Of frogs and men
This essay was written by Tim Mohun and Jim Smith and was first published in the 2000 Mill Hill Essays.
Tyger Tyger burning bright, In the forests of the night; What immortal hand or eye, Dare frame thy fearful symmetry?
William Blake (1757-1827)
The idea that we each arise from a single, tiny egg has long been a source of inspiration for artists and scientists alike. The events that transform the fertilised egg into a perfectly formed individual have provoked awe and fascination in equal measure and it is perhaps not surprising that the desire to understand embryonic development has been a central quest of modern biology. Its pursuit has resulted in discoveries that have illuminated many different branches of biology, from studies of evolution to the nature of genetic disease. Whilst the methods used by scientists may have varied with progress, time and again, decisive advances have come from the study of one familiar subject; the simple frogspawn that many of us remember collecting in jam jars from the local pond.
It is generally estimated that the human body contains about two hundred different types of cell, each specialised to perform distinct functions. Almost all of these are formed during embryonic development. The simplest and most profound question biologists have sought to answer is how does this diversity of cell types arise? Put another way, since all the cells of an embryo originate from repeated division of a single fertilised egg cell, what determines the fate of an individual embryo cell?
Anyone who has watched the development of frogspawn through a magnifying glass will have seen how quickly embryos change from clusters or balls of apparently similar cells to complex structures containing distinct tissues and possessing an increasingly recognisable shape. For such a remarkable transformation to occur with such predictable precision each time an embryo develops, suggests that each cell “knows” its fate in a reliable manner. Whether cells always know what tissue or organ they will contribute towards or whether they “learn” their fate as a result of their position within the embryo is a question that intrigued natural philosophers long before the advent of experimental biology. At one extreme, we can imagine that the entire body plan lies prefigured in some manner within the fertilised egg, cell division serving effectively to parcel up appropriate fates to daughter cells. The alternative view is that embryo cells “learn” their fate during the course of early embryo development, through interaction with neighbouring cells and their environment. Rather than independently following unique instructions inherited from the egg, cells in the developing embryo might interact with their neighbours, and this communication could provide the basis for establishing their subsequent fate.
Observation alone cannot distinguish these possibilities and embryologists in the last century looked instead to test the importance of cell neighbours, either by removing cells from early embryos or moving them from one part of the embryo to another. Such experiments needed embryos that were large enough to make microsurgery possible and robust enough to survive the procedure. In addition, the embryos had to be readily available from the earliest developmental stages. Early mammalian embryos were of little use, being far too small and inaccessible within the pregnant mother. Even chicken embryos were of limited use since the earliest stages of development were known to occur before the egg is laid. In contrast, the eggs of frogspawn are large and numerous. Furthermore, protected by their individual jelly coats, they undergo the entire course of development externally in the pond water, from fertilisation of the egg to formation of swimming tadpoles. Not surprisingly then, amphibian embryos rapidly became the favourite subject for experimental embryologists.
An obvious characteristic of frogspawn is that the newly laid eggs are similar in size to the hatching tadpoles. All the nutrients required for embryo development are already present within the fertilised egg, stored in the form of yolk protein molecules. As the egg divides, each daughter cell receives half of these nutrient stores and the process is repeated at each successive cell division. As a result, amphibian embryos, unlike those of mammals or birds, do not actually grow. Until the tadpole starts to feed, all cell divisions simply cut the cell in half. Whilst the number of cells in the tadpole is in the order of perhaps one hundred thousand, its mass is the same as, or, allowing for metabolic processes, slightly less than, the mass of the fertilised egg.
One consequence of this is that if individual cells are marked with a non-toxic dye, this will be inherited by daughter cells without dilution and remains detectable much later in development. With this simple technique, embryologists have constructed detailed maps describing the fate of different cells and regions of the early amphibian embryo. The absence of growth has other advantages for the experimenter. Since cells contain their nutrients, they will continue to divide in a culture medium, which need be no more than a simple salt solution. This provides one test to establish at what stage embryo cells know their future fate. If, after removal from the embryo, they continue to form the cells or tissue predicted by the “fate map”, at the time they were placed into the culture medium such cells must have already understood their position in the future body plan. On the other hand, failure to specialise in the expected way, suggests that the cells do not yet know their normal fate. Studies of this kind formed the foundation of experimental embryology and remain the basis of research to this day.
From work of this kind, we now know that, to a large degree, the complex pattern of cell specialisation seen within the amphibian embryo does not arise through the autonomous activity of individual cells, programmed to “know” their fate by virtue of their lineage from the fertilised egg. Rather, it depends critically upon signals passing between neighbouring cells. These first establish broad outlines of the embryo body plan, which are progressively refined as the embryo develops. The gradual acquisition of “pattern” or cellular identity is sufficiently precise that its results are virtually identical each time an embryo of a particular species develops. Yet as experimenters quickly found, it is also sufficiently flexible that it can often accommodate considerable natural or experimentally-induced variation in both the size and structure of individual embryos.
Subsequent studies of embryonic development in species ranging from fish to mammals have shown that similar mechanisms lie at the heart of embryo development in all vertebrates. The goal of developmental biologists has been to identify the nature of the signals that pass between cells of the early embryo and to establish how they direct cells down a particular developmental path. Remarkably, as we have gained insights into the molecules involved, it has become clear that similar fundamental mechanisms guide embryo development in creatures as diverse as flies and man. In short, we can be confident that the lessons we learn from studying frogspawn will advance our knowledge of human embryo development.
Before the Second World War, most embryologists studied embryos conveniently obtained from local amphibian species. Those from newts and salamanders were particularly attractive, due to their size and their slow rate of development. Whatever the species used, the supply of embryos was necessarily limited to the annual breeding season and as biochemical approaches became popular, demand for more plentiful supplies increased. Fortunately, in 1931, during the course of some endocrinology studies, investigators found that extracts of the anterior pituitary gland would induce egg laying in the South African clawed frog, Xenopus. The active agent turned out to be luteinising hormone, a hormone produced by mammals during pregnancy and excreted into the urine. This discovery had several consequences. Firstly, it led to the development of a pregnancy test, based upon the injection of urine extracts into female frogs. As a result, large numbers of Xenopus were taken for use in the laboratory, raising fears that the wild population would be decimated. Happily, immunological approaches eventually provided a more sensitive and reliable assay for luteinising hormone and the Xenopus test was abandoned. Its legacy was to establish Xenopus as a common laboratory animal and for developmental biologists the ability to induce egg-laying all year round made Xenopus the amphibian of choice for embryology experiments.
At the Institute, Xenopus were initially used to study how individual nerves make connections specifically with their target tissues, using the development of the tadpole visual system as a model. Research into the actions of steroid hormones prompted studies of tadpole metamorphosis. The profound changes that transform a tadpole into a frog proved to be controlled by thyroid hormone. Most striking amongst these was the loss of the tadpole tail, which is a result of hormone-induced tail-cell death. These research projects led to the establishment of a Xenopus laevis frog colony at the Institute enabling others to join in studying amphibian embryo development. Other pioneering work with Xenopus concentrated on spatial patterning in the early embryo. Using microsurgery to perturb normal development this work centred on the problem of how blocks of body muscle were formed during the first two days of embryo development. The visible “blocks” are arranged along the back of the embryo, on either side of the embryonic backbone. They are formed progressively from head to tail at a precise rate and in exact numbers. A mechanism to account for the precision in their pattern was sought by interfering with their formation. As is often the case, the conclusions of this work had broader implications for developmental biology since many other structures of the vertebrate embryo show evidence of similar repetitive patterning and scientists around the world are today seeking to identify the molecular signals responsible.
We have seen that the fate of different regions from the early Xenopus embryo can be reliably predicted and furthermore, because of their nutrient stores, embryo fragments can be readily cultured in isolation. Embryologists were quick to see that these properties provided a simple way of testing for and identifying signalling molecules. Specific extracts or purified molecules could be added to the culture medium and tested for their ability to alter the developmental fate of embryonic cells. Precisely this approach was used by the German embryologist Tiedemann, who showed that an extract of ten-day old chicken embryos would cause future skin cells to form muscle and embryonic backbone. This tantalizing result showed that inducing factors could indeed be derived from embryonic tissue. Unfortunately the nature of the assay made systematic study almost impossible and little progress was made in isolating the active factor.
A similar example of signalling between cells had also been described in amphibian embryos. After five hours of development, the frog embryo comprises little more than a ball of a few thousand cells, with a central internal cavity in its uppermost half. From fate maps we know that cells forming the roof of this cavity will normally form skin or nerve tissue, whilst cells from the opposite pole will form gut and liver. Cells in between, forming a belt around the equator of the embryo contribute to muscle, kidney and backbone. Amphibian embryologist Pieter Nieuwkoop elegantly demonstrated that sandwiches of cells from the two poles recreated equatorial-like tissue. In other words, the “prospective” skin/nerve tissue from the top of the embryo was redirected to form muscle, kidney and backbone by the “prospective” gut/liver tissue from the bottom. With a plentiful supply of embryos available to prepare extracts, several laboratories set out to identify the signalling molecules Nieuwkoop’s work had shown must be present.
The successful search for Nieuwkoop’s factor began at the Institute in 1984. The source of signalling activity finally identified did not come from an embryo extract, but instead from a type of Xenopus cell that originated from tadpoles but had been grown in laboratories for many years. When prospective skin cells from the five-hour old embryo were placed in contact with a pellet of these cells, the embryo cells were redirected to form muscle tissue. Indeed, simply exposing the embryo cells to the fluid in which the cultured cells had been grown was sufficient to induce muscle cells to form. Importantly, this showed that the signalling factor was secreted by the cells into their surrounding medium and could in principle be isolated. Purification turned out to be easier said than done, in part because the activity was so potent. However, after three years of efforts, one hundred and sixty-five litres of culture medium yielded 1.6 millionths of a gram of pure protein. This was then identified by protein chemist colleagues at the Institute as a hormone called activin.
Activin was already known to endocrinologists as a hormone responsible for regulating the production in the pituitary gland of another hormone, called follicle stimulating hormone, which is required for the maturation and release of eggs from mammalian ovaries. This new result was the first indication that activin might also be involved in early embryonic development. Perhaps as important, activin became a powerful experimental tool to study the mechanism by which the fate of equatorial cells, collectively known to embryologists as mesoderm tissue, was established. It still remains uncertain whether activin itself is indeed the signalling molecule present in early embryos, or whether it fortuitously mimics the real factor. However, subsequent studies have identified several other proteins present in the early embryo that are related to activin and possess similar properties. Fate maps demonstrate that the equatorial region of the early embryo forms different types of mesoderm tissue in a reliable pattern. Equatorial cells from one side of the embryo form the embryonic backbone, adjacent cells form muscle and cells from the other side form kidney. If all mesoderm cell types are induced by activin, how is such patterning to be explained? This type of problem had previously been addressed in the work of colleagues at the Middlesex Hospital Medical School, who proposed that a single signalling molecule might have different effects on target cells, depending on its concentration. Just such an effect was found with activin. Low concentrations tended to produce tissues such a kidney and blood, normally formed on the side or the belly of the tadpole. Higher concentrations formed tissues from the back of the tadpole, such as muscle and embryonic backbone.
How do molecules such as activin alter cell fate? With the advent of molecular biology it became possible to pose this question in a more specific way. As cells specialise, they begin to make the particular set of proteins needed for their unique function. Muscle cells will, for example, make all the proteins needed to form contractile muscle fibres along with those needed to provide the energy to drive muscle contraction. They do so by activating the appropriate set of genes that encode these proteins. Signals directing embryonic cells to a particular fate must therefore change the pattern of genes expressed in the embryo cell, activating some and perhaps shutting down others. These changes do not seem to occur in a single step, rather they appear to be established in a series, with some genes encoding proteins that regulate others lower in the sequential hierarchy. The final result is differentiation of a specialised cell type, a process that is generally irreversible. From the perspective of molecular biology, cells “know” their eventual fate when signals trigger such cascading changes in gene expression. Understanding how signals like activin establish or change cell fate means identifying the genes at each step in the process. This has now become a feasible goal and once again, experiments with amphibian embryos have provided decisive advances.
One way to test the developmental role of any particular gene is to see what effect the protein it codes for has on cells that do not normally make it. Xenopus embryos lend themselves to this type of experiment. Cloned nucleic acid encoding the protein can be microinjected into the embryo, causing the protein to be made in all the injected cells and their descendants. DNA cloning methods can be used to alter the microinjected nucleic acid in order that the embryo cells now make a mutant protein. This may be a super-active form of the protein, but it can also be a form that inhibits the function of the normal protein. By examining the effect of normal and mutated versions of a gene product on embryo development, we can begin to identify the role it plays in early development. For example, we now know the identity of a few genes which are switched on in skin cell precursors by exposure to activin. One of these, called Brachyury, encodes a protein which, on its own, can trigger undecided embryo cells to form muscle or notochord. This suggests that synthesis of the Brachyury protein is a very early event in the developmental pathway leading to these cell types. Consistent with this idea, the protein appears to act by binding to DNA, controlling the activity of other genes lower down in the developmental pathway.
How can we be sure that a gene we identify actually plays the role we suspect from these sorts of studies? One simple way is to ask what effect removing the gene has on normal development. Mutations that have this effect can be induced randomly into the genetic material, by chemical treatments or exposure to radiation. Individual mutants can then be identified for further study, by searching through the embryos of treated animals and identifying those which fail to develop the particular organ or cell type of interest. This approach was pioneered by geneticists using fruit flies and has been applied more recently to vertebrate development using embryos of the zebrafish. An alternative strategy is the selective mutation of individual genes using genetic engineering, a procedure that has become well established in studies of mouse embryo development. Whether mutation is random or directed, a crucial aspect of each method is the ability to breed and maintain populations for each individual genetic mutation. This has proved impossible with Xenopus laevis, not least because it generally takes at least a year to raise sexually mature adults from embryos. However, frogs of the related species Xenopus tropicalis have bred in less than four months, yet seem to retain many if not all the experimental advantages evident in their more slowly developing cousins. The combination of genetic, molecular and embryological approaches to study early development in a single vertebrate is a powerful and exciting prospect.