Faculty of Biological Sciences, University of Leeds

These pages have been left in this location as a service to the numerous websites around the world which link to this content. The original authors are no longer at the University of Leeds, and the former Centre for Human Biology became the School of Biomedical Sciences which is now part of the Faculty of Biological Sciences.

Evolutionary Developmental Biology Lecture 11

Epigenetic common ground.(OVERHEAD)

Because a Bauplan stage is common to all metazoans, and because nearly all metazoans are related we are justified in thinking that similar epigenetic mechanisms and events are concerned with all development. Remember we had a list (OVERHEAD) of factors which would probably affect the developing egg. The important omission from this list, of these, common to nearly all metazoans (there are some parthenogenetic species) is fertilisation.


Fertilisation is a very unusual event: where else do we see the fusion of two cells, let alone the fusion of two cells from different organisms? Pre-osteoclasts unite in the skeleton to make multicellular osteoclasts and myotubes unite in muscle to form myocytes, but these are cellular fusions not nuclear ones.

The first epigenetic consequence of fertilisation is usually the completion of the meiotic reduction division which halves the number of chromosomes. In flatworms the egg is diploid at the time of sperm penetration: in sponges, dogs and foxes it is a little further on: in molluscs and many insects it is in the first stage of meiosis. In amphibians and most mammals it is at the second stage of meiosis. Only in cnidarians and sea urchins does sperm penetration occur at the end of meiosis. So a usual epigenetic consequence of sperm penetration is the completion of meiosis. If the egg is not penetrated it rapidly dies.

The important thing here is not strictly fertilisation i.e. the uniting of the egg and sperm nucleus but sperm penetration. If the egg is deluded, by pricking with a needle, or pH or temperature shock development proceeds until the end of cleavage - the blastula stage. The stored maternal information in the egg is enough to take the embryo thus far, then DNA from the zygote takes over.


So sperm penetration leads to maturation of the oocyte and the initiation of the early stages of development. Fertilisation, i.e. the union of egg and sperm nuclei is another consequence of penetration. This, and the possibility of genetic recombination is seen as the major 'advantage' of sexual reproduction. In order that this may occur in an orthodox fashion another consequence of penetration is the closing of the door - the depolarisation of the egg membrane to prevent the entry of other sperm.

And finally remember that the point of entry of the sperm defines the midline in many bilaterally symmetrical animals. The other important polarity, dorsal versus ventral is probably determined by the egg itself. Most eggs are not uniform but contain yolk and other things, in an organisation imposed during maturation of the egg in the ovary.(OVERHEAD) We can upset this by centrifuging the egg gently, when the heavy yolk separates out at the bottom with layers of other stuff above it. What happens next depends whose egg it was. In Tunicate eggs (OVERHEAD) the organisation will gradually return to something near normal and the egg can be fertilised and often develops normally. In amphibian eggs the yolk stays ventral but the egg will develop, albeit in an unusual way, to give a normal frog. So the yolk gradient in fact is not necessary to determine the dorsal/ventral axis, but is incidental. It also seems to determine the relative size of the various cells produced by cleavage (OVERHEAD).

Cleavage, however, is much more important than this. All eggs must cleave to produce a multicellular organism. In very few animals (except mammals and gastropod molluscs - the food supply to the small mammal and mollusc eggs is external throughout) is it a process of increasing size. It is a reorganisation of the egg into a number of cells. This reorganisation has a variable effect on the eventual organisation of the embryo.

The most extreme degree of control is seen in nematodes and tunicates (OVERHEAD, OVERHEAD). In these organisms each product of cleavage, or blastomere is strictly limited as to what it can become. There is thus no compensation for loss or damage. We can kill off particular cells and demonstrate that the same parts are always missing from the developing embryo. This is the so called determinate or mosaic egg.

In animals like frogs the picture is quite different, and blastomeres, if separated can make a whole, small, frog up to about the 8 cell stage. These are indeterminate or regulatory eggs.

Most indeterminate eggs show radial cleavage (OVERHEAD) which is not terribly precise in its geometry. The first division is often governed, as we said, by the entry of the sperm, the next is at right angles making a sort of four slice chocolate orange and the one after that equatorial - or rather anything between equatorial and Arctic circle according to the amount of yolk present. We then have eight cells, four rather smaller ones on top and four rather larger ones directly below. After this regularity suffers, because the larger cells divide more slowly and by the sixth division regularity has been largely lost.

Now this won't do at all in a mosaic egg, because loss of control would ultimately place organs where they shouldn't be. Most mosaic eggs have spiral cleavage. We have already seen the special Nematode cleavage (OVERHEAD) allowing a series of very precise divisions at odd angles to each other, followed by cell movements often over cell diameters. Each organ has a precise number of cells, and always the same ones.

Nematodes are unique in this but in molluscs and annelids (OVERHEAD) a simpler process results in a nearly identical result. In their version of spiral cleavage the plane of the mitotic spindle is not vertical, so the first four cells are slightly dissimilar and the third division results in two layers of cells which are alternate rather than sitting on top of each other (OVERHEAD). The mitotic spindle then moves through 90o

so that the next generation of cells is angled to the last. This results in a pattern rather like a herringbone brick wall, with each cell interlocked and in a precise position. The rates of cell division are precisely controlled too, so that at least up to 128 cells the blastula is the same (apparently) across the phylum.

So far most of the organisation seems to be down to the egg structure, or the maternal genome. We can test this in strains where mum has mutated. Most snails coil clockwise when viewed from the tip of the shell. In a few places, however we can find left handed shells. The inheritance of this a simple Mendelian affair with D dextral dominant over d sinistral. But the gene doesn't affect the snail itself, it affect its offspring - whatever their genotype. So a snail's coiling is dependant on the maternal genome, not its own (OVERHEAD) The gene affects eggs in the ovary, determining the direction of the obliquity of the spindle at the third cleavage division.

The problem of dextrocardia in man - and situs inversus viscera might have a similar origin.

Body cells versus germ cells

One early choice in development is whether cells should be put aside as germ cells or not. In many phyla this is the case, but in 9 smaller phyla, including the Sponges and some annelids and platyhelminthes this seems not to be necessary, and any cell can become a germ cell at any point throughout life. This is also true of all fungi and plants. Buss, who made a study of this matter came to the conclusion that those groups which split off the germ line early in development had fewer species than those that did not do so. Perhaps the splitting of and sequestrating of the germ line minimises the time during which potentially heritable changes can occur.

Should we wish to reserve germ cells how do we go about it? In anuran amphibians and Drosophila only cells which inherit a particular type of cytoplasm (germ plasm) can become gametes. In Drosophila the vital agent is known to be a protein (mw 95,000) carried in polar granules in the egg cytoplasm. Cells getting some of this during development will be turned into germ cells. The same goes for anurans (frogs and toads), although we don't know if it is the same protein. In urodele amphibians (newts and salamanders) things seem to be different - germ cells come from presumptive ectoderm during the formation of the mesoderm. These are not the same cells that contain germ cell activation factor in anurans - but it might be a different factor stored in different cells.

Subsequent events in development - the entry of the embryonic genome

The development of the embryo up to the blastula stage is determined maternally. The splitting of the germ plasm is determined by the maternal cytoplasm. But that is as far as an embryo can go without its own DNA. Conventionally once fertilisation has occurred the brand new genes of the embryo are expressed and govern the further development of the embryo. But is it as easy as that?

Let us take an example, from the mammals as it happens. To understand the switching on of the embryonic genotype we must look at a bit of genetics. Lets think of a cross between two individuals of the same, heterozygous genotype Hp/+, where + is normal and Hp is a mutant. Now conventional genetics says that crossing these to individuals works like this:(OVERHEAD)
Father Mother
Gametes Hp + Hp +
Offspring Hp/Hp Hp/+ +/Hp +/+


The last line gives us the classic 1:2:1 ratio.

Now the hairpintail gene (Thp, really a small deletion) is lethal in utero when homozygous, so we lose a quarter of the individuals by birth. We also lose some (about half) of the heterozygotes, who have a bent tail and polydactyly . So it looks as if two doses are very bad and one dose is quite bad.

At this stage, embarrassingly, I come on the scene. The hairpin gene was worked on at first by a rather unhappy colleague who went back to India leaving some breeding records: I took it over. Now because the cross on the overhead is wasteful (you only get small litters) in practice the stock was kept by crossing Hp/+ to +/+, giving a 1:1 ratio of hairpin heterozygotes and normals.(OVERHEAD) Sometimes this was done using a hairpin father and normal mother and vice versa.

While calculating the number of lethalities in these crosses I noticed that the figure was much higher when the mother was Hp/+ - in fact no Hp/+s at all survived from this cross. Ah, good I thought, a maternal effect killing the Hp/+ in an Hp/+ mother. Then I thought what about the Hp/+ x Hp/+ cross where half of them survive? Looking more closely into this I was able to split the Hp/+ animals into two kinds (OVERHEAD).

But what was going on? Ever since Mendel it had been assumed that the 2 in a 1:2:1 ratio was due to the Hp/+ and +/Hp being the same. In this case it wasn't. I had to rule out all sorts of things, then finally proved it by putting a genetic marker on the normal chromosome and proving that survivors got the marker from mum, and hence the Hp gene from dad. This was the first demonstration of such a phenomenon in mammals, and when I was quite sure I published it in fear and trepidation, because the other explanation, that I was a Wally and had missed something seemed even more likely.

But in fact I had discovered a phenomenon known as imprinting, when the maternal and paternal genes are switched on at different times.(OVERHEAD) This later became respectable: here is an example from man. Prader-Willi and Angelman's syndrome were known as two near opposites. They have something in common, both producing mentally retarded children, but there are features in one not seen in the other, and in general feeling and behaviour are more or less opposite. They are now known to be due to the same small deletion in chromosome 15, but inherited from different parents.(OVERHEAD)

If this can happen with a deletion it looks likely that the same situation occurs normally, and the paternal and maternal genomes don't switch on together. It also seems that for some genes maternal leads paternal and for others vice versa. Reik et al (Nature 328,248-251) looked at seven different genes and found that in one of them the level of methylation differed, being low in the paternally derived allele and high in the maternal one. Other workers found four out of five genes do this. Methylation is low in the adult testis, but not elsewhere. Obviously this persists, and can in fact be seen to switch back and forth as a particular gene copy is inherited via the male or female lines. It must persist into late development because paternal duplications (which mimic these genes) give rise to abnormalities which occur late in development. One of two models might apply: either methylation level differs only in genes with this paternal/maternal difference in inheritance, or it happens all over the place and some genes respond to it by showing changes in expression.

Anyway, once the genome starts to work it may well do so in an uneven manner. When the zygote DNA is switched on varies between taxa. We can find out by looking for RNAs from the zygote nucleus, but this is complicated by a vast array of maternal RNAs. In some phyla translation i.e. RNA formation is early, transcription (i.e. protein synthesis) is late. In other cases one seems to follow rapidly on the other.

Two models are again possible: either all cells get the same intact genome and parts of it are switched on and off by something during further development, or some cells get only a given selection of genes and so can only make certain products. In amphibians at least we know the answer to that one, because amphibians have large eggs. If you remove a nucleus from an embryo pregastrulation and put it into the cytoplasm of an enucleated egg a normal embryo develops.

Any nucleus that has passed over a marker in the embryo, the lip of the blastopore, and is therefore going on to produce mesoderm will only produce certain tissues - so something unique happens to this tissue at this point in time/space.

The formation of mesoderm is in fact one of the earliest inductive processes . In the blastula (OVERHEAD) only two of the three final germ layers exist, the ectoderm and the endoderm. Ectoderm ultimately forms those structures on the outside of the embryo, such as skin and the c.n.s. Endoderm forms the alimentary canal and any structures that bud off it, like the liver. Most of the remaining embryonic structures, the skeleton, muscles, blood vessels, gonads, heart and kidneys form from mesoderm. Mesoderm arises from ectoderm at the equator of the blastula as a result of an induction by the adjacent endoderm, without any major movements being involved. In Xenopus (OVERHEAD) the inducors are growth factors of the TGF- family and FGF which are localised within the endoderm and act upon the ectoderm - we shall talk about these later when we come to cell signalling.. Because the blastocoel separates most ectoderm from most endoderm the signalling is local, around the equator only. We know some of the things the signal says. It is both activational and repressive: it says switch on a muscle actin gene and a homeobox protein (to do with pattern) and switch off the gene for cytokeratin, an epithelial product. Adding lithium at this stage messes up the induction process (OVERHEAD).

Lithium makes more anterodorsal mesoderm from what is normally ectoderm. It also enhances expression of genes normally switched on at the anterior end of the cns and diminishes the expression of posterior markers. It thus realigns the body axis and produces, at different concentrations, a whole series of Bauplans with differing proportions of dorsal and ventral structures. So here is a clue: changing the concentration of a simple molecule - a metal ion in fact can give you alternative Bauplans.

We can't pin down mesoderm inducing factor to one chemical. FGF and FGF mRNA have been found in the embryo of Xenopus, but on its own this would only produce vegetal posterior mesoderm, and not dorsal mesoderm like the notochord. So we suspect TGF - is there as well, but we haven't found it yet. But again in the mouse a single induce is known which produces both dorso-ventral and antero-posterior polarity, according to dose.

Growth factors seem to act like this: they activate DNA-binding proteins. The DNA-binding proteins bind selectively to DNA and presumably activate the genetic program for mesoderm production.

Gastrulation has therefore produced two fundamentally different types of tissue. The ectoderm and the endoderm are epithelia (OVERHEAD), sheets of cells which are essentially two dimensional with an outer and an inner surface, the inner surface marked by a basement membrane (which is secreted by the cells as an extracellular matrix) and whose cells stand side by side linked by tight junctions, desmosomes etc. The sheet of cells is polarised and coupled. If we want to change an epithelial sheet we can't do much except bend it in one or two dimensions. Bending it in one dimension rolls it into a cylinder, with basement membrane on either side: bending it in two dimensions makes it into a sphere.

The other type of tissue, mesoderm or mesenchyme is rather different (OVERHEAD), consisting of cells forming a meshwork which may be more or less connected i.e. the cells may be spread out in a 3 dimensional net, touching only at the tips or they may be quite closely packed with more contact. The mesenchymal ECM is in the form of networks or matrices of peri or exracellular matrices surrounding otherwise isolated and unpolarised mesenchymal cells. These stages are not immutable: under certain conditions epithelial cells may become mesenchyme, (OVERHEAD) but the reverse transformation has not been observed.

The first act of differentiation in the developing embryo, the formation of mesoderm is initiated by the endoderm. This is usually termed induction (OVERHEAD) Further differentiation and morphogenesis, the events in which we have an interest are almost always initiated by the interaction of epithelial and mesenchymal cells (OVERHEAD). Such epithelial/mesenchymal interactions are often reciprocal, with one interaction setting up the next. These have been called epigenetic cascades. As the tissues and organs of the embryo develop interactions begin between adjacent systems, such as nerves and muscles, muscles and the skeletal system. These interactions integrate the various systems of the developing body. We can usually ascribe a given change to one of these levels: an epidermal/ mesodermal interaction, which tend to be early, or a later inter-system integrating interaction.

Differentiation of the chorda mesoderm.

The next induction that we see is between chorda mesoderm (the future notochord and mesoderm) and the ectoderm. Just as some ectoderm does not come into contact with endoderm during cleavage, and so is not induced to form mesoderm, some ectoderm does not come into contact with notochord during gastrulation, as so is not induced to become neural ectoderm. Instead it becomes epidermal ectoderm and forms epidermis. The ectoderm that becomes epidermis is not incapable of forming either mesoderm or neural tissue - it just normally is not induced to do so because it never meets the appropriate signal. If brought experimentally into contact with notochord it will become neuralised: this discovery brought Spemann a Nobel prize in 1934.

If you like this is the default position - what ectoderm will become if it is not persuaded to become something else.

What can we say about the inductive signal? (OVERHEAD) Well it seems to be all or nothing. Either tissue makes a nervous system or it doesn't: there is no stage where a little induction makes a tissue that is part nervous and part not. And at the boundaries of the nervous system there is a clean switch to epidermis: cells don't gradually become less nerve like and more epidermal, there is a clean break. At the break adjacent cells are neural and epidermal. Induction acts as a switch, a setter of thresholds or an establisher of boundaries.

There is also a time limit. If we manipulate notochord and epidermal ectoderm of various ages we find that the ability of the ectoderm to respond to the signal (competence) is limited. By the end of neurolation the window of opportunity has closed. Similarly the notochord will only work as an inducer: there are stages that are too early to induce and others that are too late. There is also the property of regionalisation. Notochord from gastrulae at any stage of gastrulation will induce nervous tissue, but notochord from early stages always induces forebrain, and later notochord induces spinal cord. We know that two different mechanisms are at work here, because the induction of nervous tissue does not need cell-cell contact, but the induction of a specific region does.

So there are two independent properties involved, the ability to induce and the ability to respond, and both are temporary states. There is also a double spatial limitation: only some regions can induce a particular change, and only some regions can respond. This will ensure, most of the time, that the right organ is formed in the right place at the right time. And added to this is, at least in the notochord/ectoderm induction a separate cell contact mediated regional inductive process.

As well as affecting the neurectoderm the notochord also has an influence on the mesoderm lying alongside it. (Except, that is, in the head. The head seems to have been rather a developmental afterthought tacked on in front of the notochord. Here much of the skeleton and connective tissue forms from the neural crest, whose cells lie initially alongside the developing c.n.s. and acts as pseudomesoderm. Cranial neural crest cells migrate away to populate the developing head. Trunk neural crest cells are less ambitious, forming spinal ganglia, sympathetic nervous system, pigment cells and the adrenaline forming cells of the adrenal gland).

Experimentally we can show that the notochord specifies the mesoderm immediately alongside the nervous system to become somitic mesoderm, which will later form paired somites, the piece beyond that to become intermediate, kidney forming mesoderm and mesoderm more lateral still to become lateral plate. This later splits into two layers, one associated with the epidermis which produces dermis and superficial muscles and one associated with the gut, which produces smooth muscle and connective tissue covering the gut.

This page is maintained by Steve Paxton