Two radical technologies are being explored in the attempt to extend healthy lifespans. The first involves senolytics; the second, mutated mitochondrial DNA. Neither works. Mitochondrial control of cellular senescence offers a way forward.1 Mitochondrial control is exerted across the entire organism, and researchers have already devised methods for therapeutically transforming ailing tissue by introducing whole mitochondria. Many of these techniques have only been applied in vitro.2 But getting mitochondria into cells is not the problem. Cells scoop them up by using endocytotic mechanisms or by building elaborate membrane-tunneling nanotubes.3 Current clinical work lacks an overarching theory for how mitochondria are created, mined, selected, eliminated, and how they move throughout the body.4
Senolytics
Senescence occurs when cells no longer divide, but do not die. Neurons and heart cells are senescent by design; they remain permanently in the quiescent G0 growth phase. When cells reach the Hayflick limit, if they do not die, neither do they reproduce. When this phenomenon was discovered, it was thought that artificial activation of telomerase might be used to rebuild telomeres, extending the division cycles of cells. This is how most tumors evade senescence.
The indiscriminate reactivation of telomerase would be folly. Senolytics aims to eliminate senescent cells through the use of drugs. Senescent cells, and only senescent cells, must be targeted. But there is no marker unique to senescent cells. Senolytic drugs tend to be chemotherapeutic or antiviral, and their side effects have proven toxic.5
Allotopic Expression
Mitochondria are endosymbiont proteobacteria that, at the dawn of the eukaryotes, were engulfed by an archaebacterial host. Nearly all the genes of the original symbiont were copied into its nuclear DNA, and mitochondrial localization sequences (MLS) were used to move the proteins into the mitochondria. Once mitochondria had a steady source of nuclear mito proteins, those genes were dropped from their own DNA (mtDNA). In many mammals, for example, all that remains of the mtDNA are 13 protein subunits, and the minimal set of genes needed for their translation. Allotopic expression fixes mitochondria with damaged mtDNA by expressing copies of pristine mitochondrial genes in the nuclear DNA. Some molecular biologists see this as completing a natural evolutionary process. Once mitochondrial genes are encoded in the nuclear DNA, they are immune from the free radicals generated by respiration, uncoupling their evolution from the high substitution and deletion rates of mtDNA.
This process demands that proteins be imported into the mitochondria. They must first be recoded to use the nuclear codon system. The hydrophobicity of the proteins themselves must then be reduced so that they do not fold before they are imported. The mRNA code itself may need to be optimized so that it is translated by the cytosolic ribosomes at the mitochondrial surface.
Getting this process to work is a tall order.
Researchers have achieved successful allotopic expression for the ATP6 and ATP8 subunits of complex V.6 Another group has reported success for the ND4 subunit of complex I. It is possible that individuals with major mitochondrial dysfunction could benefit from saturating cells with nuclear copies of missing subunits. How to extrapolate these findings to healthy people? It is by no means clear. For one thing, current gene sequences have evolved for maximum efficiency; alterations to allow for nuclear expression would result in inferior gene products. Consider species like Drosophila and the many single-celled protists that have far fewer genes in their mitochondria than human beings. Trypanosomes, for example, get all of their tRNA from the nucleus and use the same translocators for both tRNA and proteins.7 Every product they leave in the nucleus is affected negatively; the proteins become less hydrophobic. Even more problematic is the question of regulating genes that are haphazardly integrated into the nuclear DNA. Over millions of years, flies and protists may have achieved a certain level of regulation for those genes expressed in the nucleus.
Giving up local regulation in the mitochondria comes at an evolutionary cost.8
Colocation for Redox Regulation
At the heart of the mitochondrion is the nucleoid that contains the mtDNA. It is there that mitochondrial ribosomes, or mitoribosomes, are assembled. Proteins made by these mitoribosomes are then cotranslationally inserted directly into the overlying inner membrane. By analogy with the nucleolus of the cell nucleus, this machine is called the mitochondriolus.9
Organelles with elaborate electron transport chains need to influence their expression to maintain redox balance in the bioenergetic membrane. Genes and gene products must be colocated. The initial evidence for the hypothesis that redox regulation depends on some form of colocation came from the observation that photosynthetic organelles tune the rate of their electron transport in response to changes in light quality.10 The redox state of electron carriers directly controls various protein modifications, which in turn ultimately determine how energy is distributed. Local redox regulation of transcription adjusts the stoichiometry of photosystem components within the chloroplasts. Similar redox regulation occurs in mitochondria, where the generation of radicals, the direction of electron transport, and the stoichiometry of individual respiratory complexes are all tightly controlled.11
There seems to be little doubt that the restriction or loss of organelle respiratory bioenergetics is accompanied by the reduction or loss of organelle DNA. The prediction of the colocation redox regulation hypothesis is borne out in mitosomes and hydrogenosomes that have no genomes left. Conversely, loss of the genome would lead to loss of functional oxidative respiration. The only question is how quickly.
A New Theory
Recent studies have explored the large deletions that result directly from transient ischemia. One common deletion takes out a 7.3kb chunk that includes eleven proteins. The breakpoints occur in genes near direct repeat sequences and other hotspots. While allotopic expression, or even direct germ-line editing, could theoretically be useful, a simple mitochondrial transplant could well fix this.12
Mitochondria carry nearly twice as many tRNAs as proteins, but this seemingly massive and disproportionate overhead is actually a benefit. As genetic hybrids, these organelles exploit two completely separate and uniquely optimized transcription and translation systems. This metagenetic cooperative works in tandem to construct the respiratory complexes, with mitoribosomes in the matrix and ribosomes at the outer membrane facing the cytosol. Eukaryotes use a bacterially derived glycolytic pathway, and their ribosomes ultimately stem from an archeal lineage. Mitoribosomes, on the other hand, stem from bacterial ribosomes. There is only one reported instance of mitoribosomes naturally translating nuclear RNAs, and that was for sperm mitochondria.13
Shlomi Reuveni, Måns Ehrenberg, and Johan Paulsson are the authors of a new theory that explains how different kinds of ribosomes are precisely tailored to what they need to accomplish.14 Cytosolic ribosomes are autocatalytic in the sense that they beget other ribosomes. Except for a couple of rRNAs, mitoribosome parts are made by cytosolic ribosomes. Ribosome doubling time imposes significant constraints on the cell doubling time. The smallest and fastest translating ribosomes are those of bacteria, which are under the most selective pressure for rapid biogenesis. Their ribosomes, which contain the shortest ribosomal proteins and the highest mass percentage of rRNA (70%), require only six minutes to make a new set of ribosomal proteins. While their rRNAs vary greatly in size, the short proteins all turn out to be roughly of the same length; rRNA length is less important than proper stoichiometry. Eukaryotic ribosomes are somewhere in the middle of the ribosome world: they have more proteins, each of longer length, and less need for rapidly synthesized rRNA. At the far end of the spectrum are the mitoribosomes, which have the largest protein mass (80%), and the highest number of ribosomal protein subunits (~80), each with longer than average length.
Reuveni, Ehrenberg, and Paulsson have devised a theory that neatly accounts for these distributions. It involves a formula for the minimum fraction of time ribosomes spend on their own generation. The fraction they represent as a function of cell doubling time, the number of ribosomal proteins, and the time involved in their construction.15 If their theory can accurately predict how mitochondria should outfit and apportion their mitoribosomes, perhaps we can generate theories for how mitochondria conduct the respiratory complexes. Respiratory and metabolic theories should eventually add up to a morphological theory of how mitochondria choose to fuse, fizz, die, and amble through different compartments in the cell. Beyond that, it should account for net creation and dissolution within the mitochondrial population as it migrates across cells and through the body at large.
Across the Kingdom
Mitochondria are pressed into service across the animal kingdom. Viper pit organs, eel shock boxes, and the paracrystalline lens of the planarian eye, are all packed with unusual forms of mitochondria. In human hearts, cell-wide networks of around 8,000 mitchondria fill 30% of the volume of the cell,16 controlling contractility by rapidly uncoupling intermitochondrial junctions when membranes are depolarized.17 When forming a network, individual cristae line up into a continuous reticulum.18 Consider the firefly light mantle.19 To make light, a firefly needs to send oxygen to peroxisomes, where the luciferin-based light reaction happens. For this, oxygen must be made available. Nerves first release the transmitter octopamine onto nearby tracheal cells. These form branching channels that penetrate the light mantle, causing a signal cascade that leads to the generation of nitric oxide. When the nitric acid reaches the mitochondria—some 20μm away, in the cytoplasm of photocytes—it binds to the hemoprotein of cytochrome c oxidase and inhibits the use of oxygen. This is a relatively simple transduction pathway; its basic behavior could be captured by static compartmental models involving O2, NO, and perhaps a few other molecules. If, however, we take into account autonomous mitochondria capable of moving toward or away from the peroxisomes, the situation becomes complicated.
There have been several attempts to capture global mitochondrial performance in terms of just a few parameters with respect to oxygen use and formation of radicals. One set of results points to the ratio of FADH2 to NADH electrons as the crucial determinant in radical formation. The F/N ratio would be low during glucose breakdown and high during fatty acid breakdown. The longer the fatty acid, the higher the F/N ratio. The longest fatty acids should be broken down where the extra FADH2 would not lead to excess mitochondrial radical formation.
If the power of a theory is measured by what it can explain, its usefulness lies in what it can predict. Here the theory explains both the evolution of peroxisomes and the absence of fatty acid oxidation in long-lived cells like neurons. It may also explain mitochondrial refinements like carnitine shuttles, uncoupling proteins, and multiple antioxidant mechanisms linked to fatty-acid oxidation. Depending on the species and tissue, peroxisomes perform many unique functions for the cell.20
Still another approach has involved splitting mitochondrial operation into three primary operating modes for the purpose of quantifying how radicals, particularly superoxide anions (), are produced in each. In the first, or normal mode, mitochondria make ATP (adenosine triphosphate). This results in a more oxidized NADH pool, reduced membrane potential (Δp), and negligible production. In the second mode, mitochondria do not make ATP. They have a high Δp, a reduced coenzyme Q (CoQ) pool and CoQH2/CoQ ratio, leading to reverse electron transport and significant generation of . In the third mode, is also high, but from a reduced NADH pool or high NADH/NAD+ ratio in the matrix.21 To test these kinds of theories, we would need to measure accurately in vivo, perhaps via electron paramagnetic resonance measurements.22
A Return to Reality
The genetic engineering of mitochondria is not as advanced as it is for the nucleus. There are fewer good restriction enzymes, and advanced methods like CRISPR (clustered regularly interspaced short palindromic repeats) do not work. But there are many new methods now in the pipeline to modify the mitochondrial genome.23 It is also possible to get new proteins into the mitochondria without their having full allotopic expression. One research group optically controlled mitochondrial membrane potential and ATP generation by transfecting cells with cDNAs for optically gated channels.24 With the right leader sequences, they could put proteins into the mitochondria instead of the plasma membrane. It should also be possible to transfect mitochondria with optical indicators.25
Similarly enticing would be the ability to swap out mitochondria to match a particular environment. There are specific mitochondrial haplotypes that are ideally suited for different temperatures and altitudes.26 Researchers and clinicians alike invariably speak of heteroplasmy as a negative. This does not have to be the case. Two heterogeneous populations of mitochondria can peacefully coexist. We can even watch endogenous mitochondria that have not been modified in any way. Two-photon-excited fluorescence of NADH in mitochondria can image mitochondrial reorganization in the skin, and distinguish basal cell carcinoma and melanoma.27
Conventional wisdom holds that injecting extra mitochondria into the bloodstream is nothing like a blood transfusion. Mitochondria are highly immunogenic by virtue of their formylated peptides and undermethylated CpG islands, something that our immune system exploits.28 A Chinese group recently found evidence that bloodstream injection of mitochondria in rodents can apparently fix certain kinds of brain damage.29 There was no immune activation in the acute phase for the first two hours after injection. Cells from many different organs had taken up functional mitochondria from the initial injection stock. A likely explanation is that blood or endothelial cells absorb the mitochondria right after injection and pass them on to other distribution channels.30
If the mitochondria can pass the blood brain barrier, or can be supplied more directly to key access points in the nervous system, then many new possibilities arise. The nervous system is immune privileged; the peripheral immune system rarely gains access to the brain. And nerves reach out and touch many locations that capillary beds only reach by diffusion, such as sensory corpuscles, hair follicles, bone marrow sites of hematopoiesis, and niche stem-cell populations in the epithelial sheets lining many organs in the body.
In stem cells, mitochondria directly control the fate and senescence of daughter cells, sorting themselves according to age, membrane potential, and other indicators. A daughter cell that differentiates into a postmitotic state receives more and younger mitochondria; the daughter that retains stem-cell characteristics retains fewer but older mitochondria.31 When cells or their progeny turn cancerous, it is frequently because their mitochondrial function has been compromised. What often transforms a quiescent tumor into an invasive cancer is the influx of fresh mitochondria from nearby cells.32
Nerves also control cellular fates, and particularly cancer. Nerve-controlled tumors appear in multiple cancers and organs. Cholinergic innervation of stem-cell crypts determines the timing of cell maturation, proliferation, and tumorigenicity in the gut, while nerves specializing in other factors, like GDNF (glial cell line-derived neurotrophic factor), control pancreatic cancers. Still other nerve channels feed basal-cell carcinomas or melanomas in the skin. The neurological control of cell senescence and cancer has several things in common with the neurological control of tissue regeneration. Salamander digits and tadpole limbs readily regrow when cut off, but all regenerating tissues require neurological control.
As I have written, nerves take the donation and absorption of mitochondria to an extreme. The nervous system controls the rate of aging throughout the body, and maintains its populations of cells in various states of senescence and proliferation by apportioning mitochondria. In this light, cell senescence is not quite so irreversible as one might assume.33
Brain Cancers
A unique feature of glioblastoma is its resistance to all forms of treatment. Knocked down, it reorganizes and returns with a vengeance. Another feature that defines all glioblastomas is that they form one continuous syncytium.34 By watching these networks evolve, it was discovered that nuclei and mitochondria constantly scan the entire glioma through the membranous tubes that interconnect it. When cells are irradiated, mitochondria bear the brunt of the damage.35 One source of the remarkable resistance of gliomas to radiation seems to be their ability to rapidly repopulate and energize the network with fresh mitochondria.
Many of the hidden links between nuclear alterations and mitochondrial dysfunction are now coming to light. To sense mitochondrial status, the nucleus attaches mitochondrial localization sequences to the front end of a transcription factor, and nuclear localization sequences to its back end. If the nucleus makes copies of the transcription factor ATFS1, but the mitochondria cannot import it because the membrane potential is too weak for uptake, then the factor will eventually find its way back to the nucleus.36
While we lack a complete understanding of mito-nuclear regulatory crosstalk in cancers like glioblastoma, some inferences are now possible. The regular migration of nuclei and mitochondria within the glioma syncytium might be an evolutionary throwback. Sriram Garg and William Martin have suggested that eukaryotic chromosome division arose in a filamentous, syncytial ancestor.37 Individual nuclei inside the protosyncytium with insufficient chromosome numbers could complement each other through mRNA in the cytosol, and generate new chromosome combinations through nuclear fusion. This theory explains why the mechanisms for eukaryotic chromosome separation are more conserved than those for cell division. Garg and Martin showed that the energy provided by mitochondria relieved a major constraint on the ability to produce tubulin for chromosome separation. The methods used to transmit mitochondria look like those features that the brain employs in its daily operation.38 Much of the basic allometry of nervous systems, like relative lengths and branching patterns of dendrites and axons, cannot be adequately explained by the computational theory to which traditional neurosciences continue to appeal. These features may have been shaped by what the mitochondria needed to accomplish. Rapid signal transduction and computation would have arisen as a side effect.
Under what circumstances, then, do neurons create and transmit mitochondria? When do they remove them? And when is this done with whole mitochondria and when just their nucleoids? Consider cells that use dopamine. These cells are uniquely susceptible to mitochondrial loss, particularly in Parkinson’s disease or in the presence of certain drugs. Different forms of Parkinson’s can be distinguished by analyzing mitochondria vented into the cerebrospinal fluid (CSF). During a stroke, mitochondria are also found in the CSF.39 By donating mitochondria, astrocytes can rescue ailing neurons after their blood supply is interrupted.40 In idiopathic Parkinson’s disease, a high concentration of mtDNA in the CSF has been found in patients who carry the LRRK2G2019S mutation.41 To replenish mitochondria in a disease that targets transmitter systems like dopamine, we would need a diagram of the mitochondrial flow through the striatum.
Mitochondrial Maps
Many common chemotherapies damage the hematopoietic system that generates new blood cells by destroying nerve endings that contact stem-cell niches in the bone marrow. Without these sympathetic nerves, the growth of hematopoietic cells grinds to a halt.42 In acute myeloid leukemia (AML), the bone marrow is infiltrated by poorly differentiated blast cells. Compared to normal CD34+ hematopoietic stem-cell progenitors, these cells are overloaded with mitochondria and rely on their oxidative phosphorylation to generate ATP. This contrasts with most common tumors, which generate their ATP by aerobic glycolysis.
The mitochondria responsible for transforming malignant AML blasts are supplied by bone marrow stromal cells (BMSCs). This heterogeneous population of local stromal cells includes precursors of endothelial cells, osteoclasts, osteoblasts, adipocytes, and fibroblasts. AML blasts get mitochondria by extending tunneling nanotubes to the BMSCs and by generating superoxide radicals with NADPH oxidase.43
For people with serious mitochondrial dysfunction or depletion, the status of their mitochondriome is more important than the number of mitochondria in their blood cells. Heteroplasmy is a woefully neglected issue in medicine. To make pig organs compatible with humans, researchers delete sequences known as porcine endogenous retroviruses from the pig.44 Few are concerned about introducing pig mitochondria. Human cells may in fact tolerate heteroplasmic mitochondria from other species; hybrid cow-human embryos have been created, and the cells, which retained cow mitochondria along with the human nucleus, were viable.45
Friedreich’s ataxia is a neurodegenerative disease caused by inherited deficiency of the mitochondrial protein frataxin, which makes iron-sulfur clusters needed for the respiratory complexes I, II, and III. These clusters are also needed for DNA replication and repair. Frataxin mutations cause deficiencies in mitochondrial biogenesis.46 A drug now used to treat multiple sclerosis called dimethyl fumarate boosts mitochondrial numbers. To test it, one might simply count mitochondria in bone marrow–derived platelets in the blood.47
Any drug given to the body as a whole is going to affect metabolisms throughout the system. For aging, no drug has been unambiguously proven to be worth the trouble—there are always significant side effects. Controlling the life cycle of the body will require controlling the nervous system and mitochondria distributed by it.