by Nick Lane
Against this background, it’s perhaps rash for me to attempt a categorical statement, but attempt I shall. The picture that is beginning to emerge does indeed suggest that both sides are right. It seems there is a difference in the fate of mutant mitochondria, depending on the location of the mutation—whether this lies in the control region of the mitochondrial genome, or in a coding region.
Mutations in the control region (which binds the factors responsible for copying mitochondrial DNA) can prosper, and even stage a clonal take-over of entire tissues. These mutations don’t necessarily cause much functional deficit. A ground-breaking study, published in Science in 1999 by Giuseppe Attardi (one of the pioneers of mitochondrial DNA sequencing) and his colleagues at Caltech, showed that individual mutations in the control region can accumulate at over 50 per cent of the total mitochondrial DNA in the tissues of older people, but are largely absent in young people. We can take it that certain types of mutation do build up with age to high levels, but we can’t say whether these mutations are harmful, as they don’t affect the protein-coding genes. Certainly not all of them are harmful. Another important study published by Attardi’s group in 2003 showed that one mutation in the control region is actually associated with greater longevity in an Italian population. In this case the mutation, a single letter change, cropped up five times more often in centenarians than in the general population, implying that it might have offered some kind of survival advantage.
In contrast, mutations in the functional protein- or RNA-coding regions very rarely accumulate at levels above about 1 per cent, which is far too low to cause a significant energy deficit. Interestingly, functional mitochondrial mutations, such as cytochrome oxidase deficiency, are amplified clonally within particular cells, so that the mutants come to predominate in those cells. This happens, for example, in some neurones, heart-muscle cells, and indeed the ragged red fibres of ageing muscles. However, the total proportion of such mutants in the tissue as a whole rarely rises above 1 per cent. There are two possible explanations for this. One is that different cells accumulate different mutations, so that any particular mutation is just the tip of an enormous iceberg of diverse mutations. The other explanation is that most mitochondrial mutations simply don’t accumulate at very high levels in ageing tissues. Perhaps surprisingly, it’s this latter explanation that seems closest to the truth. Several studies have shown that most mitochondria in ageing tissues have basically normal DNA, except perhaps in the control region, and moreover, are capable of virtually normal respiration. Given that the proportion of mutant mitochondria needed to undermine a cell’s performance in mitochondrial diseases is as high as 60 per cent, a total mutation load of just a few per cent seems insufficient to account for ageing, at least by the tenets of the original mitochondrial theory.
So what’s going on here? I find myself asking the obtuse question, ‘Are we really so different to yeast?’ I doubt that question will cost many readers much sleep. But it ought to! Yeasts accumulate mitochondrial mutations rapidly, yet for the most part, we don’t. From an energetic point of view, we function in a similar way to yeast; the only difference is that we depend on our mitochondria, whereas yeasts don’t. Perhaps this difference gives the game away—necessity. Let’s say we accumulate mutations in the control region simply because they don’t matter a great deal: they have little impact on function (as indeed is implicit in the studies of human inheritance discussed in Part 6), whereas most functional mutations do not accumulate because they do matter. That sounds fair enough, but it implies that selection for the best mitochondria is taking place in tissues (even in tissues composed of long-lived cells like heart and brain). So we are faced with two possibilities. Either the mitochondrial theory of ageing is completely wrong, or mitochondrial mutations do occur at a similar rate to yeasts, but the mutants are eliminated by selection within a tissue for the best mitochondria. If so, then mitochondrial function must be far more dynamic than had been envisaged in the original mitochondrial theory of ageing. Which is it?
17
Demise of the Self-Correcting Machine
After the previous chapter, you might be forgiven for supposing that the mitochondrial theory of ageing is claptrap. After all, most of its predictions seem utterly false. One prediction is that antioxidants should prolong maximal lifespan, and this does not seem to be true. Another is that mitochondrial DNA mutations should accumulate with ageing, but only the least important ones actually do. Another is that the proportion of free radicals escaping the respiratory chains is constant, so lifespan should vary with metabolic rate; but this is only true in general, and fails to explain exceptions like bats, birds, humans, and the exercise paradox (the fact that athletes, who consume more oxygen over a lifetime, don’t age faster than couch potatoes). In fact, the only prediction of the original theory that seems to be true is that mitochondria are the main source of free radicals in the cell. Hardly the lineaments of a vigorous, healthy theory.
It’s time to return to an idea that we parked in the previous chapter: the proportion of free radicals escaping from the respiratory chains is not constant and unavoidable, but is subject to natural selection. Over evolutionary time, the rate of free-radical leakage is set at the optimal level for each species. In this way, long-lived animals have a fast metabolic rate while leaking relatively few free radicals, whereas short-lived animals typically combine a fast metabolic rate with leaky mitochondria and plenty of antioxidants. We posed the question: What is the cost of well-sealed mitochondria? Why would a rat not benefit from cutting back its investment in antioxidants by sealing its mitochondria better? What does it have to lose?
Let’s think all the way back to Part 3, and in particular to John Allen’s explanation for the very existence of mitochondrial DNA (see pages 141–144). You might recall that he argued it was no fluke that a hard-core of genes survives in every species that relies on oxygen for respiration. The reason, he suggested, was to balance the requirements of respiration, as an imbalance in the components of the respiratory chain can lead to inefficient respiration and free-radical leakage. We saw that it is necessary to retain a local contingent of genes to pinpoint the need for reinforcements in particular mitochondria, rather than all mitochondria at once, regardless of need, which is what would happen if control were retained by the bureaucratic confederation of genes in the nucleus. Allen’s essential point is that the mitochondrial genes survive because the benefits of keeping them on site outweigh the disadvantages.
How might one particular mitochondrion signal its need to produce more respiratory chain components? We’re now entering into the realms of twenty-first-century science, and it’s best to admit that at present little is known. As we saw in Part 3, Allen suggests that they do so by modulating the rate of free-radical production from the respiratory chains: the free radicals themselves act as the signal to start building more respiratory complexes. This suggests an immediate reason why a rat might lose out by restricting free-radical leakage—it would muffle the strength of the signal, and so require a more refined detection system. We’ll see later how birds might have got around this problem, and why it wasn’t worth it for rats.
What might happen if there is not enough cytochrome oxidase in a particular mitochondrion? This is the scenario that we considered in Chapter 8. Respiration becomes partially blocked, and electrons back up in the respiratory chains, making them more reactive. Oxygen levels rise, as less is consumed by respiration. The combination of high oxygen with slow electron flow means that free-radical production increases. According to Allen, this is exactly the signal required to produce more complexes, to correct the deficit. How the mitochondria detect the rise in free-radical leakage is unknown, but various plausible possibilities exist. For example, mitochondrial transcription factors (which initiate protein synthesis) might be activated by free radicals; or the stability of the RNA might depend on free-radical attack. Examples of both are known, but neither has been proved to take place in the mitochondria
. Either way, the rise in free-radical leakage should lead to more core respiratory proteins being made from mitochondrial DNA. These insert themselves into the inner membrane, and once implanted they behave as beacons and assembly points for the additional proteins encoded by the nuclear genes. When the full complex is assembled, the blockage of respiration is corrected. Free-radical leakage falls again, and so the system is switched off. Overall, then, this system behaves like a thermostat, in which a fall in room temperature is itself the signal used to switch on the boiler. The rising temperature then switches off the boiler, so regulating the room temperature between two fixed limits. But of course, if the room temperature didn’t fluctuate up and down, the system couldn’t work at all. Similarly, if the rate of free-radical leakage from respiratory chains didn’t fluctuate, there could be no self-correction to an appropriate number of respiratory complexes.
What happens if the free-radical signalling fails? If the synthesis of new respiratory proteins from the mitochondrial genes fails to stem the leak of free radicals, then the lipids of the inner membrane, such as cardiolipin, are oxidized. In Part 5, we noted that cardiolipin binds cytochrome oxidase, so if cardiolipin is oxidized, cytochrome oxidase is released from its shackles. This in turn blocks the passage of electrons down the respiratory chain altogether, so respiration grinds to a halt. Without the constant flow of electrons needed to maintain it, the membrane potential collapses, and apoptotic proteins are spilled out into the cell. If this turn of events happens in a single mitochondrion, the cell does not necessarily commit apoptosis—there appears to be a threshold. When only a few mitochondria expire at one moment, then the signal for apoptosis is not strong enough to cause the cell to die, and instead the mitochondria themselves are broken down. In contrast, if a large number of mitochondria simultaneously spill out their contents, then the cell as a whole does get the point, and goes on to commit apoptosis.
This flexible signalling system is a far cry from the spirit of the original mitochondrial theory of ageing. The original theory supposed that free radicals are purely detrimental; that the continued existence of mitochondrial DNA was a diabolical fluke of evolution; and that free-radical damage spiralled out of control, leading to the degeneration and misery of ageing. We now appreciate that free radicals are not purely detrimental—they carry out an essential signalling role—and that the bizarre survival of mitochondrial DNA is not a fluke, but is actually necessary for cellular and bodily health. Furthermore, mitochondria are better protected against free-radical damage than had once been assumed. Not only is mitochondrial DNA present in multiple copies (usually five to ten copies in every mitochondrion), but recent work shows that mitochondria are reasonably efficient at repairing damage to their genes, and (as we saw in Part 6) are capable of recombination to fix genetic damage.
So where does all this leave the mitochondrial theory of ageing? Perhaps surprisingly, it’s not dead and buried, merely radically transformed. A new theory has emerged, phoenix-like, from the ashes, but it still places a premium on the free radicals generated by the mitochondria. This new theory is not attributable to any one mind in particular, but has gradually condensed out of the work of researchers in several related fields. Beyond being consistent with the data, the new theory has the immeasurable benefit that it gives a deep insight into the nature of the diseases of old age, and how modern medicine might set about curing them. Critically, the best way to tackle them is not to target each one individually, as medical research currently does, but to target all of them simultaneously.
The retrograde response
We have seen that the mitochondria operate a sensitive feedback system, in which the leaking free radicals themselves act as signals to calibrate and adjust performance. But the fact that free radicals play an integral part in mitochondrial function does not mean that they are not toxic too. Clearly they are, even if rather less so than is shouted about in health magazines. Lifespan does correlate with the rate of free-radical leakage from respiratory chains. While a good correlation doesn’t necessarily imply a causal link, it’s hard to claim causality in the absence of any correlation at all. If two factors are not linked in any way, then one can hardly be said to ‘cause’ the other; and there are remarkably few, if any, other factors which correlate with lifespan across radically different groups, such as yeast, nematodes, insects, reptiles, birds, and mammals. For the sake of argument, let’s assume that free radicals do cause ageing. How can we square their signalling role with a more conventional idea of their toxicity—and with the evidence to date?
Yeast accumulate mitochondrial mutations at least 100 000 times faster than nuclear mutations. People, too, accumulate particular types of mitochondrial mutation with age, notably those in the ‘control’ region. Importantly, the control region mutations can often stage a ‘take-over’ of whole tissues, so that the same mutation is found in practically all the cells. In contrast, mutations in the coding regions of the mitochondrial genome can be amplified within particular cells, but only very rarely attain levels of above 1 per cent in the tissue as a whole. I suggested that this smells suspiciously of selection acting in tissues. Can we perhaps link the signalling role of free radicals with a weeding-out of detrimental mitochondrial mutations? Indeed we can, and this is the crux of the ‘new’ mitochondrial theory of ageing.
What might happen to the calibration of mitochondrial function if there is a spontaneous mutation in mitochondrial DNA? Let’s think it out step by step. If the mutation is in the control region, it doesn’t affect gene sequence, but it might affect the binding of transcription or replication factors. If the effect is not entirely neutral, the mutant mitochondrion would tend to copy its genes either more often or less often in response to an equivalent stimulus. So what is the outcome? If the mutation makes a mitochondrion ‘fall asleep’ on duty, so it responds sluggishly to signals for replication, the likely outcome is that the mutant mitochondria would simply disappear from the population. In response to a signal to divide, ‘normal’ mitochondria would divide, but the mutant mitochondria would slumber on. Their population would fall relative to normal mitochondria, and they would eventually be displaced altogether in the normal turnover of cellular components.
In contrast, if the mutation made the mitochondrion more alacritous in its response to an equivalent signal, we would expect to see an expansion of its DNA. At every signal to divide, the mutant mitochondria would leap into action, and so would eventually displace the ‘normal’ mitochondria from the population. And if the mutation occurred in a stem cell (which gives rise to replacement cells in a tissue) the mutants would be more likely to be passed on every time the stem cell divided, and so would finally take over the entire tissue. It’s important to note that such mutations are most likely to stage a tissue takeover if they’re not particularly detrimental to mitochondrial function. This is likely to be true, as there is nothing the matter with the respiratory complexes themselves. Energy generation can continue normally, if a degree out of synch with requirements; and as we’ve seen (page 287), Giuseppe Attardi’s group has shown that one control-region mutation is actually beneficial.
So what happens if a mutation is in the coding region of a gene? Why do such mutations take over individual cells, but not the tissue as a whole? This time it’s more likely that mitochondrial function will be altered. Let’s imagine that the mutation affects cytochrome oxidase in some way. Given the need for nanoscopic precision in the interactions of different subunits, the likelihood is that respiration will be impaired, and electrons will back up in the respiratory chains. Free-radical leakage increases, and this signals the synthesis of new respiratory chain components. This time, however, building new complexes cannot correct the deficit, for these, too, would be dysfunctional (although if the deficit is modest, it may help a little). What happens next? The outcome is not the error catastrophe proposed in the original version of the mitochondrial theory, but more signalling. The defective mitochondrion signals its deficiency to t
he nucleus, by way of a feedback pathway known as the ‘retrograde response’, which enables the cell to compensate for its deficit.
The retrograde response was originally discovered in yeast, and was so named because it seems to reverse the normal chain of command from the nucleus to the rest of the cell. In the retrograde response, it is the mitochondria that signal to the nucleus to change its behaviour—the mitochondria, not the nucleus, set the agenda. Since its discovery in yeast, some of the same biochemical pathways have been found to operate in higher eukaryotes too, including humans. While the exact signals almost certainly differ in detail and meaning, the overall intention appears to be similar—to correct the metabolic deficiency. Retrograde signalling switches energy generation towards anaerobic respiration, such as fermentation, and in the longer term stimulates the genesis of more mitochondria. It also fortifies the cell against stress, aiding survival in the more trying times ahead. Yeast, which don’t depend on their mitochondria to survive, actually live longer when the retrograde response is active. With our dependence on mitochondria, it’s unlikely that similar benefits would apply to people; for us, the purpose of the retrograde response is to correct the mitochondrial deficiency. But I suppose we could be said to live longer in the sense that, without it, we would certainly live ‘shorter’.