The Greatest Show on Earth (23 page)

Cellular family tree of
Caenorhabditis elegans

Embryology seems complicated – is complicated – but it is easy to grasp the important point, which is that we are dealing with local self-assembly processes all the way. It’s a separate question, given that (almost) all the cells contain all the genes, how it is decided which genes are turned on in each different kind of cell. I must briefly deal with that now.

THEN WORMS SHALL TRY

Whether or not a given gene is turned on in a given cell at a given time is determined, often via a cascade of other genes called switch genes or controller genes, by the chemical environment of the cell. Thyroid cells are quite different from muscle cells, and so on, even though their genes are the same. That’s all very well, you may say, once the development of the embryo is under way, and the different kinds of tissues such as thyroid and muscle already exist. But every embryo starts out as a single cell. Thyroid cells and muscle cells, liver cells and bone cells, pancreas cells and skin cells, all are descended from a single fertilized egg cell, via a branching family tree. This is a cellular family tree going back no further than the moment of conception, nothing to do with the evolutionary tree going back millions of years, which keeps cropping up in other chapters. Let me show you, for example, the complete family tree of all 558 cells of a newly hatched larva of the nematode worm, Caenorhabditis elegans (below: please pay close attention to every detail of this diagram). By the way, I don’t know what this tiny worm did to earn its species name of elegans, but I can think of a good reason why it might have deserved it retrospectively. I know that not all my readers like my digressions, but the research that has been done on Caenorhabditis elegans is such a ringing triumph of science that you aren’t going to stop me.
Caenorhabditis elegans was chosen in the 1960s as an ideal experimental animal by the formidably brilliant South African biologist Sydney Brenner. He had recently completed his work, with Francis Crick and others at Cambridge, on cracking the genetic code, and was looking around for a new big problem to solve. His inspired choice, and his own pioneering research on its genetics and neuro-anatomy, has led to a worldwide community of Caenorhabditis researchers that has grown into the thousands. It is only a bit of an exaggeration to say that we now know everything about Caenorhabditis elegans! We know its entire genome. We know exactly where every one of its 558 cells (in the larva; 959 in the adult hermaphroditic form, not counting reproductive cells) is in the body, and we know the exact ‘family history’ of every one of those cells, through embryonic development. We know of a large number of mutant genes, which produce abnormal worms, and we know exactly where the mutation acts in the body and the exact cellular history of how the abnormality develops. This little animal is known from start to finish, known inside out, known from head to tail and all stations in between, known through and through (‘O frabjous day!’). Brenner was belatedly recognized with the Nobel Prize for Physiology in 2002, and a related species was named in his honour, Caenorhabditis brenneri. His regular column in the journal Current Biology, under the byline ‘Uncle Syd’, is a model of intelligent and irreverent scientific wit – as elegant as the worldwide research effort on C. elegans that he inspired. But I do wish molecular biologists would talk to some zoologists (like Brenner himself) and learn not to refer to Caenorhabditis as ‘the’ nematode, or even ‘the’ worm, as though there were no others.
Of course you can’t read the names of the cell types at the bottom of the diagram (it would take seven pages to print the whole thing out legibly), but they say things like ‘pharynx’, ‘intestinal muscle’, ‘body muscle’, ‘sphincter muscle’, ‘ring ganglion’, ‘lumbar ganglion’. The cells of all these types are literally cousins of one another: cousins by virtue of their ancestry within the lifetime of the individual worm. For example, I am looking at a particular body muscle cell called MSpappppa, which is a sibling of another body muscle cell, first cousin of two more body muscle cells, first cousin once removed of two more body muscle cells, second cousin of six pharynx cells, third cousin of seventeen pharynx cells . . . and so on. Isn’t it amazing that we can actually use words like ‘second cousin once removed’, with the utmost precision and certainty, to refer to named and repeatably identifiable cells in an animal’s body? The number of cell ‘generations’ that separates the tissues from the original egg is not that great. After all, there are only 558 cells in the body, and you can theoretically make 1,024 (2 to the power 10) in ten generations of cell splitting. The numbers of cell generations for human cells would be much larger. Nevertheless, you could in theory make a similar family tree for every one of your trillion-odd cells (as opposed to the 558 cells of a C. elegans female larva), tracing each one’s descent back to the one fertilized egg cell. In mammals, however, it is not possible to identify particular, repeatably named cells. In us, it is more a case of statistical populations of cells, whose details are different in different people.
I hope my euphoric digression on the elegance of Caenorhabditis research has not distracted us too far from the point I was making about how cell types change in their shape and character as they branch away from one another in the embryonic family tree. At the branching point between a clone that is destined to become pharynx cells, and a ‘cousin’ clone that is destined to become ring ganglion cells, there has to be something to distinguish them, otherwise how would they know to turn on different genes? The answer is that, when the most recent common ancestor of the two clones divided, the two halves of the cell before division were different. So, when the cell divided, the two daughter cells, though identical in their genes (every daughter cell receives a full complement of genes), were not identical in the surrounding chemicals. And this meant that the same genes were not turned on – which changed the fate of their descendants. The same principle applies right through embryology, including its very start. The key to differentiation, in all animals, is asymmetric cell division.*
Sir John Sulston and his colleagues traced each of the cells in the body of the worm back to one and only one of six founder cells – we might even call them ‘matriarch’ cells – called AB, MS, E, D, C and P4. †In naming the cells, they used a neat notation that summarized the history of each one. Every cell’s name begins with the name of one of those six founder cells, the one from which it is descended. Thereafter, its name is a string of letters, the initial letters of the direction of cell division that gave rise to it: anterior, posterior, dorsal, ventral, left, right. For example, Ca and Cp are the two daughters of matriarch C, the anterior and posterior daughter respectively. Notice that every cell has no more than two daughters (of which one may die). I am now looking at a particular body muscle cell, whose name, Cappppv, succinctly discloses its history: C had an anterior daughter, which had a posterior daughter, which had a posterior daughter, which had a posterior daughter, which had a posterior daughter, which had a ventral daughter, which is the body muscle cell in question. Every cell in the body is denoted by a comparable string of letters headed by one of the six founder cells. ABprpapppap, to take another example, is a nerve cell that sits in the ventral nerve cord running along the length of the worm. Needless to say, it is not necessary to take in the details. The beautiful point is that every single cell in the body has such a name, which totally describes its history during embryology. Every one of the ten cell divisions that gave rise to ABprpapppap, and every other cell, was an asymmetric division with the potential for different genes to be switched on in each of the two daughter cells. And in all animals that is the principle by which tissues differentiate, even though all their cells contain the same genes. Most animals, of course, have far more cells than Caenorhabditis’ 558, and their embryonic development is in most cases less rigidly determined. In particular, as Sir John Sulston kindly reminds me, and as I have already briefly mentioned, in a mammal the ‘family trees’ of our cells are different for every individual, whereas in Caenorhabditis they are almost identical (except in mutant individuals). Nevertheless, the principle remains the same. In any animal, cells differ from each other in different parts of the body, even though they are genetically identical, because of their history of asymmetric cell division during the short course of embryonic development.
Let us hear the conclusion of the whole matter. There is no overall plan of development, no blueprint, no architect’s plan, no architect. The development of the embryo, and ultimately of the adult, is achieved by local rules implemented by cells, interacting with other cells on a local basis. What goes on inside cells, similarly, is governed by local rules that apply to molecules, especially protein molecules, within the cells and in the cell membranes, interacting with other such molecules. Again, the rules are all local, local, local. Nobody, reading the sequence of letters in the DNA of a fertilized egg, could predict the shape of the animal it is going to grow into. The only way to discover that is to grow the egg, in the natural way, and see what it turns into. No electronic computer could work it out, unless it was programmed to simulate the natural biological process itself, in which case you might as well dispense with the electronic version and use the developing embryo as its own computer. This way of generating large and complex structures purely by the execution of local rules is deeply distinct from the blueprint way of doing things. If the DNA were some kind of linearized blueprint, it would be a relatively trivial exercise to program a computer to read the letters and draw the animal. But it would not be at all easy – indeed, it might be impossible – for the animal to have evolved in the first place.
And now, so that this chapter on embryos should not end up as a mere digression in a book on evolution, I must return to the sincere dilemma of Haldane’s questioner. Given that genes control processes of embryonic development rather than adult shape; given that natural selection – like God – doesn’t build tiny wings, but embryology does; how does natural selection go to work on animals to shape their bodies and their behaviour? How does natural selection go to work on embryos, in other words, to rejig them so they become ever more proficient at building successful bodies, with wings, or fins, leaves or armour plating, stings or tentacles or whatever it takes to survive?
Natural selection is the differential survival of successful genes rather than alternative, less successful genes in gene pools. Natural selection doesn’t choose genes directly. Instead it chooses their proxies, individual bodies; and those individuals are chosen – obviously and automatically and without deliberative intervention – by whether they survive to reproduce copies of the very same genes. A gene’s survival is intimately bound up with the survival of the bodies that it helps to build, because it rides inside those bodies, and dies with them. Any given gene can expect to find itself, in the form of copies of itself, riding inside a large number of bodies, both simultaneously in a population of contemporaries, and successively as generation gives way to generation. Statistically, therefore, a gene that tends, on average, to have a good effect on the survival prospects of the bodies in which it finds itself will tend to increase in frequency in the gene pool. So, on average, the genes that we encounter in a gene pool will tend to be those genes that are good at building bodies. This chapter has been about the procedures by which genes build bodies.
Haldane’s interlocutor found it implausible that natural selection could put together in, say, a billion years, a genetic recipe for building her. I find it plausible, although of course neither I nor anybody else can tell you the details of how it happened. The reason it is plausible is precisely that it is all done by local rules. In any one act of natural selection, the mutation that is selected has had – in lots of cells and in lots of individuals in parallel – a very simple effect on the shape into which a protein chain spontaneously coils up. This, in turn, through catalytic action, speeds up, say, a particular chemical reaction in all the cells in which the gene is turned on. This changes, perhaps, the rate of growth of the embryonic primordium of the jaw. And this has consequential effects on the shape of the whole face, perhaps shortening the muzzle and giving a more human and less ‘ape-like’ profile. Now, the natural selection pressures that favour or disfavour the gene can be as complicated as you like. They might involve sexual selection, perhaps aesthetic choice of a high order by would-be sexual partners. Or the change in jaw shape might have a subtle effect on the animal’s ability to crack nuts, or its ability to fight rivals. Some hugely elaborate combination of selection pressures, conflicting and compromising with one another in bewildering complexity, can bear upon the statistical success of this particular gene, as it propagates itself through the gene pool. But the gene knows nothing of this. All it is doing, within different bodies and in successive generations, is rejigging a carefully sculpted dent in a protein molecule. The rest of the story follows automatically, in branching cascades of local consequences, from which, eventually, a whole body emerges.

Even more complicated than the selection pressures in the ecological, sexual and social environments of the animals is the phantasmagoric network of influences that go on within and among the developing cells: influences of genes on proteins, genes on genes, proteins on the expression of genes, proteins on proteins; membranes, chemical gradients, physical and chemical guide rails in embryos, hormones and other mediators of action at a distance, labelled cells seeking others with identical or complementary labels. Nobody understands the whole picture, and nobody needs to understand it in order to accept the exquisite plausibility of natural selection. Natural selection favours the survival in the gene pool of the genetic mutations responsible for making crucial changes in embryos. The whole picture emerges as a consequence of hundreds of thousands of small, local interactions, each one comprehensible in principle (although it may be too hard or too time-consuming to unravel in practice) to anyone with sufficient patience to examine it. The whole may be baffling and mysterious in practice, but there is no mystery in principle, either in embryology itself, or in the evolutionary history by which the controlling genes came to prominence in the gene pool. The complications accumulated gradually over evolutionary time: each step was only a tiny bit different from the one before, and each step was accomplished by a small, subtle change in an existing local rule. When you have a sufficient number of small entities – cells, protein molecules, membranes – each at its own level obeying local rules and influencing others – then the eventual consequence is dramatic. If genes survive or fail to survive as a consequence of their influence on such local entities and their behaviour, natural selection of successful genes – and the emergence of their successful products – will inevitably follow. Haldane’s questioner was wrong. It is not in principle difficult to make something like her.
And, as Haldane said, it only takes nine months.