Morphogens: Molecules that tell cells in the embryo where they are
This essay was written by Jim Smith and was first published in the 1996 Mill Hill Essays.
Magazines and newspapers are full of articles and advice about how to make a baby and how to give birth to one. Much less is written, however, about what happens in the nine months between these two landmark events. Indeed, most people don’t think about the development of the human embryo at all unless something goes wrong with the process. Unfortunately, this is not such a rare event: recent figures show that fifteen percent of couples have difficulty conceiving a child, that twenty-three percent of human conceptions are lost in the first five months, and that two percent of children are born with a major developmental defect. Studying the development of the embryo is therefore of great medical importance because it is only by finding out what happens in normal development that one can explain, and possibly prevent, problems such as these. Why, for example, are babies born roughly the right size? Why do they have hands on the ends of their arms instead of feet? Why does the lens of the eye form so conveniently over the retina? And how does everything connect up? These are the fascinating problems addressed by the developmental biologist.
There are about two hundred different types of cell in a human being, and about a hundred million million cells altogether. There are skin cells and muscle cells, kidney cells and liver cells, nerve cells and hair cells, and so on. In coming to understand how we develop from the fertilised egg, therefore, one problem we have to solve is: how do cells know which cell type they should become? How do they decide, for example, to form kidney rather than liver? And a second question is: how does the embryo know how much to grow during its nine months in the womb? These are fundamental questions of biology that are addressed by cancer experts as well as developmental biologists, because cancer is a result of excessive growth and the return of specialised cell types to an unspecialised embryonic state – a sort of development in reverse.
But there is a third problem, unique to developmental biology, which until recently has received less attention than the control of growth and the generation of different cell types, and this is the question of how the right cell type forms in the right place. There is, after all, little point in making a hundred million million cells, comprising two hundred cell types, if they are all stuck together in a random and haphazard fashion. For the developmental biologist, this question of spatial organisation is the problem.
We can rule out immediately one mechanism by which the right cell type might form in the right place. It is perfectly clear to anyone who has looked at photographs of human foetuses that we acquire our shape gradually; we do not make a random heap of cells and then suddenly have them sort out to make a human being. But although modern techniques like ultrasound allow us to observe embryos developing in greater and greater detail, simply looking at embryos and describing the way they change cannot give us any real insights into how these things happen. To answer this question we have to do experiments – we have to say ‘what is the effect on the embryo if I disturb it in this way or that way?’. Obviously it would be unethical and unacceptable to do this to human embryos so developmental biologists have turned to a wide variety of other species. To an outsider attending a conference on developmental biology, it would perhaps seem strange that we can’t all agree on what is the best embryo to study. Why should some choose to study a worm and others a South African frog? What are the advantages of a fish from the Ganges over a fruitfly or a chicken or a mouse? It turns out that there are good reasons for studying each species, and the encouraging thing is that when you look at the results obtained with each, some general principles emerge. Among these is an idea that has been around for almost a hundred years. This idea is that cells are somehow aware of their position within the embryo, and that they use this ‘positional information’ to decide what cell type to become. How do cells learn where they are? One way seems to be by measuring the concentrations of substances that surround them.
In the same way that we can estimate how far we are from the baker or the fish and chip shop by noticing how strong is the smell of newly-baked bread or frying chips, so, if a cell were able to measure the concentration of a substance being produced by a specialised group of cells, it would be able to work out how far it was from that group of cells, the source of the substance. Obviously enough, the concentration of these substances will be high close to the cells that are making them, and lower as they diffuse further away. These special substances are called morphogens, because they influence the shapes of the structures that the cells of the embryo will form. To the developmental biologist, the appeal of this idea lies in its simplicity but, as is often the case, it is the simplest ideas which are the hardest to prove. Thus, although there have been tantalising hints over the last twenty years that morphogens do exist, there has been no definitive proof. It is only in the last eighteen months that good evidence has been obtained.
Developmental biologists work on frog embryos for many reasons. Like us they have a backbone so the lessons we learn from the frog are very likely to apply to us; the embryos are big, one and a half millimetres across, so it is easy to cut them into defined pieces; one can obtain frog embryos in large numbers, so there is no shortage of experimental material; they develop quickly, so we get results fast; and, of course, the embryos grow in pond water, so they are easy to keep. The early stages of frog development are like those of any other species, including man. After fertilisation, when the sperm meets the egg, the egg begins to divide. Ninety minutes later there are two cells, by four hours there are about a hundred, and by nine hours there are ten thousand. At this time, the embryo is just a ball of cells, with no obvious separation into different tissues. It turns out, however, that already the embryo has formed some different types of cell that can be recognised as different only because they make special proteins. These cells do not form the sorts of tissues we are familiar with, like muscle, bone, or blood. Rather, they are cells destined to form particular regions of the embryo.
The first cells of this sort to form in the embryo are called the ectoderm, the mesoderm and the endoderm. Anyone who has looked closely at newly-laid frog spawn will know that frog eggs have a darker-coloured half and a lighter-coloured half. The embryos tend to orientate themselves so that the darker half is up and the lighter half is down, an arrangement which makes them less visible, respectively, to predators from above and below. But what is inconvenient to predators is very useful to the researcher, because the darker half is that which gives rise, in every embryo, to the ectoderm, and the lighter half always goes on to form the endoderm. The mesoderm forms at the equator of the embryo, at the junction between the ectoderm and the endoderm. It is as if the spherical embryo is made up of three layers: ectoderm at the top, then mesoderm, and then endoderm. Or, increasing the scale somewhat, it is as if the ectoderm occupies that part of the globe north of the tropic of Cancer, the endoderm that part south of the tropic of Capricorn, and the mesoderm occupies the region between. The ectodermal cells eventually form the outer parts of the embryo, including the skin; the endodermal cells form the innermost structures, such as the gut; and the mesodermal cells form a layer in the embryo between the endoderm and the ectoderm, of which muscle is the major cell type. Of these three types of cells, the mesoderm is special because it is further subdivided into cells that will form the head of the embryo and cells that will form the tail. So how do cells in the very early embryo know to form head or tail? We now know that they measure the concentration of a morphogen.
The first important discovery was made by the Dutch developmental biologist Pieter Nieuwkoop about twenty years ago. Nieuwkoop showed that the endodermal cells of the embryo, those south of the tropic of Capricorn, secrete substances which cause the overlying equatorial cells to form mesoderm rather than ectoderm. He demonstrated this by removing ectodermal cells from the embryo from a region so far north as to be out of range of the substances, and growing them in contact with cells which would go on to form endoderm, the so-called future endoderm. If grown alone, each tissue would form what it normally forms: the future ectoderm would form skin and the future endoderm would form gut. When they were grown together, the fate of the endoderm was unaffected, but the ectodermal tissue formed mesodermal cell types instead. The interpretation of this experiment, therefore, is that contact with endodermal cells cause the ectodermal cells to form mesoderm.
Just looking at the mesoderm that is formed in these experiments shows that it includes both head and tail tissues. So Nieuwkoop’s experiments provide the first clue as to how the embryo makes a head and a tail. What they do not do is explain how the future head region forms in one place in the embryo and the future tail region in another. The solution to this problem awaited identification of the substances produced by the future endoderm.
To be honest we still don’t know the exact identity of these molecules, but one strong candidate is a protein called activin. If prospective ectodermal tissue from young embryos is incubated in a solution of activin it becomes mesoderm rather than ectoderm. If signalling by activin-like proteins is inhibited in the frog embryo, no mesoderm forms. Together with the fact that activin can be detected in the early frog embryo, these experiments show that an activin-like molecule is involved in mesoderm formation in the frog embryo.
At last then, we can turn to asking how head and tail cells come to form in different parts of the embryo. Or, to put it another way, how a head-specific protein is made in one part of the embryo and a tail-specific protein in another. The answer comes from two types of experiment. In the first, future ectodermal cells are exposed to different concentrations of activin. At low concentrations, neither the head nor the tail protein is made; the concentration is just too low to make anything happen at all. As levels increase, however, there comes a point at which cells begin to make tail proteins, but not head proteins. At still higher concentrations, there comes a point where cells switch to making head proteins only. In other words, there is a transition of cell type from tail to head. One remarkable observation is that just a two-fold increase in the concentration of activin is sufficient to divert cells from making tail proteins to making head proteins, from forming tail to forming head.
What might this mean in real life? The simplest and most attractive idea is that activin is made in early endodermal tissue and diffuses away towards the equatorial region of the embryo. Where the concentration is high, cells would make head proteins and become head, and where it is low, further away, cells would make tail proteins and become tail. And, sure enough, the head-forming cells are positioned closer to the endodermal tissue than are the future tail cells. But it is possible that a molecule like activin could diffuse far enough (over distances of ten to twenty cells) fast enough (within a couple of hours) and with enough precision (remember that cells are exquisitely sensitive to activin concentration) to make sure that each cell in the early embryo makes the right part of the adult.
To address these questions, scientists took two pieces of future ectodermal tissue and grew them in a dish in contact with each other. However, they were ectodermal tissues with a difference. All of the cells in one piece were made to produce activin. All the cells in the other piece contained a molecule that glows in ultraviolet light, so that they could be distinguished from the activin-producing cells. The two pieces of tissue were kept together for three hours, during which time activin should have had time to diffuse from one piece into the other, and then the positions of cells making the tail protein and the head protein were recorded. The results may be summarised simply. If low levels of activin were made by one of the pieces of ectoderm, the head protein was not made in the neighbouring ‘glowing’ tissue at all, but the tail protein was made close to the junction between the two pieces of ectoderm. If higher levels of activin were produced, head protein was made by cells at the junction, and the tail protein was only made some distance into the glowing tissue, about ten cells away from the activin-producing tissue. The results are precisely what would be predicted if activin diffused away from its site of synthesis and set up a concentration gradient. Where the concentration of this ‘morphogen’ is high, cells make head protein and become head, and where it is lower they make tail protein and become tail.
As well as examining frog eggs developmental biologists have for almost a century investigated development of the fruit fly. Why study flies? After all they are very different from animals with backbones and the embryos are tiny. On the other hand, flies like frogs develop quickly, and they have the big advantage that their life-cycle is only about ten days. This short life-cycle means that it is possible to study heritable changes – mutations – in genes which affect development. Normally such mutations are too rare for systematic study, but the frequency can be increased by treating male flies with a chemical which damages DNA – the genetic material. The offspring of such flies frequently carry mutations, making it possible to carry out large-scale ‘screens’ aimed at identifying every gene, and therefore every protein required for development of a particular structure
Recently, three proteins required for development of the fly wing have been studied. One of them is a molecule similar to activin, and this diffuses into surrounding tissues where different concentrations cause nearby cells to make the other two proteins. The three proteins are made in a nested pattern, like a set of Russian dolls, in the developing wing. The activin-like morphogen is made in cells running down the middle of the wing. Imagine straightening your arm and laying your hand, palm down, on the table. The morphogen would be expressed along your middle finger, in the middle of the back of your hand, and then in a straight line up the rest of your arm to the shoulder. The second protein is more widespread but is also centred on the midline of the wing, and the third protein is more widespread still. It is as if the cells making the activin-like morphogen were marked on one’s arm with a fine pencil, the cells making the second protein with a broad marker pen, and the cells making the third protein with a paint-brush.
Mutant flies that don’t make the activin-like morphogen don’t make the second and third proteins either, showing that the morphogen is required for their production. Now, if the morphogen works in the fly wing as it does in the frog mesoderm, one might suppose that it would diffuse from the midline cells that are making it into the surrounding tissues. Where levels of the morphogen are high, cells would make both the second and third proteins, and where it is low they would make only the third protein. This idea can be tested using exactly the same kind of experiment as used in the frog embryo. For example, if by genetic trickery a small group of cells is forced to make the activin-like morphogen in the wrong place, surrounding cells come to make the other proteins; cells nearer the source of morphogen make both and cells further away make just the third protein. If the amount of the morphogen produced in this small group of cells is reduced, the second protein is not produced in the immediately surrounding tissue at all, while the third protein is produced over a smaller area.
Together, these experiments on frog and fruitfly embryos reveal that the century-old idea of morphogens was correct. Cells learn where they are in the embryo by measuring concentrations of a substance produced by a special signalling region. In the frog embryo the signalling region is the future endoderm, and in the fruitfly it is the midline of the developing wing. There is little doubt that the same mechanism will be used over and over again during development, and indeed it already seems that morphogen gradients are established in the developing vertebrate limb, in the developing brain, and elsewhere.
It is sobering to think that it has taken almost a hundred years to provide evidence in favour of what seems such a simple idea, but I don’t think this should be taken as a criticism of developmental biologists! Sometimes the simplest ideas are the hardest to test, and the experimental proof has required use of recent advances in field as diverse as biochemistry, molecular biology, immunology and genetics. And finally, like any scientific experiment, these findings raise more questions than they answer. In my opinion the two most pressing are firstly, can we visualise the morphogens directly? and secondly, how do cells distinguish between two-fold differences of morphogen concentration so that they make different proteins and become different cell types? Fortunately, with the current pace of research in developmental biology, I don’t think we’ll have to wait a hundred years for the answers.