by Nick Lane
So we have a gap. It seems very likely that the mitochondria lost their autonomy well before the evolution of true multicellular organisms. During this period, the mitochondria could gain nothing from killing their hosts, for they could not survive independently. Nor could their hosts gain anything from being killed, for they were not yet part of a multicellular organism. Thus the current advantages of apoptosis, the ruthless maintenance of a police state in multicellular organisms, could not apply.
This is a paradox. The persistence of a dedicated machinery of death must have been actively detrimental to both host and mitochondria. We might expect it to have been jettisoned by natural selection, yet we know that it was maintained. We also know that much of the death apparatus was inherited from the mitochondria, rather than from the host (or evolving more recently). And to cap it all, I have argued in favour of the hydrogen hypothesis, which contends that the eukaryotic cell originated in a metabolic union between two peacefully cohabiting cells, neither of which could gain from killing the other. I seem to have argued us into a blind alley, to wit: a collaborative cell brought with it to a peaceful union a fully developed death apparatus, detrimental to both parties, which persisted against all the odds for a few hundred million years before it happened to find a use. Can this crazy scenario be rationalized? Yes, but only if we are prepared to make a concession—the death apparatus did not always cause death. Once upon a time, it caused sex.
Sex and the origin of death
Let’s consider the first eukaryotes from the point of view of the peaceful cohabitation proposed by the hydrogen hypothesis. In the introduction to Part 5, we discussed the different levels at which natural selection operates—the level of the individual as a whole, or its constituent cells, or the mitochondria within the cells, or of course the genes themselves. We saw that it is not necessarily helpful, when considering cells that replicate asexually like bacteria, to think about natural selection operating at the level of the genes. Instead, selection works mostly at the level of the individual cells, which in this case are the true replicating units. This background will now prove invaluable to us, for we must consider the interests of the mitochondria and their host cells separately, in the early days of the eukaryotic merger. In those days, both the mitochondria and the host cells could be thought of as separate cells (and we shall see in the next few chapters that in many ways it still helps to consider them in this way).
So what were the private interests of the proto-mitochondria and their host cells? Given their combination of autonomy and uneasy mutual dependency, how could they have acted out their own interests? A compelling answer was put forward in 1999, by one of the most fertile thinkers in evolutionary biochemistry, Neil Blackstone, at Northern Illinois University, along with Douglas Green, one of the pioneers of cytochrome c release in apoptosis, at UCSD, La Jolla.
Like all cells, it is in the interest of mitochondria to proliferate. As soon as their own future has been tied to that of their host, they can gain nothing from killing this host and moving on to another—they could not survive the interim in the wild. There’s also a limit to how far the mitochondria can proliferate within a single host cell: a mitochondrial ‘cancer’ within the host would be detrimental to the cell as a whole, which would perish, along with all of its mitochondria. So the only way that the mitochondria can successfully proliferate is in line with the host cell. Each time the host cell divides, the mitochondrial population must double, to provide a contingent for each daughter cell. Of course, there’s nothing the host cell likes better than dividing either, so the interests of host and mitochondria are in common. If they were not, it is quite doubtful that the arrangement could have persisted as a stable relationship for two billion years. It would surely have torn itself asunder early on, and we would not have been here to be any the wiser.
But the interests of the mitochondria and the host cell are not always in common. What might happen if, for some reason, the host cell refused to divide? Clearly neither the host cell nor its mitochondria could then proliferate (well, the mitochondria could proliferate, but only to a certain point: it would be detrimental to the host, and so to the mitochondria themselves, if they continued proliferating until they produced a mitochondrial ‘cancer’ inside the cell). The consequences might differ depending on the reason the host cell refused to divide. The most likely reason is lack of food. In Part 3 we noted that most bacteria spend most of their lives in stasis, despite their enormous capacity to replicate. The same must have applied to the early eukaryotes. If so, there was nothing to do but wait out the lean times, and resume proliferating again as soon as food became available. In this case, the interests of the mitochondria and the host cell are again in common: if the mitochondria pressed the host to divide without sufficient resources, both would perish. Better to devote the remaining resources to bolstering resistance to any physical stress likely to be encountered during the period of deprivation, such as heat, cold, and ultraviolet radiation. Under these conditions, many cells form a resistant spore, which survives the wait in a dormant state before springing back to life in times of plenty.
Another reason that might prevent the host cell from dividing is damage, in particular to the DNA of the cell nucleus. Now the interests of the host and the mitochondria begin to diverge. Let’s assume that food is plentiful, but the host cell is nonetheless unable to divide. You can almost picture the trapped mitochondria, faces pressed against the bars, yelling ‘Let me out! Unfairly imprisoned!’ In the meantime, their neighbouring cells grin and divide away, their mitochondria proliferating happily. What are the trapped mitochondria to do? They don’t gain anything from killing their host, as they’d soon be dead themselves. But they would gain if the host cell fused with another, and recombined its DNA with that of the partner. Recombination of DNA is common in bacteria, and is the very basis of sex in eukaryotes. The fused cell gains a new lease of life—and the mitochondria a new playground.
Why sex evolved in eukaryotes is still fiercely contested, given its twofold cost (see page 191). It seems likely that several different factors contribute. Sex tends to mask damaged DNA, as the damaged gene is likely to be paired with an undamaged copy of the same gene; and the variety generated by recombination probably gives cells a competitive edge over parasites—a theory championed by Bill Hamilton. Recent data imply that neither reason alone is sufficiently strong in all circumstances to account for the evolution of sex; but they don’t conflict with each other, and it seems likely that the benefits of sex are many pronged. On the other hand, its origin is a mystery. Bacteria recombine genes, but never fuse cells. In contrast, sexual reproduction in most eukaryotes involves the fusion of two cells, then the fusion of their nuclei, and finally the recombination of their genes, an altogether more committing act. What made eukaryotic cells fuse in the first place? Losing the unwieldy cell wall of bacteria no doubt made the physical act of fusion far more practicable, but this still does not account for the actual urge to fuse. Cells don’t fuse all the time, so there is nothing about the wall-less state in itself that promotes fusion. Might it be that early eukaryotic cells were manipulated by their mitochondria to fuse together? If so, could mitochondrial sabotage explain the origin of sexual fusion? Tom Cavalier-Smith, whom we met in Part 1, has reasoned that cell fusion would have been common in the early eukaryotes: he argues that the form of cell division in sex (meiosis), in which the chromosomal number is first doubled, and then halved, evolved via a few simple steps, as a means of restoring the original number of genes and nuclei after cellular fusions. In this case, mitochondria might have agitated for a fusion that was likely to happen in due course anyway.
The question of whether the mitochondria can manipulate the host cell is a serious one. We know they do today: they cause apoptosis. But might they have done so in the early days of the eukaryotic cell too? Neil Blackstone has suggested an ingenious way in which they could have done, and it explains both the urge to fuse and ultimately the evolution of apop
tosis.
Free-radical signal
Think about the respiratory chain. We discussed the leakage of free radicals from the chain in Part 3. Paradoxically, the rate of free-radical leakage does not correspond to the rate of respiration, as one might think intuitively, but rather depends on the availability of electrons (ultimately derived from food) and oxygen. Because these factors vary continuously, free-radical production shifts according to circumstances. Sudden bursts of free-radical production can affect the behaviour of the cell.
If a cell is growing and dividing quickly, and so has a high demand for fuel (and plenty to meet this demand), there is a fast flux of electrons down the respiratory chain to oxygen. In these circumstances, relatively few free radicals leak from the chain. This is because they are more likely to pass down the line of least resistance, from one electron acceptor to the next in the chain, and finally to oxygen. Blackstone describes the chain in these circumstances as a well-insulated wire, through which electricity flows as a current of electrons. So, fast growth with plentiful fuel equates to a low leakage of free radicals.
What about times of starvation? Now there is little fuel, and practically no electrons passing down the respiratory chain. There may be plenty of oxygen around, but no spare electrons to stray off and form free radicals. If we think of the respiratory chains as little electrical wires, then starvation equates to a grid power failure: it’s impossible to suffer an electric shock if the mains supply is dead. Free-radical leakage is low because there is no electron flow at all.
But now think about what happens if a cell is damaged, leaving it with plenty of fuel, but no longer able to divide. The mitochondria are trapped in their prison. Because there is no cell division, there is a very low demand for ATP, and the cellular stocks remain high. The rate of electron flow down the chain depends on the rate of consumption of ATP. If ATP consumption is fast, then the electrons flow swiftly to keep up, as if sucked on by a vacuum; but if there is no demand, then the respiratory chain becomes choked up with spare electrons, which have nowhere to go. Now there is plenty of oxygen as well as spare electrons. The rate of free-radical leakage is far higher. The respiratory chain behaves like a badly insulated wire, easily giving an electric shock. So if the host cells are damaged, and don’t grow or divide despite plentiful fuel, their mitochondria give them an electric shock from within: a sudden burst of free radicals.1
Any burst of free radicals tends to oxidize the lipids in the mitochondrial membranes, and release cytochrome c from its shackles into the inter-membrane space; this in turn completely blocks electron flow down the chain, as cytochrome c is an integral part of the respiratory chain. Losing cytochrome c from the chain is like clipping a live wire. The earlier part of the chain chokes up with electrons, and continues to leak free radicals, just as the live part of a clipped live wire still gives a shock. But cessation of electron flow eventually dissipates the membrane potential (for proton leak is no longer balanced by proton pumping), and as stress mounts the pores open in the outer mitochondrial membrane, spewing apoptotic proteins, including cytochrome c, into the rest of the cell. In other words, these circumstances simulate the first steps of apoptosis.
Where does all this leave us? It means the interests of the mitochondria and the host cell are aligned at most times. If both proliferate, all is well and good. The cell is in a reduced (as opposed to oxidized) state, but free-radical leakage is minimal. Conversely, if resources are scarce, then neither can proliferate, and the cell will do best to bolster its resistance, to wait out the lean times ahead. The cell is now in an oxidized state, and again free-radical leakage is minimal. When the host cell is damaged, however, and can’t divide despite having plenty of fuel, then the mitochondria signal their displeasure by producing an angry burst of free radicals. This is significant, says Blackstone, for the free-radicals attack the DNA in the cell nucleus (and the presence of cytochrome c in the cytosol would actually promote free-radical formation there). In yeasts and other simple eukaryotes, DNA damage constitutes a signal for sexual recombination. Even more strikingly, in the primitive multicellular alga Volvox carteri, a luminously beautiful hollow green ball, a twofold rise in free-radical production activates the sex genes, leading to the formation of new sex cells (gametes). Importantly, this effect can be induced by a blockage of the respiratory chain. So Blackstone’s theory can be furnished with some concrete examples. The long and short of it is this. The first few steps of apoptosis in single cells might once have stimulated sex, not death.
First steps to the individual
This view is entirely compatible with the hydrogen hypothesis, for it implies that the cells involved in the initial eukaryotic merger lived peacefully together, but nonetheless retained their own interests. These interests stretched to manipulating the host for sex, in the case of mitochondria, but not to murder, from which neither side could gain. Moreover, such a gently manipulative relationship, in which most interests are aligned at most times, explains why the machinery of apoptosis might have survived in single cells for possibly hundreds of millions of years—sex benefits both the damaged host and the mitochondria, and so would not be penalized by natural selection.
But the question remains: how did sex turn into death? We know that the mitochondria brought most of the death machinery with them, and they certainly use it to kill their hosts by apoptosis today. If we accept that the original purpose of the death machinery was sex, not death, what led to such a portentous change in purpose? When did the drive for sex become punishable by death, and why?
Sex and death are entwined. To an extent, both serve the same purpose. Consider why yeasts and Volvox are driven to recombine their genes when their DNA is damaged: recombination of genes probably enables the damaged copy to be replaced, or masked, by an undamaged copy of the same gene. Similarly, free radicals promote lateral gene transfer in bacteria (the uptake of genes from other cells or the environment). Again, the damaged genes are replaced or masked. What of programmed cell death, then? In multicellular organisms, apoptosis is also a means of repairing damage. Rather than the costly option of fixing a broken cell, apoptosis takes the cost-effective approach of eliminating it from the body, making space for an undamaged replacement—the first steps towards our modern ‘throwaway’ culture. So sex helps to eliminate damaged genes, while apoptosis eliminates damaged cells. Seen from the point of view of the ‘higher’ organism, sex repairs damaged cells and apoptosis repairs damaged bodies.
In Blackstone’s view, the machinery of apoptosis originally signalled cells to fuse, instigating recombination and repair of damage. At a later stage, in multicellular organisms, the machinery was rededicated to death. In principle, all that was required was the insertion of a new step—the caspase cascade. We noted earlier that the caspase enzymes were inherited from α-proteobacteria (probably by way of the mitochondrial merger), but that they serve a different purpose in bacteria—they slice up some proteins, but do not bring about the death of the cell. In this respect, it’s interesting that different groups of eukaryotes appear to have integrated the caspase enzymes into programmed cell death quite independently. Plants, for example, bring about cell death using a group of related proteins known as meta-caspases, whereas mammals use the familiar caspase cascade. Both, however, trigger cell death through the release of cytochrome c and other proteins from the mitochondria. This implies that the death machinery of apoptosis arose independently more than once in the eukaryotes, in response to a common signal (free radicals, and the release of proteins from stressed mitochondria) and a common selection pressure—the need to eliminate damaged cells from a multicellular organism.
If apoptosis is linked with the need to police the multicellular state, rather than a parasite war, and multicellular organisms evolved independently more than once, which they certainly did, then it is not surprising that the detailed execution of apoptosis differs in different groups. On the face of it, it is more surprising that there is so much in common—that somewha
t similar machinery was pressed into service more than once. Why was this?
Again Blackstone suggests an answer. He has spent many years studying some of the most primitive animals, such as marine colonial hydroids (colonies of cells that are capable of reproducing sexually or asexually, by fragmentation). He argues that a multicellular colony offers various advantages over individual cells, but as soon as the cells within a colony begin to differentiate—so that some are obliged to fulfil menial tasks, like paddling (moving the colony around), while others form fruiting bodies that pass on their genes—then a tension must develop. What stops the menial ‘slave’ cells from revolting?
Although they are all genetically identical (for a period at least), the cells in a colony don’t have equal opportunities—a ‘caste’ system develops in which some cells reap privilege at the expense of others. Blackstone argues that redox gradients are set up by the food and oxygen supply, which varies with currents, other local fluctuations, and the position of cells within the colony (at the surface or buried under other cells). Some cells have plenty of oxygen and food while others are deprived of one or the other, and so find themselves in a different redox state. The differentiation of cells is controlled by their redox state, by way of signals from the mitochondria. We have already noted, for example, that a lack of respiratory electrons, due to starvation, generates a signal for stress resistance.