The mtDNA is maternally inherited and shows striking regional genetic variation. This regional variation was a major factor in permitting humans to adapt to the different global environments they encountered and mastered. Moreover, the mitochondria burn the calories in our diet with the oxygen that we breathe to make chemical energy to do work and heat to maintain our body temperature. As a by-product of energy production, the mitochondria also generate most of the endogenous reactive oxygen species (ROS) of the cell, and these damage the mitochondria, mtDNAs, and cell. Hence, the mitochondria are the only human genetic system that embodies the features necessary to explain the observed characteristics of the common age-related diseases (237). In this review, I make the case that the mitochondrial decline and mtDNA damage are central to the etiology of the age-related metabolic and degenerative diseases, aging, and cancer. This is because the rate of mitochondria and mtDNAs damage and thus decline is modulated by the extent of mitochondrial oxidative stress. Mitochondrial ROS production, in turn, is increased by the availability of excess calories, modulated by regional mtDNA genetic variation, and regulated by alterations in nDNA expression of stress response genes. MITOCHONDRIAL BIOGENESIS AND BIOENERGETICS: AN AGING PARADIGM The proto-mitochondrion entered the primitive eukaryotic cell between two and three billion years ago. Initially, the bacterial genome encoded all of the genes necessary for a free-living organism. However, as the symbiosis matured, many bacterial genes were transferred to extrabacterial plasmids (chromosomes), such that today the maternally inherited mtDNA retains only the genes for the 12S and 16S rRNAs and the 22 tRNAs required for mitochondrial protein synthesis plus 13 polypeptides of the mitochondrial energy generating process, oxidative phosphorylation (OXPHOS) (Figure 1). The remaining ~1500 genes of the mitochondrial genome are now scattered throughout the chromosomal DNA. These nDNA- encoded mitochondrial proteins are translated on cytosolic ribosomes and selectively imported into the mitochondrion through various mitochondrial protein import systems. For example, certain proteins destined for the mitochondrial matrix are synthesized with an amino terminal, positively charged, amphoteric targeting peptide that is cleaved off once the protein enters the mitochondrial matrix (237). The 13 mtDNA-encoded polypeptide genes are translated on mitochondrial ribosomes and all are structural subunits of OXPHOS enzyme complexes. These include 7 (ND1, 2, 3, 4L, 4, 5, 6) of the 46 polypeptides of complex I (NADH dehydrogenase), one (cytochrome b, cytb) of the 11 polypeptides of complex III (bc1 complex), 3 (COI, II, III) of the 13 polypeptides of complex IV (cytochrome c oxidase), and 2 (ATP 6 & 8) of the 16 proteins of complex V (ATP synthetase) (Figure 1). The nDNA codes for all other mitochondrial proteins including all four subunits of complex II (succinate dehydrogenase), the mitochondrial DNA polymerase �� Wallace Page 3 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

(POLG) subunits, the mitochondrial RNA polymerase components, the mitochondrial transcription factor (mtTFA), the mitochondrial ribosomal proteins and elongation factors, and the mitochondrial metabolic enzymes. The organization of the mtDNA is unique in that its structural genes have no 5��� or 3��� non-coding sequences, no introns, and no spacers between the genes. The mtDNA is symmetrically transcribed from two promoters, one for the G-rich heavy (H) stand and the other for the C- rich light (L) strand. These H- and L-strand promoters (PH and PL) are contained in the approximately 1121-np control region (CR) that encompasses four mtTFA binding sites, the H-strand origin of replication (OH), and three conserved sequence blocks (CSB I, II, and III). The L-strand origin of replication (OL) is two thirds of the way around the mtDNA from OH. Transcription is initiated at PH or PL and progresses around the mtDNA, generating a polycistronic message. The tR-NAs, which punctuate the genes, are cleaved out and the mRNAs and are then polyadenylated (237) (Figure 1). The mtDNA gene repertoire has remained relatively constant since the formation of the fungal- animal lineage. This is because the mtDNA genetic code began to drift in the fungi such that the mtDNA genes can no longer be interpreted by the nuclear-cytosol system (233). Consequently, when a mtDNA sequence is transferred to the nDNA, it remains as a nonfunctional pseudogene (147). The mitochondria generate energy by oxidizing hydrogen derived from our dietary carbohydrates (TCA cycle) and fats (��-oxidation) with oxygen to generate heat and ATP (Figure 2). Two electrons donated from NADH + H+ to complex I (NADH dehydrogenase) or from succinate to complex II (succinate dehydrogenase, SDH) are passed sequentially to ubiquinone (coenzyme Q or CoQ) to give ubisemiquinone (CoQH��� ) and then ubiquinol (CoQH2). Ubiquinol transfers its electrons to complex III (ubiquinol: cytochrome c oxidoreductase), which transfers them to cytochrome c. From cytochrome c, the electrons flow to complex IV (cytochrome c oxidase, COX) and finally to ��O2 to give H2O. Each of these electron transport chain (ETC) complexes incorporates multiple electron carriers. Complexes I, II, and III encompass several iron-sulfur (Fe-S) centers, whereas complexes III and IV encompass the b + c1 and a + a3 cytochromes, respectively. The mitochondrial TCA cycle enzyme aconitase is also an iron-sulfur center protein (234,235,237). The energy released by the flow of electrons through the ETC is used to pump protons out of the mitochondrial inner membrane through complexes I, III, and IV. This creates a capacitance across the mitochondrial inner membrane, the electrochemical gradient (��P = ����+ ��pH). The potential energy stored in ��P is coupled to ATP synthesis by complex V (ATP synthase). As protons flow back into the matrix through a proton channel in complex V, ADP and Pi are bound, condensed, and released as ATP. Matrix ATP is then exchanged for cytosolic ADP by the adenine nucleotide translocator (ANT) (Figure 2). Because the ETC is coupled to ATP synthesis through ��P, mitochondrial oxygen consumption rate is regulated by the matrix concentration of ADP. In the absence of ADP, the consumption of oxygen is regulated by proton leakage across the inner membrane and thus is slow (state IV respiration). However, when ADP is added, it binds to the ATP synthase and is rapidly converted into ATP at the expense of ��P. As protons flow through the ATP synthase proton channel, the proton gradient is depolarized. Stored fats and carbohydrates are then mobilized to provide electrons to the ETC, which reduce oxygen to water and pump the protons back out of the mitochondrial matrix. The resulting increased oxygen consumption on addition of ADP is known as state III respiration. ��P is also used for the import of cytosolic proteins, substrates, and Ca2+ into the mitochondrion. Drugs such as 2,4-dinitrophenol (DNP) and nDNA-encoded uncoupler proteins 1, 2, and 3 (Unc1, 2, and 3) render the mitochondrial inner membrane leaky Wallace Page 4 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

for protons. This short-circuits the capacitor and ���uncouples��� electron transport from ATP synthesis. This causes the ETC to run at its maximum rate, causing maximum oxygen consumption and heat production, but diminished ATP generation (Figure 2). The efficiency by which dietary calories are converted to ATP is determined by the coupling efficiency of OXPHOS. If the ETC is highly efficient at pumping protons out of the mitochondrial inner membrane and the ATP synthesis is highly efficient at converting the proton flow through its proton channel into ATP, then the mitochondria will generate the maximum ATP and the minimum heat per calorie consumed. These mitochondria are said to be tightly coupled. By contrast, if the efficiency of proton pumping is reduced and/or more protons are required to make each ATP by the ATP synthase, then each calorie burned will yield less ATP but more heat. Such mitochondria are said to be loosely coupled. Therefore, in an endothermic animal, the coupling efficiency determines the proportion of calories utilized by the mitochondrion to perform work versus those to maintain body temperature. As a toxic by-product of OXPHOS, the mitochondria generate most of the endogenous ROS. ROS production is increased when the electron carriers in the initial steps of the ETC harbor excess electrons, i.e., remain reduced, which can result from either inhibition of OXPHOS or from excessive calorie consumption. Electrons residing in the electron carriers for example, the unpaired electron of ubisemiquinone bound to the CoQ binding sites of complexes I, II, and III can be donated directly to O2 to generate superoxide anion (O2������). Superoxide O2������ is converted to H2O2 by mitochondrial matrix enzyme Mn superoxide dismutase (MnSOD, Sod2) or by the Cu/ZnSOD (Sod1), which is located in both the mitochondrial inner membrane space and the cytosol. Import of Cu/ZnSOD into the mitochondrial intermembrane space occurs as the apoprotein, which is metallated upon entrance into the intermembrane space by the CCS metallochaperone (166,207). H2O2 is more stable than O2��� ��� and can diffuse out of the mitochondrion and into the cytosol and the nucleus. H2O2 can be converted to water by mitochondrial and cytosolic glutathione peroxidase (GPx1) or by peroxisomal catalase. However, H2O2, in the presence of reduced transition metals, can be converted to the highly reactive hydroxyl radical (���OH) (Figure 2). Iron-sulfur centers in mitochondrial enzymes are particularly sensitive to ROS inactivation. Hence, the mitochondria are the prime target for cellular oxidative damage (241,242). For tightly coupled mitochondria, in the presence of excess calories and in the absence of exercise, the electrochemical gradient (��P) becomes hyperpolarization and the ETC chain stalls. This is because without ADP the ATP synthase stops turning over, thus blocking the flow of protons back across the mitochondrial inner membrane through the proton channel of the ATP synthase. However, the ETC will continue to draw on the excess calories for electrons to pump protons out of the mitochondrial inner membrane until the electrostatic potential of ��P inhibits further proton pumping. At this point, the ETC stalls and the electron carriers become maximally occupied with electrons (maximally reduced). These electrons (reducing equivalents) can then be transferred to O2 generating increased ROS and oxidative stress. By contrast, in individuals who actively exercise and generate ADP, the ATP synthase keeps ��P hypopolarized, electrons flowing through the ETC to sustain ��P, and the electron carriers retain few electrons (remain oxidized). The low electron density of the ETC electron carriers limits ROS production and reduces oxidative stress. This same effect can be achieved if the mitochondria become partially uncoupled, either by decreasing the number of protons pumped per electron pair oxidize of by permitting protons to flow back through the inner membrane without making ATP. Uncoupling can be achieved by disconnecting electron transport from proton pumping by alterations of complexes I, III, or Wallace Page 5 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

IV by increasing the number of protons required by the ATP synthase to make an ATP or by expression of an alternative proton channel such as an uncoupling protein (UCP). The mitochondria are also the major regulators of apoptosis, accomplished via the mitochondrial permeability transition pore (mtPTP). The mtPTP is thought to be composed of the inner membrane ANT, the outer membrane voltage-dependent anion channel (VDAC) or porin, Bax, Bcl2, and cyclophilin D. The outer membrane channel is thought to be VDAC, but the identity of the inner membrane channel is unclear since elimination of the ANTs does not block the channel (98). The ANT performs a key regulatory role for the mtPTP (98). When the mtPTP opens, ��P collapses and ions equilibrate between the matrix and cytosol, causing the mitochondria to swell. Ultimately, this results in the release of the contents of the mitochondrial intermembrane space into the cytosol. The released proteins include a number of cell death- promoting factors including cytochrome c, AIF, latent forms of caspases (possibly procaspases-2, 3, and 9), SMAD/Diablo, endonuclease G, and the Omi/HtrA2 serine protease 24. On release, cytochrome c activates the cytosolic Apaf-1, which activates the procaspase-9. Caspase 9 then initiates a proteolytic cascade that destroys the proteins of the cytoplasm. Endonuclease G and AIF are transported to the nucleus, where they degrade the chromatin. The mtPTP can be stimulated to open by the mitochondrial uptake of excessive Ca2+, by increased oxidative stress, or by deceased mitochondrial ��P, ADP, and ATP. Thus, disease states that inhibit OXPHOS and increase ROS production increase the propensity for mtPTP activation and cell death by apoptosis (Figure 2) (241,242). The mtDNA is maternally inherited (63) and semi-autonomous. The genetic independence of the mtDNA has been demonstrated by enucleating cells harboring a putative mtDNA marker, such as resistance to the mitochondrial ribosome inhibitor chloramphenicol (CAPR), and fusing the cytoplasmic fragments containing the donor mitochondria and mtDNA to recipient cells that lack the genetic marker, e.g., CAPS recipient cells (26,240). The resulting transmitochondrial cybrids have the mtDNA of the donor and thus inherit the donor���s mitochondrial phenotype (CAPR), yet they have the nDNA of the recipient, demonstrating that mtDNA inheritance can be independent of nDNA inheritance. The mtDNAs of the donor cell can more completely repopulate the recipient cells if the recipient cells are depleted, or cured, of their resident mtDNAs. The recipient cells can be depleted of their mtDNA by treatment with the mitochondrial poison rhodamine-6-G (220) or they can be cured of their mtDNAs by previous selection in ethidium bromide plus glucose, pyruvate and uridine, resulting in cells that are permanently mtDNA-deficient (��0 cells) (94). The mtDNA has a very high mutation rate, presumably due to its chronic exposure to mitochondrial ROS. When a new mtDNA mutation arises in a cell, a mixed intracellular population of mtDNAs is generated, a state known as heteroplasmy. As the heteroplasmic cell undergoes cytokinesis, the mutant and normal molecules are randomly distributed into the daughter cells. As a consequence of this replicative segregation, the proportion of mutant and normal mtDNAs can drift toward homoplasmic mutant or wild type. Furthermore, in post- mitotic tissues, mtDNAs harboring deleterious mutations have been found to be preferentially, clonally, amplified within cells. This may be a consequence of the nucleus attempting to compensate for an energy deficiency by making more mitochondria. Therefore, cells with defective mitochondria are preferentially stimulated to replicate their mitochondria and mtDNAs. As the proliferating mitochondria are turned over, presumably by autophagy (2, 116,136), the mutant mtDNA can become enriched in some cells by genetic drift. As the percentage of mutant mtDNAs increases, the mitochondrial energetic output declines, ROS production increases, and the propensity for apoptosis increases. As cells are progressively lost through apoptosis, tissue function declines, ultimately leading to symptoms. Thus the accumulation of mutant mtDNA creates the aging clock (241,242) (Figure 3). Wallace Page 6 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Clinical symptoms appear when the number of cells in a tissue declines below the minimum necessary to maintain function. The time when this clinical threshold is reached is related to the rate at which mitochondrial and mtDNA damage accumulates within the cells, leading to activation of the mtPTP and cell death, and to the number of cells present in the tissue at birth in excess of the minimum required for normal tissue function. Given that the primary factor determining cell metabolism and tissue structure is reproductive success, it follows that each tissue must have sufficient extra cells at birth to make it likely that that tissue will remain functional until the end of the human reproductive period, or about 50 years. If the mitochondrial ROS production rate increases, the rate of cell loss will also increase, resulting in early tissue failure and age-related disease. However, if mitochondrial ROS production is reduced, then the tissue cells will last longer and age-related symptoms will be deferred (236,238,241) (Figure 2). PATHOGENIC mtDNA MUTATIONS The first evidence that mtDNA mutations might be a key factor in aging and age-related degenerative diseases came with the identification of systemic diseases caused by mtDNA mutations. Pathogenic mtDNA mutations fall into three categories: rearrangement mutations (74), polypeptide gene missense mutations (243), and protein synthesis (rRNA and tRNA) gene mutations (200,244). As the number of mtDNA-associated diseases identified increased, it became clear that mitochondrial diseases commonly have a delayed onset and progressive course and that they result in the same clinical problems as observed in age-related diseases and in the elderly. Clinical manifestations that have been linked to mtDNA mutations affect the brain, heart, skeletal muscle, kidney, and endocrine system, the same tissues affected in aging. Specific symptoms include forms of blindness, deafness, movement disorders, dementias, cardiovascular disease, muscle weakness, renal dysfunction, and endocrine disorders including diabetes (51,149,237,241,242). Pathogenic mtDNA Rearrangement Mutations Systemically distributed mtDNA rearrangement mutations can be either inherited or spontaneous. Inherited mtDNA rearrangement mutations are primarily insertions. The first inherited insertion mutation to be identified caused maternally inherited diabetes and deafness (13,14). Spontaneous rearrangement mutations, primarily deletions, generally result in a related spectrum of symptoms, irrespective of the position of the deletion end points. This is because virtually all deletions remove at least one tRNA and thus inhibit protein synthesis (154). Thus the nature and severity of the symptoms from mtDNA deletion rearrangements is not a consequence of the nature of the rearrangement, but rather of the tissue distribution of the rearranged mtDNAs. Rearrangements that are widely disseminated prevent bone marrow stem cells from proliferating and lead to the frequently lethal childhood pancytopenia, known as Pearson marrow pancreas syndrome. Widely distributed mtDNA rearrangements that spare the bone marrow result in mitochondrial myopathy with ragged red fibers (RFF). Mitochon-drial myopathy is frequently associated with ophthalmoplegia and ptosis, which is referred to as chronic progressive external ophthalmoplegia (CPEO). More severe, earlier-onset cases with multisystem involvement are known as the Kearns-Sayre syndrome (KSS). Pathogenic mtDNA missense mutations���Missense mutations in mtDNA polypeptide genes can also result in an array of clinical manifestations. A mutation in the mtDNA ATP6 gene at np 8993 (T to G) is associated with neurogenic muscle weakness, ataxia, and retinitus pigmentosum (NARP) when present at lower percentages of mutant (75) and lethal childhood Leigh syndrome when present at higher percentages of mutant (213). This mutation causes a Wallace Page 7 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

marked inability of the ATP synthase to utilize the electrochemical gradient to make ATP (219) and an associated increase in mitochondrial ROS production (128). Missense and nonsense mutations in the cytb gene have been increasingly linked to progressive muscle weakness (8,52). Rare nonsense or frameshift mutants in COI have been associated with encephalomyopathies (25,37). Missense mutations in complex I genes have been linked to Leigh syndrome (205), generalized dystonia and deafness (88), and to Leber hereditary optic neuropathy (LHON), a form of midlife, sudden-onset blindness (22,23,243). Surprisingly, the phenotype of LHON can be caused by mutations in a number of ND genes that change amino acids, with very different amino acid conservation resulting in varying severities of complex I defects (22). This anomaly has been explained by the discovery that the mildest LHON mtDNA mutations, particularly those in ND6 at np 14,484 and ND4L at np 10,663, are usually found on a particular mtDNA background, the European mtDNA lineage J (21,23). Lineage J mtDNAs have been found to harbor mtDNA missense mutations that partially uncouple OXPHOS, thus exacerbating the ATP defect of the pathogenic mutations (194). Pathogenic mtDNA proteins synthesis mutations���Base substitutions in mtDNA protein synthesis genes can also result in multisystem disorders with a wide range of symptoms, including mitochondrial myopathy, cardiomyopathy, deafness, mood disorders, movement disorders, dementia, diabetes, intestinal dysmotility, etc. As with the ATP6 np 8993 (T G) mutation, studies on the tRNALys np 8344 A to G mutation associated with myoclonic epilepsy and ragged red fiber (MERRF) disease have revealed that the level of mutant heteroplasmy plus the age of the patient influence the severity of the clinical symptoms (200,244). Perhaps the most common mtDNA protein synthesis mutation is the np 3243 (G) in the tRNALeu(UUR) gene (64). This mutation is remarkable in the variability of its clinical manifestations. When present at relatively low levels (10%���30%) in the patient���s blood, the patient may manifest only type II diabetes with or without deafness (223). This is the most common known molecular cause of type II diabetes, purportedly accounting for between 0.5% and 1% of all type II diabetes worldwide (241). By contrast, when the A3243G mutation is present in 70% of the mtDNAs, it does not cause diabetes, but instead causes more severe symptoms including short stature, cardiomyopathy, CPEO, and mitochondrial encephalomyopathy, lactic acid and stroke-like episodes, the MELAS syndrome (64). Mitochondrial Defects in Diabetes The genetic linkage of mtDNA rearrangement (13,14) and tRNA mutations (223) to type II diabetes directly implicated mitochondrial defects in the etiology of diabetes and the metabolic syndrome. Evidence that mtDNA defects are a common factor in the etiology of diabetes comes from the observation that as the age-of-onset of the proband increases, the probability that the mother was the affected parent also increases, reaching a ratio of 3:1 for patients with a mean age-of-onset of 46 years (241). These observations have been linked to the larger human metabolic syndrome through the identification of a mtDNA tRNAIle mutation at np 4291 (T C) that causes hypertension, hyperc-holesterolemia, and hypomagnesemia (renal ductal convoluted tubule defect). This mutation was found in a maternal pedigree in association with reduced mitochondrial ATP production and the secondary clinical findings of migraine, hearing loss, hypertrophic cardiomyopathy, and mitochondrial myopathy (248). Consistent with these genetic results, studies of patients with type II diabetes have revealed that mitochondrial function and gene expression are generally down-regulated. Insulin- resistant offspring of type II diabetic patients have been found to have impaired mitochondrial energetics, as assessed by 31P-MR spectroscopy (176). Furthermore, type II diabetes patients Wallace Page 8 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

consistently show a down-regulation in the expression of nDNA-encoded mitochondrial genes, in association with alterations in the levels of the peroxisome-proliferation-activated receptor �� (PPAR�� )-coactivator 1 (PGC-1) (153,175), a major regulator of mitochondrial biogenesis and fat oxidation (105,251). Diabetes mellitus is also seen in Friedreich ataxia. Patients with Friedreich ataxia manifest cerebellar ataxia, peripheral neuropathy, hypertrophic cardiomyopathy, and diabetes as the result of inactivation of the frataxin gene on chromosome 9q13. Frataxin binds iron in the mitochondrial matrix, thus minimizing mitochondrial ���OH production. The loss of the frataxin protein results in excessive ROS generation which inactivates all mitochondrial iron-sulfur- containing enzymes including complex I, II, III, and aconitase. Thus increased mitochondrial ROS production and decreased mitochondrial OXPHOS must be the cause of diabetes in Friedreich ataxia (103,192,249). Type II diabetes has also been associated with a Pro121A polymorphism in the PPAR�� gene (5) and with a Gly482Ser polymorphism in the PGC-1 gene in Danish (53) and Pima Indian (156) populations. Maturity onset of diabetes in the young (MODY) is an early onset autosomal dominant form of type II diabetes that can also be linked to mitochondrial dysfunction. The molecular defects of four forms of MODY have been identified. MODY 2 is the result of mutations in glucokinase, MODY 1 is due to mutations in the hepatocyte nuclear factor (HNF)-1��, MODY 3 to mutations in HNF-4 ��, and MODY 4 to mutations in insulin promoter factor (IPF)-1. Glucokinase has a very high Km for glucose and is thought to be the glucose sensor. HNF-1�� is a transcription factor and mutations in its gene are associated with post-pubertal diabetes, obesity, dyslipidemia, and arterial hypertension, all features of mitochondrial diseases. HNF-1�� is also important in regulating nDNA-encoded mitochondrial gene expression and the expression of the GLUT 2 glucose transporter (250). HNF-4��, a member of the steroid/thyroid hormone receptor superfamily, acts as an upstream regulator of HNF-1�� (231) (Figure 4). The importance of mitochondrial defects in �� cell insulin secretion deficiency has been confirmed in two mouse models. In the first, the mitochondrial transcription factor Tfam was inactivated in the pancreatic �� cells. This resulted in increased blood glucose in both fasting and nonfasting states, and the progressive decline in �� cell mass by apoptosis (201). In the second, the ATP-dependent K+-channel (KATP) affinity for ATP was reduced, resulting in a severe reduction in serum insulin, severe hyperglycemic with hypoinsulinemia, and elevated D-3-hydroxybutyrate levels (102). These models demonstrate that mitochondrial ATP production is critical in the signaling system of the �� cell to permit insulin release (238). Thus pancreatic �� cell mitochondrial defects are important in both glucose sensing through glucokinase and insulin release through the KATP channel. Type II diabetes thus involves mutations in energy metabolism genes including the mtDNA and glucokinase mutations in the transcriptional control elements PPAR��, PGC-1, HNF-1��, HNF-4��, and IPF-1 and alterations in insulin signaling. These seemingly disparate observations can be unified through the energetic interplay between the various organs of the body. The human and mammalian organs can be divided into four energetic categories: the energy-utilization tissues including skeletal muscle, heart, kidney and brain the energy-storage tissues including brown adipose tissue (BAT) and white adipose tissue (WAT) the energy- homeostasis tissue, liver and the energy-sensing tissues including the �� and �� cells of the pancreatic Islets of Langerhans. All of these tissues interact to coordinate the utilization and storage of energy, based on the availability of calories in the environment. For our hunter- gather ancestors, the primary variation in available dietary calories was due to the cyclic growing seasons of edible plants caused by either warm versus cold or wet versus dry seasons. Wallace Page 9 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

During the growing season, plants convert the Sun���s energy into glucose, which the plants store as starch. When humans and animals ingest these plant tissues the concentration of glucose in their blood rises. Hence, serum glucose is the metabolic surrogate for monitoring plant calorie abundance. When plant calories are abundant and consumed, the elevated serum glucose is detected by the energy-sensing pancreatic �� cells, which respond by secreting insulin into the blood stream. The insulin signal then informs the energy-utilizing heart and muscle tissues to down-regulate mitochondrial energy utilization, since food-seeking behavior is less pressing. It also informs the energy-storing WAT and BAT tissues to store the excess calories as fat for when the season changes and plant calories again become limiting. When plant calories do become limited, insulin secretion declines and the pancreatic �� cells secrete glucagon. These low blood sugar hormonal signals inform the energy-utilization tissues to up-regulate the mitochondrial OXPHOS system, thus enhancing food-seeking capacity. They also mobilize the energy- storage tissues to transfer the stored triglycerides into the blood to fuel the increased mitochondria OXPHOS. Furthermore, low blood glucose stimulates the energy-homeostasis tissue, liver, to synthesis glucose to maintain the basal level of blood sugar, which is particularly critical for brain metabolism. The molecular basis of this primeval system for adapting to energy fluctuation can now be partially understood through recent discoveries pertaining to the transcriptional regulation of mitochondrial gene expression. The energy-sensing tissue, pancreatic �� cells, detects abundant plant calories when the blood glucose level exceeds the high Km of the �� cell glucokinase. Since glucokinase is bound to the mitochondrial VDAC and ANT through the mitochondrial inner and outer membrane contact points, the ATP binding site of glucokinase is continuously occupied (3,60,125). The binding of glucose by glucok-inase immediately generates G-6-P and then pyruvate. Pyruvate is oxidized by the mitochondria to generate ATP. The elevated ATP production increases the ATP/ADP ratio, which results in the closure of the KATP channel, depolarizing the �� cell plasma membrane. The depolarized plasma membrane opens the voltage-sensitive Ca2+ channel. The influx of Ca2+ leads to the formation of the exocytosis core complex and activation of protein kinases (255). The insulin-containing vesicles then fuse to the plasma membrane and release their stored insulin into the blood (Figure 4). Thus, when blood glucose is high, the insulin concentration in serum increases. The insulin binds to the insulin receptor of target cells throughout the body, activating the insulin receptor tyrosine protein kinase to phosphorylate insulin receptor substrates. These activate the phosphotidylinositol 3-kinase (PI3K), which phosphorylates phosphotidylinositol 2-phosphate (PI2) to phophatidylinositol 3-phosphate (PI3). PI3 then activates the AKT 1/2 kinases, which phosphorylate the FOXO forkhead transcription factors FOXO1, FOXO3A, and FOXO4 in the target tissues. Phosphorylation of the FOXOs results in their transport out of the nucleus and the transcriptional inactivation of genes whose promoters contain insulin response elements (IRE) (1,59). The FOXOs can also be removed from the nucleus by acetylation via Cbp or p300 (1) and reactivated by deacetylation via the NAD+-dependent SIRT1 (Sirtuin 1) (24). When blood sugar is low, the pancreatic �� cells secrete glucagons, which binds to the glucagon receptor in target cells. This stimulates the production of cAMP, which activates protein kinase A (PKA). PKA phosphorylates and activates the cAMP response element binding protein (CREB), which activates transcription of genes harboring a cAMP response element (CRE) (48). The FOXOs and CREB regulate numerous genes through the IRE and CRE cis elements. The PGC-1�� promoter contains three IREs that bind unphosphorylated and deacetylated FOXO1, Wallace Page 10 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

plus one CRE that binds phosphorylated CREB (48). PGC-1�� together with its companion genes PGC-1�� and PRC (PERC) are transcriptional co-activators that regulate genes for mitochondrial biogenesis, thermogenesis, and fatty acid oxidation (92,196). PGC-1�� was cloned through its interaction with PPAR��. It is strongly induced in BAT in response to cold acting through the cAMP-mediated ��-adrenergic and thyroid hormone (TR) systems. The up-regulation of PGC-1�� in turn induces UCP-1, which then creates a proton channel through the mitochondrial inner membrane, uncoupling OX-PHOS. This causes the rapid burning of the stored fats in BAT, generating heat to maintain body temperature (184). PGC-1�� also induces mitochondrial biogenesis and UCP-2 in muscle cells (C2C12), acting through the nuclear respiratory factors 1 and 2 (NRF1 and NRF2) (251). NRF1 and NRF2, in turn, activate the transcription of a wide range of nDNA-encoded mitochondrial genes, including components of OXPHOS such as the ATP synthase �� subunit (ATPsyn��) and cytochrome c (cytc) and the mtTFA, which regulates mtDNA transcription (92,196). The PGC-1 family of transcriptional coactivators interacts with a broad spectrum of tissue- specific transcription factors. This provides the tissue-specific response to the generalized insulin and glucagon hormonal signals. In the energy utilization tissue, muscle, PGC-1 induces mitochondrial biogenesis through the interaction with NRF1, PPAR��, PPAR��, MEF2, ERR2 (estrogen-related receptor 2), HDAC5, and Gabpa/b (GA-repeat binding protein) (47,78,117, 152). In heart, PGC-1�� interacts with PPAR�� and NRF1 (78,111). In the brain, the transcriptional partners of PGC-1�� have not yet been identified (118). In the energy-storage tissues, PGC-1�� interacts with PPAR��, PPAR��, and the thyroid hormone receptor (TR) in BAT to induce mitochondrial biogenesis and UCP-1 during cold stress (78,251), while in 3T3-Li preadipocytes, PGC-1�� interacts with PPAR�� to stimulate fatty acid oxidation, including induction of the medium-chain acyl CoA dehydrogenase (MCAD) (230). Finally, in the energy- homeostasis tissue, liver, PGC- 1�� interacts with FOXO1, HNF-4��, and the glucocorticoid receptor (GR) to induce gluconeogenesis enzymes, including phosphoenolpyruvate carboxylase (PEPCK) and glucose-6-phosphatase (G6Pase) to maintain basal blood glucose levels (78,121,187,254). PGC-1�� is abundantly expressed in BAT, heart, kidney, and skeletal muscle but also in stomach and in white adipose tissue. In 3T3-L1 preadipocytes, PGC-1�� is inducible, though PGC-1�� is not, and PGC-1�� then interacts with the ERRs to induce MCAD and fatty acid oxidation (89). Thus both insulin through the FOXOs and glucagon through cAMP modulate mitochondrial energy metabolism in response to the availability of carbohydrates via the regulation of members of the PGC-1 co-activator family. Mitochondrial energy metabolism is further modulated by PGC-1�� through deacety-lation by SIRT1. PC12 cells that overexpress SIRT1 experience a 25% reduction in oxygen utilization (161). Hepatocytes that over-express SIRT1 induce the gluconeogenesis genes PEPCK and G6Pase via interaction between PGC-1�� and HNF-4�� (187). Furthermore, upon caloric restriction, the SIRT1 in fat cells binds to the nuclear receptor corepressor (NCoR), blocking its interaction with PPAR��. This inhibits fat storage and enhances fat mobilization (178,179). These observations link all of the known genetic mutations associated with diabetes to defects in mitochondria bioenergetics and energy metabolism. Mutations in the mtDNA and in glucokinase would result in energetic dysfunction. Mutations in PPAR�� and PGC-1�� would reduce mitochondrial gene expression. Mutations in HNF-4�� and its target HNF-1�� would result in loss of glucose homeostasis. Thus, diabetes is an energy deficiency disease centered upon mitochondrial bioenergetics. Given a mitochondrial etiology for type II diabetes, the various stages in the progression of type II diabetes can be understood. In individuals harboring a partial defect in OX-PHOS, the Wallace Page 11 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

capacity of the energy-utilization cells to oxidize carbohydrates and fats to make ATP is reduced. Given a high caloric diet, individuals with partial OXPHOS defects overload their mitochondrial ETC with excessive calories (reducing equivalents), hyperpolarizing ��P, stalling the ETC, and blocking the tissue utilization of glucose. As a result, the nonmetabolized glucose remains in the blood. The chronically high serum glucose signals the �� cells to secrete insulin, creating concurrently elevated glucose and insulin: the hallmark of insulin resistance. The excessive reduction of the mitochondrial ETC electron carriers in the energy-utilization tissues maximizes mitochondrial ROS production. The high serum insulin activates their Akt pathway, which phosphorylates the FOXOs. The departure of the FOXOs from the nucleus stops transcription of the stress response genes, including the mitochondrial antioxidant enzymes. It also suppresses PGC-1�� transcription, which down-regulates mitochondrial OXPHOS, further exacerbating the mitochondrial energy deficiency. The resulting chronic mitochondrial oxidative stress erodes mitochondrial function and increases insulin resistance. The excess of reducing equivalents also increases the NADH/NAD+ ratio. The conversion of NAD+ to NADH inhibits SIRT1, resulting in increased acetylation of the FOXOs and their removal from the nucleus. This further down-regulates PGC-1�� and OXPHOS. The sustained high serum glucose and insulin also activates the PGC-1�� and PGC-1�� in the energy-storage tissues BAT and WAT to store the excess calories as fats. Also, the liver accumulates lipid, and this can lead to nonalcoholic steatohepatitis (NASH), which is observed in over 30% of individuals with indications of insulin-resistance, hypertriglyc-eridemia, and hypercholesterolemia (174). In the �� cells, the excessive mitochondria ROS inhibits mitochondrial ATP production, eventually leading to a decline in insulin secretion due to inadequate ATP for glucokinase and a low ATP/ADP ratio that cannot activate the KATP channel. The resulting high glucose but reduced serum insulin is termed insulin-independent diabetes. Continued calorie overload in the pancreatic �� cells and associated mitochondrial ROS production ultimately activates the �� cell mtPTP, resulting in �� cell death by apoptosis, producing insulin-dependent diabetes. Chronic mitochondrial oxidative stress on the peripheral tissues subsequently damages the retina, vascular endothelial cells, peripheral neurons, and the nephrons, leading to the clinical sequelae of end-stage diabetes. Thus chronic mitochondrial dysfunction can explain all of the features of type II diabetes, and is thus the likely cause of the disease. ANCIENT MIGRATIONS AND CLIMATIC ADAPTATION In addition to having to adapt to changing caloric availability due to seasonal changes, ancient human hunter-gathers had to adapt to the rigors of different climatic zones. While induction of UCP-1 in BAT and UCP-2 in muscle permits acute adaptation to thermal stress in rodents, this is not an adequate mechanism for long-term cold adaptation by humans. UCP-1 induction in BAT cannot generate sufficient heat to effectively regulate the temperature of the much larger human body. Therefore, as humans migrated into the more northern latitudes they had to adapt to the chronic cold using another mitochondrial strategy. It now appears that this global climatic adaptive strategy was achieved by the acquisition of mtDNA mutations that partially uncoupled OXPHOS and thus resulted in perpetually increased mitochondrial heat production. The first clear evidence that ancient mtDNA polymorphisms influence human physiology came from the observation that LHON patients with mild complex I mtDNA missense mutations generally encompassed the same European mtDNA lineage, J. It is now appears that lineage J mtDNAs harbor specific ���uncoupling mtDNA variants��� that reduce mitochondrial ATP output Wallace Page 12 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

in favor of heat production. Thus this lineage exacerbates the partial ATP defect generated by the pathogenic mtDNA mutations. Phylogeographic studies of the human mtDNAs have revealed a remarkable correlation between mtDNA lineages and the geographic origins of indigenous populations. These regional mtDNA lineages are groups of related individual mtDNA sequences (haplotypes) known as haplogroups. The various regional haplogroups form the branches on a single human dichotomous mtDNA phylogenetic tree, generated by the accumulation of sequential mtDNA mutations on radiating maternal lineages. The human mtDNA tree is rooted in Africa, and it has specific branches radiating into different geographic regions that we now believe to have been constrained by the climatic zones (Figure 5) (28,86,144,239). African mtDNAs are the most diverse and thus most ancient, with an overall age of about 150,000 to 200,000 YBP. African mtDNAs fall into four major haplogroups: L0 (oldest), L1, L2, and L3 (youngest). L0, L1, and L2 represent about 76% of all sub-Saharan African mtDNAs and are defined by a HpaI restriction site at np 3592 [macrohaplogroup L (L0, L1, and L2)] Figure 1. In northeastern Africa, two mtDNA lineages, M and N, arose from L3 about 65,000 YBP. These were the only mtDNA lineages that succeeded in leaving sub-Saharan Africa and radiating into Eurasia to give all of the Eurasian mtDNAs. In Europe, haplogroups L3 and N gave rise to haplogroups H (about 45% of European mtDNAs), H in Figure 1, T, U, V, W, and X (about 2%) as well as I, J (about 9%), and K(Uk). Europeans separated from Africans about 40,000���50,000 YBP. In Asia, lineages M and N radiated to give rise to a plethora of mtDNA lineages. These include from N, haplogroups A, B, F, and others and from M, haplogroups C, D, G, and others (239) (Figure 5). As Asians migrated northeast into Siberia, haplogroups A, C, D (A, C, D in Figure 1) became progressively enriched, such that they became the predominant mtDNA lineages in the indigenous peoples of extreme northeastern Siberia, Chukotka. When the Bering land bridge appeared about 20,000 to 30,000 YBP, people harboring these mtDNA haplogroups were in a position to migrate into the New World, where they founded the Paleo-Indians. After the land bridge submerged, haplogroup G arose in Central Asia and moved into northeastern Siberia to populate the area around the Sea of Okhotsk. Later, a migration carrying haplogroup B (B in Figure 1) started from eastern Central Asia and moved along the coast to the New World, by- passing Siberia. Haplogroup B then mixed with A, C, and D in southern North America, Central America, and northern South America to generate the Paleo-Indians. In addition, haplogroup X was brought to the New World in a migration that took place about 15,000 YBP. These immigrants settled in the Great Lakes region, and haplogroup X mtD-NAs are found in 25% of the Ojibwa mtD-NAs today. Since haplogroup X is found primarily in northeastern Europe, it has been speculated that an ancient European migration carrying this haplogroup might also have contributed to the Paleo-Indian populations, perhaps bringing the progenitors of the Clovis lithic culture to the Americas (239). Later migrations from northeastern Siberia, carrying a modified lineage of haplogroup A, founded the Na-D��n�� populations about 9500 YBP. More recently, immigrants from Siberia bearing derived lineages of haplogroups A and D moved along the Arctic Circle to found the Eskimos and Aleuts (239). This phylogeographic distribution of mtDNA haplogroups reveals two striking discontinuities in human mtDNA diversity. The first occurs between sub-Saharan Africa and Eurasia, in which all of the sub-Saharan African mtDNA diversity remained in Africa and only derivatives of lineages M and N colonized temperate Eurasia. The second occurs between temperate Central Asia and arctic Siberia, where the plethora of Asian mtDNA types is markedly reduced to only three ancient mtDNA lineages (A, C, and D). The resulting fivefold enrichment of haplogroups Wallace Page 13 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

A, C, D, and G in the arctic over Central Asia is unlikely to be the result of genetic bottlenecks, since there are no obvious geographic barriers that separate Asia and Siberia. It is more plausible that many Asian mtDNA lineages entered the arctic, but only a few survived the intense cold to become permanent residents. A similar logic could also apply to the African- Eurasian discontinuity. Evidence that climatic adaptation has influenced the geographic distribution of mtDNA diversity was first obtained from analyzing the amino acid replacement (nonsynonymous, NS) (Ka) to silent (synonymous, S) (Ks) mutation ratios (Ka/Ks) from the 13 mtDNA open reading frames (83,239). This revealed that the amino acid sequence of the ATP6 gene was highly variable in the arctic, but was strongly conserved in the tropics and temperate zone cytb was hypervariable in temperate Europe, but conserved in the tropics and arctic and COI was variable in the tropical Africa, but invariant in the temperate and arctic regions. Regional variation was also observed in multiple ND subunits (148). Such regional gene-specific variation would not be expected if all mtDNA mutations were random and neutral. The geographic constraints on mtDNA protein variation were further validated by positioning all of the mtDNA variants from 1125 complete mtDNA coding sequences collected from around the world and organized into a sequentially mutational tree. This tree could be assembled because the maternally inherited mtDNA can only change by sequential mutations along radiating female lines. Therefore, mutations shared between different mtDNAs must be derived from a common female ancestor, and those that are confined to a single mtDNA haplotype must be new mutations. Analysis of the mtDNA nucleotide variants revealed three different categories of variants: (a) neutral, including synonymous and weakly conserved nonsynonymous amino acid substitutions (b) deleterious, altering highly conserved amino acids but located at the tips of the branches of the tree indicating they are recent and (c) adaptive, altering highly conserved amino acids but located within the internal branches of the tree indicating that they are ancient. The mutations in class (c) must be adaptive because they alter highly conserved amino acids, yet they have persisted in the face of intense purifying selection for tens of thousands of years. Hence, as these variants cannot be neutral, they must have been adaptive. To quantify the potential effect of selection on particular mtDNA variants, the inter-specific amino acid conservation [Conservation Index (CI)] was determined for 22 known pathogenic mtDNA replacement mutations (194). This gave an average CI for deleterious mutations of 93.3 �� 13.3%. Using two standard deviations around the mean CI of the pathogenic missense mutations to distinguish between functionally important versus neutral replacement mutations, 26% of the internal branch replacement mutations were found to be functionally significant, with a mean CI = 85.1 �� 9.2%. The remaining 74% were found to be essentially neutral, with a mean CI = 23.3 �� 14.9%. Since these conserved internal branch missense mutations frequently initiate new limbs of the mtDNA tree, they must have permitted individuals carrying these mutations to survive and multiply in new geographic regions, generating the region-specific mtDNA haplogroups. The most obvious environmental factor that differentiates tropical and subtropical sub-Saharan Africa from temperate Eurasia and temperate Central Asia from arctic Siberia is temperature. This implies that an important functional effect of many of the conserved internal branch missense mutations was to permit adaptation to cold. That these adaptive internal branch mutations were important in human radiation was confirmed by analyzing the ratio of NS to S missense mutations in the three major climatic zones. The internal branch NS/S ratio for the tropical and subtropical African L haplogroups was 0.31. The internal NS/S ratio for the temperate Eurasian haplogroups M and N(R) were Wallace Page 14 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

0.42 and 0.44, whereas the internal NS/S ratio for the primarily arctic macro-haplogroup N (nonR) was 0.62. Thus, the farther north that the mtDNAs were found, the more missense mutations they harbored. The excess of internal branch missense mutations in the colder latitudes is particularly striking for haplogroups that reside in the arctic and subarctic. The mean internal branch NS/S ratio for northeastern Siberian-North American haplogroups A, C, D, and X was 0.61, much higher than the mean for the non-arctic haplogroups of 0.39, the mean for the African L haplogroups of 0.31, or the mean for the Native American haplogroup B, which by-passed the cold selection of Siberia, of 0.38. The internal branch replacement mutations of the arctic and subarctic haplogroups also have a higher average CI. The internal branch CIs of haplogroups A, C, D, and X was 51% that for the remaining global haplogroups was 39% that of L was 36% and that of B was 31%. The European haplogroup internal branch replacement mutations were also higher than the African L value of 0.31. Haplogroup H was 0.48, J was 0.66, and IWX was 0.63. By contrast, haplogroup T was only 0.31. However, this is because T harbors a single highly conserved founding replacement mutation in the ND2 gene (np 4917), which contributed virtually all of the adaptive advantage to this lineage. Hence, the CI of haplogroup T was 20.3, the highest of any European haplogroup. Thus, adaptive changes fall into two categories, either the lineage accumulates multiple missense mutations, each changing a somewhat less conserved amino acid, or the lineage changes only a few amino acids, each change of which is highly evolutionarily conserved. Examples of the adaptive mtDNA mutations that occur in the arctic mtDNAs include two replacement mutations [ND2 np 4824G (T119A) and ATP6 np 8794T (H90Y)] for haplogroup A two replacement variants [ND4 np 11969A (A404T) and cytb np 15204C (I153T)] for haplogroup C a ND2 np 5178A (L237M) variant for haplogroup D and a ND5 np 13708A (A458T) variant for a sublineage of haplogroup X. This latter variant also appears in European haplogroup J. The European sister-haplogroups J and T provide the clearest example of the two classes of adaptive mutation strategies: several less conserved mutations versus a few highly conserved mutations. J and T share a common root involving two amino acid substitutions: ND1 np 4216C (Y304H) and cytb np 15452A (L236I). J and T then diverge. Haplogroup T is founded by the single nodal adaptive mutation: ND2 np 4917G (N150D), the most conserved ND2 polymorphism found (194). Haplogroup J has two replacement mutations at its root: ND3 np 10398G (T114A) and ND5 np 13708A (A458T), the second being the same variant found in haplogroup X. Haplogroup J then splits into to sub-haplogroups J1 and J2, each defined by a major cytb mutation. The J2 cytb variant is at np 15257A (D171N) and the J1 cytb mutation is at np 14798C (F18L). The np 14789C mutation is also found at the root of sub-haplogroup Uk. The np 15,257 and np 14,789 variants alter well-conserved amino acids with CIs of 95% and 79%, respectively. The 15,257 variant alters the outer coenzyme Q binding site (Qo) of complex III, which contacts the Rieske iron-sulfur protein, while the np 14,798 site alters the inner coenzyme Q binding site (Qi) of complex III (194). Since the Qo and Qi binding sites are essential for complex III proton pumping via the Q-cycle, the np 14,798 and 15,257 variants are both likely to have disconnected electron flow through complex III with proton pumping. This would reduce the coupling efficiency of mitochondrial OXPHOS by one third and proportionately increase heat generation. Wallace Page 15 Annu Rev Genet. Author manuscript available in PMC 2010 February 12. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript