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 16. Controlling segmentation (OVERHEAD).

Last time we looked at differentiation of the limb, and the way in which it was modified in various vertebrates to give a different configuration of bones, muscles etc. And before that we dealt with different patterning mechanisms which might set up a mechanism to control such differences.

One way that has been suggested to control pattern formation is the production of morphogens, that is small molecules that diffuse from cell to cell, that originate at a source in one place in the embryo and are resorbed at a sink in another place. This would set up a gradient, or gradients of morphogen across a field of cells, and cells might receive positional information which told them what to do. What to do might include switching various genes on or off, to make them into mesoderm instead of ectoderm or cartilage instead of muscle.

There is, of course, another possibility. If clones of cells are formed which represent an area or a structure ( in the way we know mosaic eggs do things) they might, perhaps using the mitotic timescale, regulate themselves and turn their own genes on and off, thus removing the need for morphogens which, awkwardly, no one is able to verify. If this happens then we need genes that do something new, turn other genes on and off, rather than make proteins. These do in fact exist, and are known as transcription factors.

Research on this hypothesis developed more or less accidentally in Drosophila. Drosophila is a small fruit fly used as a model for genetic experiments. Because it eats rotting fruit, can be bred in milk bottles, has a short generation time and only 4 chromosomes Drosophila has been a long term favourite with geneticists, and a formidable catalogue of mutations has been assembled.

Amongst these are a group, the homeotic mutations, which specialise in changing body parts into something else. Now Drosophila, being an insect, has a very specialised development which need not concern us here. Suffice it to say that the adult fly, the imago, develops from the cells of a series of discs within the grub. The grub does its own thing, and does not need the discs: mutant grubs lacking discs are healthy, but won't turn into flies.

Mutants are known which interfere with the development of these discs. Some of them just produce deformations, but homeotic mutants actually change the structure that should be there into something else (OVERHEAD). Aristopedia makes the feathery arista or antenna into a leg: ophthalmoptera turns an eye into a wing, In the 1940s it was found that many of these mutants are temperature sensitive, and that temperature shock can modify a structure up to the third instar, that is just before pupation.

If you dissect out these small discs and put them into an adult fly they divide but don't differentiate, so you can build up a stock of disc material.(OVERHEAD) If part of this is periodically put back into a grub it will differentiate. The odd thing is that as the disc material is cultured through successive generations of adults it changes. And not at random - it goes through a distinct series of changes, a hierarchy (OVERHEAD) .

Homeotic genes also affect body divisions. The classic work here was on a group of genes known as bithorax. The insect (OVERHEAD) thorax has three segments, the pro- meso and metathorax, or T1,T2 and T3, each distinguished by patterns of bristles and appendages . In the bithorax mutant the metathorax (T3) turns into a mesothorax (T2). Other regions are unaffected. This condition is known as the result of single mutations, temperature shocks and as a result of selection, presumably for the small, cumulative effects of many genes.

The affected unit here is actually not the segment but the front or rear half of the segment. In selection lines strongly affected individuals have all T3 made into T2 but others just have the front or rear half of the segment transformed.

There are five distinct alleles of bithorax with different effects arranged along the chromosome like this (OVERHEAD):-.
Bithorax ant T3 ant T2
Contrabithorax post T2 post T3
Ultrabithorax ant T3 ant T2

post T3 ant T2

ant A1 ant T3

Postbithorax post T3. post T2


Within this sequence are the seeds of a more logical order A1 1st, then moving anteriorly. Could it be that the normal logical order of the wild type had been distorted by time, duplication, unequal crossing over, deletion and other genetic trickery in the laboratory involved in making these mutations? Perhaps the order of the normal genes was such that they would be 'read' in an order corresponding to an anterio-posterior sequence.

The problem was proving it. Waddington, in reviewing this work in a book (New Patterns in Genetics and Development) was rather sniffy about 'geneticists who wish to explain everything without lifting their eyes from the chromosomes' and decried the idea. Which was a pity, because Lewis was quite right.

Subsequent analysis of homeotic genes in Drosophila and elsewhere led to some surprising findings. Homeotic genes form the basis of the segmental bodyplan.

This is probably the most successful bodyplan, both in numbers of phyla and numbers of species: that which combines bilateral symmetry with segmentation. It is seen in arthropods and vertebrates amongst others.

We know quite a lot about how this body plan develops through the study of Drosophila larvae. These have now, almost accidentally, become embryological tools: the normal development of the Drosophila larva involves a series of maternal and embryonic genes which were first collected and studied because of the odd effects seen when they went wrong, i.e. mutated.

Dorso-ventral polarity (OVERHEAD) is established in larvae by the products (protein or mRNA) of a series of maternal genes ( at least 10 including Toll, zerknullt, cactus) which are differentially distributed in the egg. Antero-posterior polarity is similarly governed by maternal genes (bicoid, okar, caudal, bicaudal). Mutations of genes in these sets produces larvae with missing or duplicated heads and/or tails, dorsal and ventral structures.


Once we know which is front and back and up and down a series of larval genes are expressed which govern segmentation (OVERHEAD). These include the bithorax locus and other homeotic genes. Segmentation seems to work in a complex way: first the front and rear body halves are identified, then each half is halved again. We can isolate mutants in gap genes (Krüppel, hunchback ) which establish domains (head, thorax, abdomen) pair rule genes (paired, even skipped, fushi tarzu) which remove every second segment, Bithorax and its mates establish segmental identity. Segment polarity genes (wingless, engrailed) which specify the polarity within a segment - so the unit, as with bithorax seems to be half a segment.

Changes in these genes produce homeotic mutations and the genes governing them, including the bithorax complex, are now named homeobox genes. Genetics has advanced since they were located and we can now find out much more about gene structure.

Homeobox genes are remarkable for being very heavily conserved i.e. the 180 nucleotides which make them up (coding for 60 amino acids) varies very little between genes, and indeed between species or even more widely separated organisms. For example 59 of the 60 amino acid residues specified by Drosophila genes are identical to those seen in frogs, and there is 70% identity between Drosophila and man. Don't be too impressed though, flies and man also share haemoglobin, specified by a much larger gene with less commonality, but still producing a very similar protein. Whatever homeobox genes make must be a small molecule whose amino acid residue order is all important: larger molecules just need key areas, like active sites in enzymes, or corners, specified. In fact what they make are products that bind to DNA - i.e. probably gene regulators or transcription factors which need to fit onto a DNA sequence accurately.

Homeotic genes are obviously worth studying further: in fact there are two very good reasons for studying them. First of all they tend to be very widely distributed in almost all phyla, and secondly they do other things as well as make segments. Let's take these in order.

In Drosophila homeobox genes are arranged in two major complexes (OVERHEAD) on the same chromosome, each containing about 9-11 genes evenly spread over 100-150 kilobases. The antennapedia group contains three genes, deformed, reversed sex combs and antennapedia, expressed in that order, and which control appendages on the head and the three thoracic segments. The bithorax complex contains ultrabithorax, abdominal a and abdominal b which specify mainly the eight abdominal segments. But if we look even at different insects, like the flour beetle, we find that this is a Drosophila speciality and most insects, and lots of other animals, just have one long homeobox region, HOM-C.

Because of this arrangement it is thought that a single region, HOM-C, probably present in the common ancestor of insects and arthropods has split into two.

In mammals 38 genes are recognised which have an antennapedia like sequence Remarkably these corresponding genes form four stable clusters on four separate chromosomes in lampreys, teleosts, anurans and mammals (SLIDE) And again the order in which the genes sit on the chromosome is the A-P order of the bits they control.

These four clusters are on four different chromosomes: this multiplication is thought to be due to splitting of HOM-C with some repetition having occurred in the process. The four different clusters seem, where present, to control different areas of development, or different systems within the same area.

Within each set of Hox genes in mammals and in Drosophila the genes located towards the 3' end of the chromosome are read before those towards the 5' end, and are concerned with more anterior structures. The homeobox gene sequences make RNA and the RNA makes a small polypeptide. This does not leave the nucleus but binds on to a specific region of DNA: it is made up of two helical parts with a turn between them, so may well bind over two copies of a gene on different DNA strands and so deactivate both copies.

The sequence of these genes in Drosophila is well known, but constantly being added to. At the moment the sequence along the chromosome is the old familiar one: bicoid acts early to establish A-P polarity, then switches on hunchback. Gap genes then segregate the embryo into domains, pair rule genes divide domains into parasegments (2 segment blocks) segment polarity genes specify the front end of each segment and homeoitic genes specify the characteristic of each segment.

These two complexes are thus read in A-P order and specify what goes on in different segments. When these genes mutate structures appear in the wrong place i.e.,. out of sequence. The homeotic genes appear to build the body using the first two thoracic segments as a ground plan. Deletion of all antannnapeda and bithorax genes produces individuals in which all thoracic, all abdominal and the posterior two head segments are unmodified T1. If only the bithorax region is deleted all segments past T2 stay as T2. Thus the first two thoracic segments are basic, and are sequentially modified by the homeobox genes.

It is tempting to make some jumps from this springboard. Could the duplication of the single, split hox group in arthropods and insects to the four in chordates be a key event triggering the origin of the chordate body plan? Does this mean a common ancestry for flies and man? The organisation of the rhombomeres (segments) in the chick and mouse hindbrain is such that boundaries correspond to the edges of homeobox gene expression. Rhombomeres behave as independent units, very much as insect segments do. It appears that it is easier to modify existing mechanisms than to build a new one.

There does not, however seem to be a simple 1:1 correlation between homeobox genes and specific areas in mammals. Hox genes are expressed in clearly segmented tissue, such as paraxial mesoderm, and in regions like the neural tube which is not segmented in the adult. Let us take as an example the developing vertebral column in mammals. In the mouse we have good data from 12.5 day old mouse embryos. We can use fluorescent dyes to indicate just where a particular gene product is being expressed i.e. where a hox gene is switched on. Since they run from anterior to posterior there is usually a sharp boundary where the gene is switched on, but a rather indistinct on at the posterior end of the area of interest, as whatever we are staining decays after synthesis ceases. Kessel & Gruss (OVERHEAD) assumed that up to about 9 vertebrae would show a particular gene as being active. Quite clearly more than one gene is involved in a particular vertebra. Now there are problems here, because there is disagreement between embryologists on the way in which the vertebrae form. For many years the classic theory, as we saw when we were talking about clones, was that vertebrae developed from contributions from two somites, the anterior half of the bone coming from the posterior part of one somite and the posterior part from the anterior half of the one behind. Recently this has been disputed and another theory of one somite one vertebra is more popular. In this case Gruss labelled his table with vertebrae: is there any correlation between Hox genes and the eventual shape of the vertebrae? After all mutations in homeotic genes transform vertebrae into others, just as they transform segments in insects.

We have developed a way of looking at shapes and differences between shapes mathematically, and applied this, at first, to the vertebrae from C1 to T2 (OVERHEAD) in the mouse. The first question that occurred to us was this: do any vertebrae share the same Hox genes, and if so are they the same shape? Yes, some of them (C3,4,5) do share a hox code, and the differences in shape are always minor. Remember that we are dealing with a pattern expressed at 12 days in the fetus and comparing it with adult mouse bones: muscles, blood vessels, nutrition, all sorts of factors act late in skeletal development and into the adult. Anyway similar Hox genes give similar vertebral shapes.

Next question: do vertebrae that differ from each other by one hox gene differ more in shape - yes. Do vertebrae that differ by two hox genes differ by more? Yes, it looks to be a linear relationship, except in the case of T2 which is an unusual bone.

Next question: Is the shape information in the Hox genes? This is a tough one, but we can get at it like this: some vertebrae are simple shapes, some are complicated by all sorts of processes. Is there a relationship between complexity of shape and the number of genes expressed? The answer seems to be no. Perhaps this isn't the right question, but it helps us to answer another. If what I have been saying about Hox genes is right, that they are highly conserved and that they are correlated with vertebral shape then why do related animals have such different vertebral shapes? Either the Hox code is different in different species, which seems unlikely, but testable or the shape differences are all late in origin: we ought to be able to test that as well by looking at the shapes of embryonic or juvenile vertebrae. In birds (OVERHEAD) we know that markers for the anterior limb occur at the same point in the hox series, so this many bird vertebrae are the same as 9 mouse ones as far as hox genes go. In swans there are 25 neck vertebrae - so perhaps they should look more similar to each other than mouse ones do? Or perhaps there should be repeats in pairs or triplets? We don't yet know.

A third alternative fits better with some Drosophila work. In Drosophila deletions of hox genes lead to the segment affected reverting to a T1. Does the same thing happen in mice? Somewhere there is either an archetypical vertebra or the plan for one (like the T1 segment in Drosophila) and the Hox code tells us how to modify it: more modification needs more genes.

And how do vertebral transformations work? We know that these occur in broadly the same way as do homeotic mutation in Drosophila - a different vertebra is substituted. Perhaps significantly we can cause these by adding retinoic acid to the embryo; this works because the retinoic acid induces hox genes. Homeotic genes are affected by growth factors and other potential morphogens as well. For example fibroblast growth factor, important in mesoderm induction, selectively activates posterior acting genes. Transforming growth factor B activates anterior acting genes. Retinoic acid induces anterior neural tissue to become posterior: it also activates posterior acting but not anterior acting homeobox genes in Xenopus, suggesting a role in regionalisation of nervous system and mesoderm - the tissues that determine body axis.

Genetic engineering is also possible: injection of homeoitic DNA into mice embryos leads to ectopic expression of the gene and the formation of additional structures, such as additional anterior vertebrae. Specification and induction are obviously coupled.

If we do this as an overkill experiment, blasting the embryo with retinoic acid at different stages we can produce transforms between some vertebrae. We don't understand the rules yet, but the following seem to be true

1. not all transforms are possible

2. transforms may be anterior or posterior

3. those that occur do so at different frequencies - perhaps some are less likely or more difficult?

4. partial transforms i.e. deformities are not uncommon.

It looks to me as if transforms occur by changing the Hox code: a one gene change might be easier (or more probable) than a two gene change. 'Deformed ' vertebrae are formed when we use a particular combination of Hox genes which do not normally occur - we recognise the new structure as a deformity. If this is so, then perhaps these new vertebrae resemble vertebrae from another animal where that particular code is used?

The scale of this problem is large (OVERHEAD). The tail of a mouse and that of a fish are both coded for by the same hox genes, the same hox genes, moreover, that code for the rear most parts of the fly's abdomen.

The anterio posterior axis is not the only one where we see Homeobox genes at work. In mammals different paralogues i.e. genes from different chromosomal groups tend to appear in different cell lineages - once again we have to ask if the duplication of these Hox genes led to new opportunities in body building. Hox genes appear in the formation of the head and the teeth, in feather germs and in limbs. They seem to be concerned with more than determining the body axis. Watch this space.

Bithorax alleles along the Drosophila chromosome:
Bithorax ant T3 ant T2
Contrabithorax post T2 post T3
Ultrabithorax ant T3 ant T2
Bithoraxoid post T3 ant T2
ant A1 ant T3
Postbithorax post T3. post T2

A more logical gene sequence would be:

post A1, ant A1, post T3, ant T3, post T2, ant T2, post T1, ant T1., or its reverse.

Dorso-ventral polarity is established in larvae by the products (protein or mRNA) of a series of maternal genes ( at least 10 including Toll, zerknullt, cactus) which are differentially distributed in the egg.

Antero-posterior polarity is similarly governed by maternal genes (bicoid, okar, caudal, bicaudal).

Mutations of genes in these sets produces larvae with missing or duplicated heads and/or tails, dorsal and ventral structures.First front and rear body halves are identified, then each half is halved again.

Gap genes (Krüppel, hunchback ) establish domains (head, thorax, abdomen).

Pair rule genes (paired, even skipped, fushi tarzu) make units of 2 segments.

Segment genes (Bithorax etc.) establish segment within pairs.

Segment polarity genes (wingless, engrailed) specify the polarity within a segment.

and so on.......

This page is maintained by Steve Paxton