Mutants

IN 1890 THE CITIZENS OF AMSTERDAM bought Willem Vrolik’s anatomical collection for the sum of twelve thousand guilders. It contained 5103 specimens, among them such rarities as the skull of a Sumatran prince named Depati-toetoep-hoera who had rebelled, apparently with little success, against his colonial masters. There was also a two-tusked Narwhal skull that had once belonged to the Danish royal family, an ethnographic collection of human crania, and the remains of 360 people displaying various congenital afflictions. Some of the specimens were adult skeletons, but most were infants preserved in alcohol or formaldehyde.

The Vrolik is just one of the great teratological collections that were built up during the eighteenth and nineteenth centuries. London’s Guy’s and St Thomas’s Hospital has the Gordon collection, while the Royal College of Physicians and Surgeons has the Hunterian; Philadelphia has the Mütter; Paris the Muséum d’Histoire Naturelle as well as the Orfila and the Dupuytren. Vrolik’s collection, which was given to the medical faculty of the University of Amsterdam, now occupies a sleek gallery in a modern biomedical complex on the outskirts of the city. What makes it unusual, if not unique, is that where most teratological collections are closed to all but doctors and scientists, the curators of the Vrolik have opened their collection to the public. In a fine display of Dutch rationalism they have decided that all who wish to do so should be allowed to see the worst for themselves.

And the worst is terrible indeed. Arrayed in cabinets, Vrolik’s specimens are really quite horrifying. The gaping mouths, sightless eyes, opened skulls, split abdomens and fused or missing limbs seem to be the consequence of an uncontainable fury, as if some unseen Herod has perpetrated a latter-day slaughter of the innocents. Many of the infants that Vrolik collected were stillborn. A neonate’s skeleton with a melon-like forehead is a case of thanatophoric dysplasia; another whose stunted limbs press against the walls of the jar in which he is kept has Blomstrand’s chondrodysplasia. There is a cabinet containing children with acute failures in neural tube fusion. Their backs are cleaved open and their brains spill from their skulls. Across the gallery is a series of conjoined twins, one of which has a parasitic twin almost as large as himself protruding from the roof of his mouth. And next to them is a specimen labelled ‘Acardia amorphus’, a skin-covered sphere with nothing to hint at the child it almost became except for a small umbilical cord, a bit of intestine, and the rudiments of a vertebral column. Until one has walked around a collection such as the Vrolik’s it is difficult to appreciate the limits of human form. The only visual referent that suggests itself are the demonic creatures that caper across the canvases of Hieronymus Bosch – another Dutchman – that now hang in the Prado. Of course, there is a difference in meaning. Where Bosch’s grotesques serve to warn errant humanity of the fate that awaits it in the afterlife, Vrolik’s are presented with clinical detachment, cleansed of moral value. And that, perhaps, suggests the best description of the Museum Vrolik. It is a Last Judgement for the scientific age.

THE CYCLOPS

Of Willem Vrolik’s published writings, the greatest is a full folio work that he published between 1844 and 1849 called Tabulae ad illustrandam embryogenesin hominis et mammalium tarn naturalem quam abnormem (Plates demonstrating normal and abnormal development in man and mammals). The teratological lithographs that it contains are of a beauty and veracity that have never been surpassed. The richest plates are those devoted to foetuses, human and animal, that have, instead of two eyes, only one – a single eye located in the middle of their foreheads. By the time Vrolik came to write the Tabulae he had been studying this condition for over ten years, had already published a major monograph on it, and had assembled a collection of twenty-four specimens – eight piglets, ten lambs, five humans and a kitten – that displayed this disorder in varying degrees of severity. Following Geoffroy he gave the condition a name that recalled one of the more terrible creatures in the Greek cosmology: the Cyclops.

Hesiod says that there were three Cyclopes – Brontes, Steropes and Arges – and that they were the offspring of Uranus and Gaia. They were gigantic, monstrous craftsmen who in some accounts made Zeus’ thunderbolts, in others, the walls of Mycenae. The Cyclopes of the Odyssey are more human and more numerous than those of the Theogony, but their single eye is still a mark of savagery. Homer calls them ‘lawless’. Polyphemus is more lawless than most: he has a taste for human flesh and dashes out the brains of Ulysses’ companions ‘as though they had been puppies’ before eating them raw. Homer does not identify the island where the renegade Cyclops lived, but Ovid put him on the slopes of Etna in Sicily and gave him an affecting, if homicidal, passion for the nymph Galatea. Painted on vases, cast in bronze or carved in marble, Polyphemus was depicted by the Greeks throwing boulders or else reeling in agony as Ulysses drives a burning stick into his single eye.

Many teratologists have linked the deformity to the myth. They argue that the iconographic model for the semi-divine monster was a human infant. Certainly the model, if it ever existed, must have been only faintly remembered. Differences in size and vigour aside, even the earliest representations of Polyphemus put his single eye where you would expect it, above his nose. But the single eye of a cyclopic infant invariably lies beneath its nose – or what is left of it. Others have argued, more or less plausibly, that the Cyclopes were inspired by the semi-fossilised remains of dwarf elephants that litter the Mediterranean islands.

Whatever its origins, Homer to Vrolik, the iconography of the Cyclops shows a clear evolutionary lineage. Homer’s Polyphemus is monstrous; Ovid’s is too, although he is also a sentimentalist. But within sixty years of the poet’s death in 17 ad, the Cyclops would appear in a different guise. It would become a race of beings that had ontological status, supported by the authority of travellers and philosophers. In 77 ad Pliny the Elder finished his encyclopaedic Historia naturalis. Drawing on earlier Greek writers like Megasthenes, who around 303 bc travelled as an ambassador to India in the wake of Alexander the Great’s conquests, Pliny peopled India and Ethiopia (the two were barely distinguished) with a host of fabulous races. There were the Sciapodes, who had a single enormous foot which they used as a sort of umbrella; dwarfish Pygmies; dog-headed Cynocephali; headless people with eyes between their shoulderblades; people with eight fingers and toes on each hand; people who lived for a thousand years; people with enormous ears; and people with tails. And then there were the single-eyed people: Pliny calls them the Arimaspeans and says that they fight with griffins over gold.

This was the beginning of a tradition of fabulous races that persisted for about fifteen centuries. By the third century ad, Christian writers had adopted the tradition; by the fifth century, St Augustine is wondering whether these races are descended from Adam. In the Middle Ages, the Cyclopes appears essentially unaltered from antiquity in manuscripts of wonder-books such as Thomas a Cantimpré’s De Naturis Rerum which was composed around 1240. In the fourteenth century, their biblical parentage is settled: they are the deformed descendants of Cain and Ham. Around the same time they appear in illuminations of Marco Polo’s travels (the Italian unaccountably fails to mention their existence); in the early 1500s one appears on the wall of a Danish church dressed in the striped pantaloons, floppy hat and leather purse of a late-medieval Baltic dandy. With time, the Cyclops becomes smaller, tamer and moves closer to home.

The first illustration of a cyclopic child, as distinct from a Cyclops, was given by Fortunio Liceti. In the 1634 edition of his De monstrorum he describes an infant girl who was born in Firme, Italy, in 1624 and who, he says, had a well-organised body but a head of horrible aspect. In the middle of her face, in place of a nose, there was a mass of skin that resembled a penis or a pear. Below this was a square-shaped piece of reddish skin on which one could see two very close-set eyes like the eyes of a chicken. Although the child died at birth she is depicted with the proportions of a robust ten-year-old, a legacy of the giants that preceded her.

Liceti describes another case of cyclopia as well, this time in a pair of conjoined twins whose crania are fused so that they face away from each other in true Janus style. Conjoined twinning and cyclopia is an unusual combination of anomalies, and one would be inclined to doubt its authenticity but for a 1916 clinical report of a pair of conjoined twins who showed much the same combination of features. And then there is the unusual provenance of Liceti’s drawing. It is, he says, a copy of one preserved in the collection of His Eminence the Reverend Cardinal Barberini at Rome, and the original, which now seems to be lost, was drawn by Leonardo da Vinci.

Looking at his bottled babies, Willem Vrolik recognised that some were more severely afflicted than others. Some had only a single eyeball concealed within the eye-orbit, but in others two eyeballs were visible. Some had a recognisable nose, others had none at all. Modern clinicians recognise cyclopia as one extreme in a spectrum of head defects. At the other extreme are people whose only oddity is a single incisor placed symmetrically in their upper jaw instead of the usual two.

The single eye of a cyclopic child is the external sign of a disorder that reaches deep within its skull. All normal vertebrates have split brains. We, most obviously, have left and right cerebral hemispheres that we invoke when speaking of our left or right ‘brains’. Cyclopic infants do not. Instead of two distinct cerebral hemispheres, two optic lobes and two olofactory lobes, their forebrains are fused into an apparently indivisible whole. Indeed, clinicians call this whole spectrum of birth defects the ‘holoprosencephaly series’, from the Greek: holo – whole, prosencephalon – forebrain. It is, in all its manifestations, the most common brain deformity in humans, afflicting 1 in 16,000 live-born children and 1 in 200 miscarried foetuses.

The ease with which foetuses become cyclopic is frightening. Fish embryos will become cyclopic if they are heated, cooled, irradiated, deprived of oxygen, or exposed to ether, chloroform, acetone, phenol, butyric acid, lithium chloride, retinoic acid, alcohol or merely table salt. In the 1950s an epidemic of cyclopic lambs in the western United States was caused by pregnant ewes grazing on corn lilies, a plant of the subalpine meadows which has leaves rich in toxic alkaloids. In humans, diabetic mothers have a two-hundred-fold increased risk of giving birth to cyclopic children, as do alcoholic mothers.

Most cases of cyclopia are not, however, caused by anything the mother did (or did not do) during her pregnancy. Mutations in at least four and perhaps as many as twelve human genes also cause some form of holoprosencephaly. One of these genes encodes a signalling protein called sonic hedgehog. This molecule received its name in the early 1980s when a mutant fruit fly was discovered whose maggot progeny had a surplus of bristles covering their tiny bodies. ‘Hedgehog’ was the obvious name for the gene, and when a related gene was discovered in vertebrates, ‘sonic hedgehog’ seemed the natural choice to a postgraduate student who perhaps loved his gaming-console too much. The sonic hedgehog mutations that cause cyclopia in humans are dominant. This implies that anyone who has just a single copy of the defective gene should have cyclopia or at least some kind of holoprosencephaly. But for reasons that are poorly understood, some carriers of mutant genes are hardly affected at all. They live, and pass the defective gene on to their children.

The fact that sonic hedgehog-defective infants have a single cerebral hemisphere tells us something important. When the forebrain first forms in the normal embryo it is a unitary thing, a simple bulge at the end of the neural tube – only later does it split into a left and right brain. This split is induced by sonic which, like so many signalling molecules, is a morphogen. During the formation of the neural tube, sonic appears in a small piece of mesoderm directly beneath the developing forebrain. Filtering up from one tissue to the next it cleaves the brain in two. This process is especially obvious in the making of eyes. Long before the embryo has eyes, a region of the forebrain is dedicated to their neural wiring. This region – the optic field – first appears as a single band traversing the embryo forebrain. Sonic moulds the optic field’s topography, reducing it to two smaller fields on either side of the head. Mutations or chemicals that inhibit sonic prevent this – thus the single, monstrous, staring eye of the cyclopic infant.

But sonic does more than give us distinct cerebral hemispheres. Mice in which the sonic hedgehog gene has been completely disabled have malformed hearts, lungs, kidneys and guts. They are always stillborn and have no paws. Their faces are malformed beyond cyclopic, reduced to a strange kind of trunk: they have no eyes, ears or mouths. These malformations suggest that sonic is used throughout the developing embryo, almost anywhere it is growing a part. It even seems to be used repeatedly in the making of our heads.

An embryo’s face is formed from five lumpy prominences that start out distinct, but later fuse with each other. Two of them become the upper jaw, two become the lower jaw, while one in front makes the nose, philtrum and forehead. These five prominences secrete sonic hedgehog protein. Sonic, in turn, controls their growth, and in doing so the geometry of the face. More exactly, it regulates its width. It sets the spaces between our ears, eyes and even our nostrils. We know this because chicken embryos whose faces are dosed with extra sonic protein develop unusually wide faces. If the dose is increased even further their faces become so wide that they start duplicating structures – and end up with two beaks side by side. Something like this also occurs naturally in humans. Several genetic disorders are marked by extremely wide-set eyes, a trait known as hypertelorism. One of these is caused by mutations in a gene that normally limits sonic’s activity. Patients with another hypertelorism syndrome even resemble the sonic-dosed chickens in having very broad noses, or else noses with two tips, or even two noses.

Disorders of this sort prompt the question of just how wide a face can be. If, as a face becomes wider and wider, parts start duplicating, might one not ultimately end up with a completely duplicated face – and so two individuals? It is not an academic question. One San Francisco-born pig arrived in the world with two snouts, two tongues, two oesophagi and three eyes each with an optic stalk of its own. It may have started out as two twin embryos that later conjoined in extraordinary intimacy. But given that the duplication was confined to the face and forebrain it may also have grown from a single primordial embryo, but one with a very wide head. The pig’s head is preserved in a jar at the University of California San Francisco, a suitable object for philosophical reflection. Was it one pig or two? It’s a question that would have stumped Aquinas himself. Not so the scientists who cared for the beast. They ignored the metaphysics, hedged their bets, and dubbed their friend(s) ‘Ditto’.

SIRENS

Among the disorders that appear regularly in the great teratology collections – the Vrolik devotes a whole cabinet to it – is a syndrome called sirenomelia. The name is taken from siren, the creatures that tempted Ulysses, and melia, for limb, but the English name, ‘mermaid syndrome’, is no less evocative. Instead of two good legs, sirenomelic infants have only one lower appendage – a tapering tube that contains a single femur, tibia and fibula. They resemble nothing so much as the fake mermaids concocted by nineteenth-century Japanese fishermen from the desiccated remains of monkeys and fish. More than Homeric echoes link cyclopia and sirenomelia. Just as cyclopia is a disorder of the midline of the face, a failure of its two sides to be sufficiently far apart, so sirenomelia is a failure in the midline of the lower limbs. A sirenomelic infant has neither a left nor a right leg but rather two legs that are somehow fused together.

The causes of sirenomelia are still not entirely known. But recently two groups of scientists independently engineered mouse strains that were defective for a particular gene. Unexpectedly, when the mice were born they had no tails and, just as sirenomelic infants do, fused hind limbs. To all appearances they were mermaid mice.

The mermaid mice were made by deleting the CYP26A1 gene. It encodes an enzyme that regulates a substance called retinoic acid. Most of the important molecules that control the construction of the embryo – that are a part of the genetic grammar – are proteins, long chains of amino acids. Retinoic acid, however, is not. Rather it is a much smaller and simpler sort of molecule, just a hydrocarbon ring with a tail. It is also one of the more mysterious of the embryo’s molecules. Because it is not a protein it has been difficult to study. For one thing, it can’t be seen in the embryo. The special stains that can be used to visualise proteins can’t be used for hydrocarbon rings. And then, because it is not a protein there is no ‘retinoic acid gene’ – no single stretch of DNA that directly encodes the information needed to make it. Instead there are just genes which encode enzymes that manufacture retinoic acid or degrade it – a frustratingly indirect relationship between gene and substance.

Even so, there have long been hints that retinoic acid is important. Embryos manufacture their retinoic acid from vitamin A – the need of which has been clear since 1932, when a sow at a Texas agricultural college that had been fed a vitamin A-deficient diet gave birth to eleven piglets all of which lacked eyeballs. Conversely, the consequences of too much retinoic acid became apparent in the 1980s when a related molecule called isotretinoin was extensively prescribed for severe acne. The drug was taken orally, and though its teratogenic effects were by this time well known some women took it while unwittingly pregnant. In one study of thirty-six such pregnancies, twenty-three superficially normal infants were born, eight ended in miscarriages, and five infants were malformed, their defects including cleft palates, heart defects, disordered central nervous systems and missing ears.

Some scientists have tried to repeat this unplanned experiment by bathing animal embryos in retinoic acid and then looking for malformations. Often the outcome is just a miscellany of deformities, rather like those shown by isotretinoin-exposed infants. But sometimes the results can be spectacular. If a tadpole’s tail is amputated, it normally grows another one in short order. But if the tail is amputated and the stump is painted with a solution of retinoic acid, the tadpole grows a bouquet of extra legs. This experiment clearly shows that retinoic acid is powerful stuff. It also suggests that tadpoles may use retinoic acid to regulate their rears. It does not, however, prove it. One could object that retinoic acid is, in effect, an exotic sort of poison, one that interferes in a completely unnatural way with the normal course of the embryo’s progress.

Hence the importance of the mermaid mice. They give, for the first time, some real insight into what embryos use retinoic acid for. It seems it is a morphogen, one of the most important in the embryo. Indeed, one might almost call it an Über-morphogen that acts the length and breadth of the embryo. Being a hydrocarbon ring, however, it works rather differently from most other morphogens. Where protein-signalling molecules are too big to enter cells and so bind to receptors on their surfaces, retinoic acid penetrates the cell membrane and attaches to receptors within the cell that go right to the nucleus where they turn genes on and off.

Where does retinoic acid come from? And what, exactly, does it do? The CYP26A1 gene encodes an enzyme that degrades retinoic acid. Thus CYP26A1-defective mice have too much of it. Their mermaid-like limbs are caused by an anomalous surplus of retinoic acid in the embryo’s rear. The rear of an embryo is not the only place affected by high levels of retinoic acid. Sirenomelic infants and mice also usually have head defects – implying that retinoic acid is normally lacking there too. Indeed, it is currently thought that could the concentration gradient of retinoic acid across an embryo be seen, it would resemble a hill with a peak somewhere near the embryo’s future neck and slopes in all directions: sides, front and back. It would show a carefully constructed topography maintained by a balance of enzymes that make and degrade the morphogen, which in frogs with extra legs, mermaid mice, sirenomelic infants and foetuses exposed to acne-medications has been eroded away leaving only an ill-defined plateau.

THE CALCULATOR OF FATE

The morphogens that traverse the developing embryo – be they protein or hydrocarbon ring – provide cells with a kind of coordinate grid that they use to find out where they are and so what they should do and be. A cell is thus rather like a navigator who, traversing the wastes of the ocean, labours with sextant and chronometer to find his longitude and latitude. But there is one difference between navigator and cell: while the navigator’s referents, the stars and planets, are always where they should be, the cell’s sometimes are not. Sirenomelia and cyclopia are two instances where mutation has warped the universe that cells refer to or even caused its total collapse.

Yet even bearing this difference (inevitable when comparing the clockwork motions of the physical world with the jerry-built devices of biology) in mind, the analogy still has force. For all the constancy of the heavens, navigators have always lost their way – perhaps because the instruments by which they read the heavens become maladjusted. In the same way, the receptors which allow cells to perceive morphogens and measure their concentrations can also go awry – and any number of congenital disorders are caused by mutations that affect them.

But perhaps the deepest level of the analogy comes when we consider the calculations that navigators must make in order to establish where they are. Cells, too, calculate – and they do so with great precision, absorbing information from their environment, adding it up and arriving at a solution. This calculator – one might call it a calculator of fate – is composed of a vast number of proteins that combine their efforts within each cell to arrive at a solution. Of course, the calculator is not infallible: just as navigators occasionally get their sums wrong, so too, occasionally, do cells.

The consequences of cells making mistakes of this sort are beautifully illustrated by one of the more curious pieces of erotica dug from the ruins of Herculaneum. It is a small marble statue – no larger than a shoebox – that depicts Pan the goat-god, whom the Romans knew as Faunus, raping a nanny goat. Masterfully combining the animal and the human in equal parts, the unknown artist has given his Pan shaggy legs, cloven hooves, thick lips, a flattened snout and an expression of concentrated violence. He has also given the god an unusual anatomical feature. Suspended from his neck, just above the clavicles, are two small pendulous lobes that in life would be no more than a few centimetres long.

These lobes, which are very distinctive, only appear in Pans of the second or third century bc, or, as in this statue (now in the Secret Cabinet of the Naples Archaeological Museum), in later Roman copies of Greek originals. The innumerable goat-gods who chase across the black- or red-figure vases of the Classical period wooing shepherds or grasping at nymphs do not have them, nor do the allegorical Pans of the Renaissance and Baroque such as those in Sandro Botticelli’s Mars and Venus or Annibale Carracci’s Omnia vincit Amor. Neck lobes would also be quite out of place in the beautiful but vapid Pans of the Pre-Raphaelites.

The origin of the god’s lobes is plain enough: they are echoed by an identical pair of appendages on his victim, the neck lobes frequently found on domesticated goats (German goatherds call them Glocken – bells). The sculptor of the original Pan Raping a Goat was clearly an acute observer of nature, and incorporated the lobes as one more detail to signify the goatishness of the god. Neck lobes, however, occur not only in goats but also, albeit rarely, in humans. In 1858 a British physician by the name of Birkett published a short paper describing a seven-year-old girl who had been brought to him with a pair protruding stiffly from either side of her neck. The girl had had them since birth. Birkett was not sure what they were, but he cut them off anyway and put them under the microscope, where he discovered that they were auricles – an extra pair of external ears.

Extra auricles are an instance of a phenomenon called homeosis in which one part of a developing embryo becomes anomalously transformed into another. The particular transformation that causes neck-ears has its origins around five months after conception, when five cartilaginous arches form on either side of the embryo’s head, positioned much where gills would be were the embryo a fish. Indeed, were the embryo a fish, gill arches are what they would become. In humans they form a miscellany of head parts including jaws, the tiny bones of the inner ears, and sundry throat cartilages. The visible, protuberant parts of our ears develop out of the cleft between the first and the second pair of arches. The remaining clefts usually just seal over, leaving our necks smooth, but occasionally in humans and often in goats, one of the lower clefts remains open and develops into something that looks much like an ear. The resemblance, however, is only superficial: the ‘ears’ have none of the internal apparatus that would enable them to hear.

Homeosis was first identified as a distinct phenomenon by the British biologist William Bateson, who in an 1894 book, Materials for the study of variation, coined the term and collected dozens of examples of such transformations. The Materials has something of the flavour of a medieval bestiary – Bateson called it his ‘imaginary museum’ – in which infants with supernumerary ears and heifers with odd numbers of teats jostle for space with five-winged moths, eight-legged beetles and lobsters that have antennae where their eyes should be. A strange book, then. Yet the Materials remains important to, and is cited by, molecular biologists in a way that few nineteenth-century zoological compendia are. This is because the transformations that Bateson identified pointed the way to one of the embryo’s most beautiful devices: the genetic programme that permits cells, and so tissues and organs, to become different from each other. Homeosis pointed the way to the calculator of fate.

The calculator of fate was first discovered in fruit flies. Flies, like earthworms, are divided into repeating units or segments. These segments are especially obvious in maggots, though metamorphosis obscures some of their boundaries. Many segments in the adult fly are specialised in some way. Head segments carry labial palps (with which the fly feeds) and antennae (with which it smells); thoracic segments carry wings, legs, or small balancing organs called halteres; abdominal segments have no appendages at all. The organs of a given segment are established when the fly is only an embryo, long before they can actually be seen. To put it a bit more abstractly, in the embryo each segment is given an identity.

Over the last eighty-odd years, Drosophila geneticists have sought and found dozens of mutations that destroy the identities of segments. Some of these mutations cause flies to grow legs instead of antennae on their heads – and make a fly that cannot smell; others cause halteres to become wings – and make a four-winged dipteran that defies its own definition. Yet other mutations cause wings to become halteres – and leave the fly irredeemably earthbound.

These mutations disrupt a series of genes that, in homage to William Bateson, have come to be known as the homeotic genes. There are eight of them, and they have names like Ultrabithorax, Antennapedia or, less euphemistically, ‘deformed’, that recall the strange flies produced when they are disrupted by mutation. They are the variables in a calculation that makes each segment distinct from any other.

The segmental calculator is a thing of beauty. It has the economical boolean logic of a computer programme. Each of the proteins encoded by the homeotic genes is present in certain segments. Some are present in the head, others in the thorax, others in the abdomen. The identity of a segment – the appendages it grows – depends on the precise combination of homeotic proteins present in its cells. The calculation for the third thoracic segment, which normally bears a haltere, looks something like this:

Which simply implies that Ultrabithorax is necessary if the third thoracic segment is to grow a haltere, that is, to be a third thoracic segment. Should the gene be crippled by a mutation, the protein that it encodes, if present at all, will be unable to do its work. The segment’s unique identity is lost; it becomes a second thoracic segment instead and carries wings.

When, in the 1980s, the homeotic genes were cloned and sequenced they proved to encode molecular switches: proteins that turn genes on and off. Molecular switches work by controlling the production of messenger RNA. Most genes contain information to make proteins. But this information requires a means of transmission. That is the job of messenger RNA, a molecule much like DNA except that it is neither double nor a helix, but only a long string of nucleotides. Messenger RNA is a copy of DNA, produced by a device that travels down gene sequences rather as a locomotive travels down a track. Molecular switches – or, to give them their proper name, ‘transcription factors’ – control this. Binding to ‘regulatory elements’, small, exact DNA sequences that surround every gene, transcription factors reach over to the molecular engine that makes messenger RNA and attempt to influence its workings. Some transcription factors seek to speed the engine up; others to shut it down. Attached to their regulatory elements, transcription factors face each other over the double helix and dispute for control. Like all negotiations, the outcome depends on the balance of power: the diversity of the opposing forces, or just their numbers.

The sequences of the eight fly homeotic genes are quite different. Yet each has a region, a sequence of only 180 base-pairs, that encodes, with small variations, the following string of amino acids:

This is the homeobox. In the sub-microscopic bulges and folds of a homeotic protein’s three-dimensional topology it is the homeobox sequence, nestling within the grooves of the double helix of the DNA, that brings the homeotic proteins to their targets, the hundreds, perhaps thousands, of genes under their control. Subtle differences in the homeobox of each protein allows it to control particular suites of genes.

The discovery of the homeobox in 1984, distinctive as a Hapsburg’s lip, suggested that the homeotic genes were all related to each other, that they were a family. Other animals, it quickly became apparent, had homeobox genes as well. They were found in worms and in snails, in starfish, fish, mice, and they were found in us. Perhaps they were present in the very first animals that crawled out of the Pre-Cambrian ooze a billion years ago. Most excitingly, if homeobox genes formed the circuits of the fly’s calculator of parts, might they not do so for all creatures, even for humans? Molecular biologists are not a breed much given to hyperbole, but when they found the homeobox, they spoke of Holy Grails and of Rosetta Stones.

They were right to do so. Another of Vrolik’s specimens, this time a skeleton, shows why. At first glance it seems a rather dull sort of skeleton. It isn’t bent with rickets or bowed with achondroplasia; there is nothing unusual about it (though its skull, limbs and pelvis have evidently long gone astray). It is only an undulating vertebral column with brownish ribs on a rusted metal stand – an altogether abject thing. It is not even on display in the public galleries, but lives in a basement where it is shelved with dozens of other skeletons accumulated over a century but now largely surplus to requirements. And yet this skeleton enjoys a quiet renown. Each spring it sees the light of day as it is displayed to a new batch of the Rijkuniversiteit’s medical students who are invited to identify its anomaly. This is surprisingly hard to spot, though obvious once pointed out – it is an extra pair of ribs.

Extra ribs have always caused trouble. In his Pseudodoxia epidemica Sir Thomas Browne relates how once, when the anatomist Renaldus Columbus dissected a woman at Pisa who happened to have thirteen ribs on one side, ‘there arose a party that cried him down, and even unto oaths affirmed, this was the rib wherein a woman exceeded’. ‘Were this true,’ Browne continues, ‘this would oracularly silence that dispute out of which side Eve was framed.’ The influence of Genesis II: 21–22 on popular anatomy has been a baleful one. I recently asked a class of thirty biology undergraduates (among them Britain’s best and brightest) whether men and women had the same number of ribs: about half a dozen of them thought not. ‘But,’ as Sir Thomas says with customary vigour, ‘this will not consist with reason or inspection. For if we survey the Sceleton of both sexes, and therein the compage of bones, we shall readily discover that men and women have four and twenty ribs, that is, twelve on each side.’ Just so. And yet extra ribs are surprisingly common: one in every ten or so adults has them (but they are no more or less frequent in women than men).

Most of us have thirty-three vertebrae. Starting at the head, there are seven neck vertebrae, then twelve rib-bearing vertebrae, then five vertebrae in the lower back, and another nine fused together to make the sacrum and coccyx or tail bones. In most people with extra ribs, this pattern is disrupted. A vertebra that normally does not bear ribs has become transformed into one that does. Sometimes this means the loss of a neck vertebra, sometimes the loss of one in the lower back; either way, homeotic transformations are much like the segment transformations that geneticists seek in their mutant flies.

It is no surprise, then, that the identity of each vertebra is controlled by homeotic genes much like those that keep a fly’s segments in order. Of course, matters are rather more complicated for us. Flies have only eight homeotic genes while mammals have thirty-nine, so many that the evocative Latinate names have been dropped: no Ultrabithorax or proboscipedia for us, but only the prefix Hox followed by unmemorable letters and digits: Hoxa3, Hoxd13 and so on. In mammals, as in flies, homeotic genes begin their work early in the life of the embryo. Vertebrae develop from blocks of mesoderm called somites that form on either side of the nerve cord like rows of little bricks. Each homeotic protein is present in just some of the somites. All thirty-nine are present in the tail somites, but then they fall away, in ones and twos, so that finally only a handful remain in the somites closest to the head. The vertebral calculator is not very economical. For the seventh neck vertebra it looks something like this:

Should a mutation cripple any one of the genes that encode these five proteins, the seventh vertebra will transform into its neighbour, the eighth vertebra, and gain a pair of ribs.

Distinguishing one vertebra from another is merely one instance of a problem that the embryo must solve repeatedly: the differentiation of parts along the head-to-tail axis. The embryo must solve this problem for the neural tube, uniform at first, but which later forms a brain at one end. It must solve it for the bones of the head – so that maxillae are formed next to mandibles and each is attached to its appropriate nerves and muscles. And it must solve this problem for the gut tube that becomes the stomach, liver, pancreas and intestines as well as the ventral blood vessel that becomes the four chambers of the heart. The Hox gene calculator is involved in all this.

How it works in mammals is known from mice in which one or more Hox genes have been deleted. Such mice are often profoundly disordered. Some have fore-limbs that are strangely close to their heads; others are missing parts of their hindbrains or cranial nerves. Some have hernias that cause their intestines to bulge into their thoracic cavities, or else open neural tubes. Some are missing their thymus, thyroid and parathyroid glands and have abnormal hearts and faces; some walk on their toes instead of on the soles of their feet, even as their hindquarters convulse uncontrollably. Most mice in which even one Hox gene has been deleted die young.

The Hox gene calculator is thought to work in humans in much the same way. The evidence for this belief is indirect and comes from a single 1997 study in which a group of London researchers stained six RU486 – ‘morning after pill’ – aborted embryos with molecular probes to reveal the times and place of homeotic gene expression. The embryos were four weeks old, about five millimetres long, and came from unwanted pregnancies. In autoradiographs of the sliced and stained embryos, Hox gene activity appears as grainy streaks and patches of white against the dark outlines of nascent rhombocephalons and pharyngeal arches. The patterns of Hox gene activity are just what one would expect from mice.

This is important and gratifying to know. But the study has not been repeated. Studies on human embryos are rare. In the United Kingdom they can only be done once formidable regulatory hurdles have been cleared; in the United States they can’t be done at all, at least not in federally funded institutions. The autoradiographs that are the raw data of such studies certainly have a disquieting quality about them. Perhaps this is because in death these embryos reveal a property – gene activity – that truly belongs to the living.

THREE THOUSAND SWITCHES

Writing of the ‘calculator of fate’ I have emphasised the roles of the thirty-nine Hox genes. But the human genome encodes some three thousand other transcription factors. Like the signalling molecules to which they respond, transcription factors come in families, of which the homeobox genes are only one. These transcription factors are the circuit components, the switches if you will, that are thrown as cells calculate their fate. This computational process is a progressive one in which the earliest cells of the embryo, naive and confronted with a world of possibilities ahead of them, are ever more channelled into becoming one thing rather than another.

Some of these calculations, such as those that go into the vertebrae, are understood; others we are just learning about. In 1904, a Tyrolean innkeeper slaughtered one of the chickens wandering around his yard and found that it had no fewer than seven hearts. A curiosity? Perhaps. But in 2001 it was discovered that if a gene called ?-catenin is deleted in mice, the result is an embryo with a string of extra hearts each of which beats and pumps blood. The extra hearts are made from tissue normally destined for the guts; and so a small part of another calculation – the one that decides whether a naive cell in the embryo becomes endoderm or mesoderm – stands revealed. Other disorders suggest the existence of calculations about which we know nothing. There is, it seems, a row of obscure glands in our eyelids (the Meibomian glands) that sometimes, albeit rarely, tranform into hair follicles. Infants who have lost their Meibomian glands have, instead, two or even three rows of eyelashes on each lid. It’s a trait that runs in families, but the gene responsible for directing eyelid epidermis into a gland rather than a hair follicle has not yet been found (and one doubts that anyone is looking).

And then there is Disorganisation. A mouse mutant of unparalleled obscurity – it has been the subject of only three papers – it is also one of the strangest. Three properties make Disorganisation strange. The first is the pervasiveness of its effects upon the mice that carry it. It would be gratuitously macabre to detail the appearance of these mutant mice: it is enough to say that the deformities of a single litter would embrace the contents of a sizeable teratology museum. And yet, the mutation is not inevitably lethal. Disorganisation’s second strange property is that no two mutant mice have the same set of defects. Some are hardly afflicted at all and can survive and breed, others are born mutilated but alive, yet others die in the womb. This variability extends to within a given mouse: a left kidney (or lung, or leg) may be destroyed even as its right cognate remains untouched. Finally, there is the strange propensity of the mutant mice to generate extra parts, not only supernumerary limbs (which can appear almost anywhere on the body), but also extra internal organs such as livers, spleens and intestines. They also have odd tumor-like structures embedded in their musculature and skin that seem to be the remains of supernumerary organs which never made it all the way. Is there a human Disorganisation gene? No human family showing Disorganisation-like properties seems to be known. However, some clinical geneticists have pointed to infants with especially bizarre suites of congenital anomalies as possible carriers of a cognate mutation. One such infant, a boy born in 1989, had nine toes on one foot and tumor-like pads of tissue scattered around his body. He also had a finger, complete with fingernail, growing from the right side of his ribcage. The Disorganisation gene has not yet been found, though it surely will be soon. Meanwhile, the mice speak. They tell of some critical, global, and quite unknown component of the embryo’s calculator of fate, one that has gone utterly awry.

MUTATIS MUTANDIS

The power of the homeotic genes over the number and kinds of body parts has led some scientists to propose that they must be important in evolution; that they have somehow, worms to whales, provided animals with their staggering variety of forms. There may be something to this. People with extra ribs, specifically those who have extra ribs located on what should be their necks, are, for example, a bit like snakes. Snakes don’t have necks at all: they have rib-bearing vertebrae that run all the way to their heads. This is because the pattern of Hox gene activity in the somites of snake embryos is quite different from that of necked reptiles, birds and mammals – a difference that also explains, incidentally, why snakes don’t have arms. The position of arms, more generally fore-limbs, is dictated by the same Hox gene calculation that decides the allocation of vertebrae between neck and ribcage. No neck, no arms; it is as simple as that.

The beguiling quality of the homeotic genes has, however, less to do with differences among species than with similarities. These genes have a universality that is simply breathtaking. Flies use them to order their segments; we use them to sort out our vertebrae – but in both there is the common theme of ordering parts along the head-to-tail axis of the body. The similarities between the homeotic genes of vertebrates and insects also go far deeper than their general uses: they go right to the genome.

Homeotic genes come as clusters: groups of genes arrayed side by side on a single chromosome. The first few genes in the fly’s homeotic cluster are involved in giving the head segments of the fly their identities; the next few genes along do the same for the thoracic segments; and the last few do the same for the abdominal segments. There is, it seems, a uncanny correspondence between the order of genes on the chromosome and the order of the fly itself. So, too, mutatis mutandis, is it for us. We have four clusters of homeotic genes on four chromosomes against the fly’s one, but within each cluster the genes preserve the order along the chromosome that their cognates have in flies. Just as in flies, the first genes of each cluster are needed for our heads, the last for our tails, and the rest for the parts in between.

Why the homeotic genes should work in this way, and why they should have stayed doing so, is not clear. Nevertheless, they point to a system of building bodies that evolved perhaps as much as a thousand million years ago in some worm-like ancestor and that has been retained ever since. Indeed, the homeotic genes were merely the first indication that many of the molecular devices that make our bodies are ancient. Over the last ten years it has become plain that we are, in many ways, merely worms writ large. A gene called ems is needed to make a fruit fly’s minute brain. So vast is the evolutionary gulf, both in time and complexity, between a fly’s brain and the hundred-thousand-million-neuron edifice perched upon our own shoulders, that one could hardly expect that the same devices are used in both. Yet mutations in a human cognate of ems cause an inherited disorder that results in a brain abnormally riven with fissures (and so mental retardation and motor defects). Another fly gene called eyeless is needed to make a fly’s compound eyes. Flies devoid of eyeless are, well, eyeless. So, in effect, are humans who inherit mutations in the cognate gene. They are born without irises.

In the cyclical way of intellectual fashion, all this has been said before, albeit far more obliquely. More than 150 years ago, that eccentric genius Étienne Geoffroy Saint-Hilaire – Linnaeus of deformity, discoverer of the universal law of mutual attraction – sought to construct a scientific programme, a philosophic anatomique, that would demonstrate that the animal world, seemingly so vast and various, was in fact one.

His initial goal was modest enough. Geoffroy attempted to show that structures that appear in mammals were the same, only modified, as those that appeared in other vertebrates, such as fish, reptiles and amphibians. In other words, he attempted to identify what we now call homologues, arguing, for example, that the opercular bones of fish (which cover the gills) were essentially the same as the tiny bones that make up the middle ears of mammals (the malleus, stapes and incus).

But opercular bones were small beer for a truly synthetic thinker: Geoffroy went on to find homologies between the most wonderfully disparate structures in the most wildly different creatures. Confronted with the exoskeleton of an insect and the vertebrae of a fish, he proposed that they were one and the same. To be sure, insects have an exoskeleton (all their guts inside their hard parts) while fish have an endoskeleton (bones surrounded by soft parts), but where other anatomists saw this as ample reason to keep them distinct, Geoffroy explained with the simple confidence of the visionary that ‘every animal lives within or without its vertebral column’. Not content with this, he went on to show how the anatomy of the lobster was really very similar to that of a vertebrate – if only you flipped it on its back. Where lobsters carry their major nerve cord on their ventral sides (bellies) and their major blood vessels on their dorsal sides (backs), the reverse is true for vertebrates. And then there was the curious case of cephalopods: if one took a duck and folded it in half backwards so that its tail touched its head (an exercise performed, I believe, on paper alone), did its anatomy not resemble that of a cuttlefish?

It did not. Geoffroy’s speculations attracted the wrath of Cuvier, his powerful rival at the Museum. The result was a debate in front of the Académie Française in 1829 that Geoffroy lost – a duck doesn’t look like a cuttlefish no matter how you bend it; even homologies between fish opercula and the mammalian middle ear didn’t bear serious scrutiny. Yet if the particular homologies that he proposed sometimes seemed absurd, even in his day, his general method was not. Different organisms do have structures that are modified yet somehow similar. Indeed, the idea of homology is so commonplace in biology today (we speak of homology among genes as easily as among fore-limbs) that it is easy to read into Geoffroy’s claims an evolutionary meaning he did not intend. The homologies that he saw, or thought he saw, were, as far as he was concerned, placed there by the Creator. It was the age of what would be called Transcendental Anatomy.

Today it is scarcely possible to study the development of any creature without comparing it to another. This is because animals, no matter how different they look, seem to share a common set of molecular devices that are the legacy of a common evolutionary history, that are used again and again, sometimes to different ends, but which remain recognisably the same wherever one looks. Indeed, the results of the genome sequencing projects suggest as much. Humans may have thirty thousand genes, but flies have thirteen thousand – a difference in number that is far smaller than one would expect given the seemingly enormous difference in size and complexity between the two species. Another creature much loved by developmental biologists, the nematode worm Caenorhabditis elegans, has nineteen thousand genes – even though the adult worms are only 1.2 millimetres long and have bodies composed of only 959 cells.

Some of Geoffroy’s specific ideas are even being revived. One of these is his notion – on the face of it utterly absurd – that a vertebrate on its four feet is really just a lobster on its back. In the previous chapter I spoke of the signalling molecules that oppose each other to form the front and the back of vertebrate embryos. These same molecules – more precisely, their cognates or homologues – also distinguish back from belly in fruit flies; but with a twist. Where in a vertebrate embryo a BMP4 signal instructs cells to form belly, in flies the cognate molecule instructs cells to form back. And where in vertebrate embryos chordin instructs cells to form back, in flies the cognate molecule instructs cells to form belly. Somewhere in the evolutionary gulf that separates flies and mice there has, it seems, been an inversion in the very molecules that form the geometry of embryos, one that looks uncomfortably like the kind of twist that Geoffroy postulated. Absurd? Perhaps not. It is the sort of uncanny correspondence that one comes to expect in an age of Transcendental Genetics.