The Science of Discworld III - Darwin's Watch tsod-3

by Terry Pratchett

  They have also taught us that an organism's development is not rigidly determined by the DNA of its cells. Environmental insults can push the course of development along pathological paths. In addition, the genetics of organisms, particularly wild organisms, are usually organised so that `normal' development happens despite a variety of environmental insults, and even despite changes in some of the genes. This so-called `canalised' development is very important for evolutionary processes, because there are always temperature variations, chemical imbalances and assaults, parasitic bacteria and viruses; the growing organism must be `buffered' against these variations. It must have versatile developmental paths to ensure that the `same' well-adapted creature is produced, whatever the environment is doing. Within reasonable limits, at any rate.

  There are many developmental tactics and strategies that help to accomplish this. They range from simple tricks like the HSP90 protein to the very clever mammalian trade-off.

  HSP stands for `heat shock protein'. There are about 30 of these proteins, and they are produced in most cells in response to a sudden, not very severe, change of temperature. A different array of proteins is produced in response to other shocks; this one is called HSP90 because of where it sits in a much longer list of cell proteins. HSP90, like most HSPs, is a chaperonin: its job is to hug other proteins during their construction, so that when the long line of amino acids folds up it achieves the `right' shape. HSP90 is very good at making the `right' shape - even if the gene that specifies the chaperoned protein has accumulated a lot of mutations. So the resulting organism doesn't `notice' the mutations; the protein is `normal' and the organism looks and behaves just like its ancestral form.

  However, if there's a heat shock or other emergency during development, HSP90 is diverted from its role as chaperonin, and other less powerful chaperonins permit the mutational differences to be expressed in most of the progeny. The effect this has on evolution is to keep the organisms much the same until there's an environmental stress, when suddenly, in one generation, lots of previously hidden, but hereditable, variation appears.

  Most books that describe evolution seem to assume that every time there's a mutation, the environment promptly gets to judge it good or bad ... but one little trick, HSP90, which is present in most animals and many bacteria, makes nonsense of that assertion. And from Lewontin's discovery that a third of genes have common variants in wild populations, and that all organisms carry lots of them, it is clear that ancient mutations are continually being tested in different modern combinations, while the potential effects of more recent mutations are being cloaked by HSP90 and its ilk.

  The trick employed by mammals is much more complex and farreaching. They reorganised their genes, and got rid of a lot of genetic complication that their amphibian ancestors relied on, by adopting a new and more controlled developmental strategy. Most frogs and fishes, whose eggs usually encounter great differences and changes of temperature during each embryology, ensure that the `same' larva, and then adult, results. Think of frog spawn in a frozen English pond, warming up to 35°C during the day while the delicate early development proceeds; then the little hatchling tadpoles have to endure these temperature changes. Now think of the frogs that so few of the tadpoles become.

  Most chemical reactions, including many biochemical ones, happen at different rates if the temperature is different. You only get a frog if all the different developmental processes fit together effectively, and timing is crucial. So how does frog development work at all, given that the environment is changing so quickly and repeatedly?

  The answer is that the frog genome `contains' many different contingency plans, for many different environmental scenarios. There are many different versions of each of the enzymes and other proteins that frog development requires. All of them are put into the egg while it is in mother frog's ovary. There are perhaps as many as ten versions of each, appropriate to different temperatures (fast enzymes for low temperatures, sluggish ones for higher temperatures, to keep the duration of development much the same)[49], and they have `labels' on the packages that make them, so the embryo can choose which one to use according to its temperature. Animals whose development must be buffered in this way use a lot of their genetic programme to set up contingency plans for many other variables, in addition to temperature.

  The mammals cleverly avoided all of this faffing around, by making their females thermostatically controlled -'warm-blooded'. What counts is not the warmth of the blood, but the system that maintains it at a constant temperature. The beautifully controlled uterus keeps all kinds of other variables away from the embryos, too, from poisons to predators. It probably `costs' much less in DNA programming to adopt this strategy, too.

  This trick, evolved by the mammals, carries an important message. To ask how much information passes across the generations in the DNA blueprint, as textbooks and sophisticated research manuals often do, is to miss the point. How the genes and proteins are used is far more important, and far more interesting, than how many genes or proteins there are in a given creature. Lungfishes and some salamanders, even some amoebas, have more than fifty times as much DNA as we mammals do. What does this say about how complex these creatures are, compared to us?

  Absolutely nothing.

  Tricks like HSP90, and strategies like warm-bloodedness and keeping development inside the mother, mean that bean-counting of DNA `information' is beside the point. What counts is what the DNA means, not how big it is. And meaning depends on context, as well as content: you can't regulate the temperature of a uterus unless your context (that is, mother) provides one.

  The simple-minded `mutation' viewpoint, allied to trendy interpretations of DNA function in terms of `information theory', is often allied with ignorance of biology in other areas. One example is radiation biology and simple ecology as seen by `conservation activists'. Some of these volunteers found five-legged frogs and other 'monsters' downwind of the Chernobyl site, years after the nuclear accident but while radiation levels were still noticeably high. They claimed that the monsters were mutants, caused by the radiation. Other workers, however, then found just as many supposed mutants upwind of the reactor site.

  It turned out that the best explanation had nothing to do with mutant frogs. It was the absence of their usual predators, owls and hawks and snakes, because there were so many humans trudging about. Rana palustris tadpoles from Chernobyl produced no more of these pathologies than did other frogspawn samples from ponds some tens of kilometres away that had not been subjected to radiation, when a high percentage of both was allowed to survive. Usually, in British Rana temporaria frogs, it is very difficult to achieve ten per cent normal adults, or even ones that are viable in the laboratory, but they don't produce extra limbs as palustris does. It is normally the case, of course, that a female frog's lifetime production of some 10,000 eggs results in a few highly selected, and therefore `normal', survivors, and on average just two breeders. But conservationists don't like thinking about this reproductive arithmetic, with all those deaths.

  Here is another issue, again chosen from the thalidomide literature, that demonstrates how talk of Lamarckism, or of `mutations', misses the point.

  Some of the children affected by thalidomide have married each other, and several of these pairings have produced phocomelic children. The obvious deduction, from the folk-DNA point of view, is that the DNA of the first generation must have been altered, so that it produced the same effect in the next generation. In fact, this effect looks, at first glance, like Lamarckism: the inheritance of acquired characters. Indeed, it seems a classic demonstration of such inheritance, as convincing as if cutting off terriers' tails resulted in puppies being born with short tails. However, it is actually a lesson in not attempting to explain things `at first glance', like the conservationists did with the abnormal frogs.

  It is very tempting to do just that, when the idea of heredity in your mind is that one gene leads to one character, so if you've got the character yo
u've got the gene, and vice versa. Figures from the epidemiological literature suggest that in the space of a few years either side of 1960, about 4 million women took thalidomide at the critical time during gestation. Of those, about 15,000-18,000 foetuses were damaged; 12,000 came to birth with defects, and about 8,000 survived their first year. That is to say, the natural course of development selected just 1 in 500 who showed adverse effects. The proportion of children born with no detectable defect was much, much higher. And that fact changes our view of the likely reason for the children of two thalidomide parents to suffer from phocomelia, for the following reason.

  Conrad Waddington demonstrated a phenomenon called `genetic assimilation'. He started with a genetically diverse population of wild fruit flies, and found that about one in 15,000 of their pupae, when warmed, produced a fly with no cross-vein in its wing. These 'crossveinless' flies looked just like some very rare mutant flies that turned up occasionally in the wild, just as occasional genetically phocomelic children turned up before thalidomide. By breeding from the flies that responded to the treatment, Waddington selected for a lower and lower threshold of response. In a few tens of generations, he had selected flies that bred true for the cross-veinless trait, exhibiting it regularly without anyone warming the pupae. This may look like Lamarckian inheritance, but it's not. It's genetic assimilation. The experiments were selecting flies that had no cross-vein at lower and lower temperature thresholds. Eventually, they selected flies that had no cross-vein at `normal' temperatures.

  Similarly, genetic assimilation provides a much better explanation than Lamarckism for the phocomelic children of thalidomidemodified parents. We have selected, from some 4 million foetuses, those that respond to thalidomide with phocomelia. It is not surprising that when they marry each other, they produce a few progeny whose threshold is very low - below zero in fact. They are so liable to produce phocomelia that they do it without thalidomide, just as Waddington's flies came to produce cross-veinlessness without warming the pupae.

  One of the things that really worried Darwin was the existence of parasitic wasps - a fact that has influenced our Discworld tale, but has gone unremarked until now in the scientific commentaries. Parasitic wasps lay their eggs in other insects' larvae, so that as the wasp eggs grow into wasp larvae, they eat their hosts. Darwin could see how this might have happened on evolutionary grounds, but it seemed to him to be rather immoral. He was aware that wasps don't have a sense of morality, but he saw it as some kind of flaw on the part of the wasps' creator. If God designed each species on Earth, for a special purpose - which is what most people believed at the time - then God had deliberately designed parasitic wasps, whose purpose was to eat other species of insect, also designed by God. To be so eaten, presumably.

  Darwin was fascinated by such wasps, ever since he first encountered them in Botafogo Bay, Brazil. He eventually satisfied himself - though not his successors - that God had found it necessary to permit the existence and evolution of parasitic wasps in order to get to humans. This is what the quote at the end of Chapter 10 alludes to. That particular explanation has fallen out of favour among biologists, along with all theist interpretations. Parasitic wasps exist because there is something for them to parasitise - so why not? Indeed, parasitic wasps play a major role in controlling many other insect populations: nearly one-third of all of the insect populations that humans like to label `pests' are kept at bay in this manner. Maybe they were created in order for humans to be possible ... At any rate, the wasps that so puzzled Darwin still have much to tell us, and the latest discovery about them threatens to overturn several cherished beliefs.

  Strictly, the discovery is not so much about the wasps, as about some viruses that infect them ... or are symbiotic with them. They are called polydnaviruses.

  When mother wasp injects her eggs into some unsuspecting larva, such as a caterpillar, she also injects a solid dose of viruses, among them said polydnaviruses. The caterpillar not only gets a parasite, it gets an infection. The virus's genes produce proteins that interfere with the caterpillar's own immune system, stopping it reacting to the parasite and, perhaps, rejecting it. So the wasp larvae munch merrily away on the caterpillar, and in the fullness of time they develop into adult wasps.

  Now, any self-respecting adult parasitic wasp obviously needs its own complement of polydnaviruses. Where does it get them? From the caterpillar that it fed on. And it gets them (just as mother did) not as a separate infective `organism', but as what is called a provirus: a DNA sequence that has been integrated into the wasp's own genome.

  Many genomes, probably most if not all, include various bits of viruses in this way. Our own certainly does. Transport of DNA by viruses seems to have been an important feature of evolution.

  In 2004 a team headed by Eric Espagne worked out the DNA sequence of a polydnavirus - as one does - and what they found was dramatically different from what anyone had expected. Typical virus genomes are very different from those of 'eukaryotes'- organisms whose cells have a nucleus, which includes most multi-cellular creatures and many single-celled ones, but not bacteria. The DNA sequences of most eukaryote genes consist of `exons', short sequences that collectively code for proteins, separated by other sequences called introns, which get snipped out when the code is turned into the appropriate protein. Viral genes are relatively simple, and typically they do not contain introns. They consist of connected code sequences that specify proteins.-This particular polydnavirus genome, in contrast, does contain introns, quite a lot of them. The genome is complex, and looks much more like a eukaryote genome than a virus genome. The authors conclude that polydnavirus genomes constitute `biological weapons directed by the wasps against their hosts'. So they look more like the enemy's genome than that of an ordinary virus.

  Numerous examples, old and new, disprove every aspect of the folk version of evolution and DNA. We end with one that looks especially important, discovered very recently, and whose significance is just becoming seriously apparent to the biological community. It is probably the most severe shock that cell biology has received since the discovery of DNA and the wonderful `central dogma': DNA specifies messenger-RNA which specifies proteins. The discovery was not made through some big, highly publicised research programme like the human genome project. It was made by someone who wondered why his petunias had gone stripy. When all the world is chasing `the' human genome, it's not easy to get research grants to work on stripy petunias. But what the petunias revealed is probably going to be far more important for medicine than the entire human genome project.

  Because proteins are the structure of living creatures, and because as enzymes they control the processes of life, it has seemed obvious that DNA controls life, that we can `map' DNA code on to all the important living functions. We could assign a function to each protein, so we could assume that the DNA that coded for that protein was ultimately or fundamentally responsible for the corresponding function. Dawkins's early books reinforced the idea of one gene, one protein, one function (although he carefully warned his readers that he didn't want to give that impression), and this encouraged such media exaggerations as calling the human genome the Book of Life. And the `selfish gene' image made it entirely credible that huge stretches of the genome were present for solely selfish reasons - that is, for no reason related to the organism concerned.

  Biologists employed - as so many now are - in the biotechnology industries serving agriculture, pharmacy, medicine, even some engineering projects (we don't mean just `genetic engineering' but making better motor oils), all subscribe to the central dogma, with a few minor modifications and exceptions. All of them have been informed that nearly all of the DNA in the human genome is `junk', not coding for proteins, and that although some of it may be important for developmental processes or for controlling some of the `real' genes, they really don't need to worry about it.

  Admittedly, quite a lot of junk DNA seems to be transcribed into RNA, but these are just short lengths that sit about bri
efly in the cell fluids and don't need to be considered when you're doing important proteinmaking things with the real genes. Recall that the DNA sequences of real genes consist of a mosaic of `exons' which code for proteins, separated by other sequences called introns. The introns have to be cut out of the RNA copies to get the `real' protein-coding sequences, called messenger-RNAs, which lace into ribosomes like tapes into a tape player. Messenger-RNAs determine what proteins get made, and they have sequences on their ends that label them for making many copies of a protein or for destruction after only a couple of protein molecules.

  Nobody worried much about those snipped-out introns, just bits of RNA drifting aimlessly around in the cell till they got broken up by enzymes. Now, they do worry. Writing in the October 2004 Scientific American, John Mattick reports that The central dogma is woefully incomplete for describing the molecular biology of eukaryotes. Proteins do play a role in the regulation of eukaryotic gene expression, yet a hidden, parallel regulatory system consisting of RNA that acts directly on DNA, RNAs and proteins is also at work. This overlooked RNA signalling network may be what allows humans, for example, to achieve structural complexity far beyond anything seen in the unicellular world.

  Petunias made that clear. In 1990 Richard Jorgensen and colleagues were trying to breed new varieties of petunias, with more interesting, brighter colours. An obvious approach was to engineer into the petunia genome some extra copies of the gene that coded for an enzyme involved in the production of pigment. More enzyme, more pigment, right?

  Wrong.

  Less pigment?

  No, not exactly. What previously was a uniformly coloured petal became stripy. In some places the pigment was being produced, elsewhere it wasn't. This effect was so surprising that plant biologists tried to find out exactly why it was happening. And what they found was `RNA interference'. Certain RNA sequences can shut down a gene, prevent it making protein. It happens in many other organisms, too. In fact, it is extremely widespread. And it suggests something extraordinarily important.