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 13: Cell behaviour and pattern (OVERHEAD)


O'Higgins, Johnson & McAndrew (1986) Journal of Embryology and experimental Morphology 96,171-182

Hinchliffe & Johnson (1980) The Development of the Vertebrate Limb. Clarendon, Oxford.

Last time we looked at pattern and at the end I asked what sorts of behaviour do cells exhibit to other cells and could modifications in this behaviour change the way in which embryos develop, perhaps defining pattern?

Let's look at a series of simple experiments. Remember the clever trick with the sponge strained through a cloth (OVERHEAD)? We can do a similar experiment with a mixture of adult cells from , say, a chick which have been enzymatically separated.

After a period of time a group of a million or so mixed cells in culture will separate into groups of single cell types. If these groups are wrapped in ectoderm jackets and grown on cells separate into groups and remarkably complete organs form.

What cells recognise as 'like' is variable. A mixture of kidney and cartilage cells from chick and mouse will aggregate by type not species. But a mixture of cells from various newts aggregates according to species. And timing is important. A mixture of embryonic ectoderm, endoderm and mesoderm from Xenopus forms a ball with the cells in the proper relative positions. But if the ectoderm is allowed to aggregate for a few hours before the other cells are added we get a mesodermal ball with a mixed ectoderm/endoderm cover: this carries on to form a structure with endoderm outside mesoderm inside and ectoderm as the meat in the sandwich.

We presume that what makes cells aggregate in this way is some kind of stickiness. We don't know if this stickiness which makes cells aggregate is very specific, like lock and key or antigen and antibody, or more general. Two cell types which just differ in unspecific stickiness would eventually segregate from each other. Embryonic cells which stay together over long periods tend to reinforce the adhesion by interdigitating their membranes, thus increasing the contact area by creating desmosome-like attachments.

Cell adhesion molecules (OVERHEAD)

Molecules on the surfaces of cells which fit on to something have to have a specific spatial and perhaps temporal arrangement of hooks or valences. Recently three classes of such molecules have been identified:-

CAMS cadherins or cell adhesion molecules which stick cells to each other

CJMS or cell junctional molecules that stick cells together whilst allowing cell-cell communication

SAMS or substrate adhesion molecules which stick cells to substrates, such as intercellular matrices (and thus facilitate cell separation)

Two of the most studied CAMS are N-CAM discovered in the nervous system and L-CAM found in the liver. Both of these occur early in development and in specific areas of time and space, so as to allow them to be candidates for epigenetic molecules.

The early chick blastoderm has both N-CAM and L-CAM. With the onset of primary embryonic induction N-CAM is preferentially expressed (i.e. present) in chordamesoderm, but not in ectoderm or epidermis which both have L-CAM. Some structures, such as eye and ear placodes have both.

Current evidence suggests that the presence of CAMS and SAMS on the surfaces of cells is a reaction to induction - the inducer is telling a cell to stick close to its buddies, or to stick to the intercellular matrix, and thus spread out. They may, of course, indirectly affect induction by changing the properties, such as speed of movement, of a group of cells.

A group of cells increasing its stickiness with time would lead to greater contact areas and cells moving towards each other. This is a formal description of condensation as seen in precartilaginous and premuscular blastemata in the limb and elsewhere in the skeleto- muscular system. We have already seen that embryonic cells from different organs and different species differ in cell adhesion. We can show the same thing happens within the limb. Chick muscle, cartilage and undifferentiated mesenchyme cells can all segregate late in development: but late is the operative word: cartilage cells cannot segregate from limb mesenchyme until after they have started to secrete matrix. Thus segregation is a very late event in cartilage and blastemata might grow by recruiting more mesenchyme cells.

This sort of aggregation behaviour is also seen in a well studied slime mould Dictyostelium. Slime moulds are the nasty greyish or reddish lumps of goo that you sometime see around taps. Their chief claim to fame is that they exist for part of the time as a slug or grex, and the rest of the time as a gang of amoeboid free living cells. They split up and get together again by changes in cell adhesion, and by shouting to each other. This shout, produced only by some cells consists of cyclic pulses of AMP.

Cells close to a signalling cell relay the signal by producing their own pulse of AMP, cells further away just move towards the signal. If just one cell initiates this signal then the result will be a circular wave form of regular wavelength (OVERHEAD). Signalling between two cells would tend to form a closed loop, but if the signal were relayed between many cells single or nested spiral wave fronts would result. Cells in blastemata do this too, although it is not known if they secrete AMP. Time lapse photography (OVERHEAD) shows that cells in condensations tend to spiral inwards. And many blastemata have this spiral appearance in sections (OVERHEAD).

The change of adhesiveness generated in a blastema has been documented. Blastemal cells are becoming specialised and so exhibit an almost universal tendency to divide more slowly. At the stage when mesoderm cells are capable of becoming either cartilage (if left in situ) or fibroblasts (if moved elsewhere) they make contact across large areas of intercellular space by means of long filopodia. In later stages, as they become more closely packed areas of cell content increase. This is either a change in the behaviour of cells which are becoming cartilaginous, or it is a necessary step in the process. If the latter we have an explanation as to why cartilage precursor cells cultured at low densities do not form cartilage.

Subsequently fibroblasts and chondroblasts become even more different: for instance both produce collagen, but cartilage produces one type, with two different protein chains (presumably derived from two different genes) and fibroblasts produce a monomer.

So the hypothesis is that a change in cell adhesiveness causes condensation i.e. a simple change in a physical property is producing a condition driven pattern. Can we verify this?

There is a chick mutant called talpid - after Talpa the mole. Talpids have many abnormalities, including broad limbs with unusual numbers of condensations (OVERHEAD) with carpals and metacarpals represented by a broad band of condensations, and those of the phalanges often consisting of aggregates of more than one digit. Ede and his co-workers showed that the mesenchyme from talpid, if cultured in suspension, was stickier than normal. More cell adhesion, bigger condensations.

So a simple change in biochemistry, leading to a simple physical change in adhesiveness can differentiate cells that will become the skeleton from cells that will become fibroblastic connective tissue. If the change is overdone we get a sticky mess with a few large condensations, formed apparently from what should be smaller ones stuck together. If the system works as it does in Dictyostelium then somewhere in the middle of a condensation are different cells, the ones that produce the signal. The difference could come from inside the cells or outside They could be a little older or younger than the others (in terms of mitotic divisions), or form a clone from a specific part of the egg, or have unique positional information or unique genetics.

The way in which cartilaginous condensations could form is only one aspect of pattern however: another problem is the relationship between one condensation and the next. What determines the overall arrangement of condensations within a limb bud, how many rows, how many columns? Well mesenchymal condensations tend to form regularly spaced patterns wherever they are. Ede found that limb mesenchyme in culture forms regularly spaced chondrogenic centres whose spacing depends on the size of the aggregate. Patou took mesenchyme from two different bird species (so that he might recognise the resulting limb) mixed them and cultured them under a cap of ectoderm. The resulting limbs (OVERHEAD) had no identifiable digits, but they did have rows and columns of elements that broaden out from the base towards the tip.

Remember Dictyostelium? The outer cells move inwards but don't signal - this would limit condensation size and hence affect spacing. So perhaps the ratio of the width of the limb bud to its dorso-ventral size does determine the number of condensations per row, as in our condition generated pattern model. But once again part of the pattern is missing - what determines the shape of the individual condensations, and how are they modified between, say, duck and hen? Back again to that niggling question - is this element of pattern formation intrinsic - from the genes - or extrinsic - a signal coming into the cell from elsewhere?


We know that in non-regulatory animals the answer to this question is from within the cell, by virtue of its lineage. If nematodes or rotifers had legs they would make them in this way. We also know that in other animals, including mammals, the mosaicism is much less pronounced or absent. But how absent is it? Totally? No because all cells acquire the ability to specialise or lose the ability to generalise at some stage.

Bonner argued that in the regulatory embryo differentiation was an orderly switching on or off of a series of subsets of genes. If a change in circumstances were imposed upon an undifferentiated region a different subset might be activated and a different tissue or organ formed. The normal result of these subsets is a flower or a foot. The decision must always be taken by one cell, and its offspring form a clone.

If we accept this it is necessary that clones exist, clones doing the same thing, and, if not necessarily derived from the same blastomere then ultimately derived from a single founding cell. We can check for clones in mammals quite easily. It is perfectly possible to stick two early mouse embryos, say at the two or four cell stage, together. The resultant four or eight cell embryo behaves normally, and is eventually born as a normal mouse.

Because the fusion is made after fertilisation any clones formed will resemble each other genetically. If we put suitable markers on each parent we can therefore see which bits derive from which parent. Significant cloning should make large patches which look like one parent, no cloning should give a random distribution of parental cells. If this is done for coat colour mutants significantly large patches of a single colour are found. Is this true elsewhere in the body?

Moore & Mintz (1972) looked at the vertebrae of chimeras from two inbred strains of mice. They found that there was evidence that the vertebrae were produced from a minimum of four clones of cells. This is reasonable, as vertebrae are made up of contributions from two adjacent somites on the left and the corresponding two on the right (OVERHEAD).

However we didn't believe this (O'Higgins et al 1986). In a more sophisticated experiment we also looked at chimeric vertebrae. First of all we demonstrated that we could tell them apart using a computerised measuring system (OVERHEAD). We used a technique called Fourier analysis which produces a series of numbers which describe the shape. Our view was of the superior surface of the bone, so we could see two of the four quarters of the vertebrae. Fourier numbers as we used them distinguish both dorsal versus ventral and left versus right differences. Two strains of mice carrying seven recessive alleles (REC) or their 7 dominant normal forms (DOM) had different shaped vertebrae, reflected in different Fourier numbers.

Then we classified vertebrae from fused embryos allowed to grow up (chimeric vertebrae) according to parental type (OVERHEAD). Quite a lot were just like one parent or the other suggesting that the mice were not really chimeras, perhaps that one fused bit had died, or the other had overgrown it. Those that did not belong to one or other parental type, the ones we assumed to be chimeric, were not like either parent and also unlike each other. Now if two clones made up the part of the vertebra we could see, and if they were from different parents the only possible arrangement is DOM on left REC on right or vice versa. The chimeras should thus cluster into two mirror image, asymmetrical groups. They didn't, and also showed no more asymmetry than parental types.

So we don't believe that vertebrae are built from such a small number of clones: but it doesn't necessarily mean that smaller clones exist - that certain groups of cells don't have specific sets of genes switched on and off in a temporal sequence - in fact they do.

Later stages of limb development - atavism and all that jazz

So we have now seen the way in which the pattern of the skeletal elements of the limb might be generated. This course is supposed to be about evolution as well as development, so I want to go on and consider how the pattern of the limb might be varied by various embryological manoeuvres to produce different adult forms.

The limb as seen in you and me is a representative of what is usually considered a fairly primitive form: the five digits seen in man and almost all primates, and in the rodents is thought of as a primitive mammalian pattern. If we look at mammals, birds, reptiles and amphibians we can find a series of variations on this pentadactyl theme: the pattern starts with one bone, the humerus or femur, then two the radius and ulna or tibia and fibula then a variable number of carpals or tarsals which seem to depend on the number of digits present. This is usually five or less, but may be more, especially in marine forms where we need a paddle. The phalangeal formula, the number of phalanges per finger also varies, being highest at 2,3,4,5,3 in reptiles and low at 2,3,3,3,3 in amphibia and mammals.

These differences are rather difficult for us to understand at the moment because we can't find intermediates between mammals and reptiles very easily, and if we can find the fossils we can't do embryology on them. So lets just stick to comparisons within classes and look at mammal versus mammal or bird versus bird. We can still find digits lost, digits apparently gained, and changes in proportion of the various elements. All these must be due to modifications in embryology. Once the basic pattern is recognised we can ask 'what is it fairly easy to change?' on the basis of the changes we see around us. Limb shape and pattern is as good an example as any of this, but remember that it applies to other things as well.

Functional diversity.

All the changes we see are the result of functional diversity. Limbs are different shapes for different purposes. Form follows function. Perhaps one of the easiest changes to understand is proportion (OVERHEAD)

You probably know already that speed in a running animal depends on length of stride and rapidity of step. That was sorted out by Gregory in 1912. The power developed by a muscle of a given mass depends on the distance through which it acts. So power and speed are opposed and we can tune an animal for one or other by changing the overall length of its limbs and their proportions, the distal elements tending to elongate for speed. The ratio of femur:tibia is often used to measure this - it is 0.6 in the elephant, 0.92 in the racehorse and 1.25 in the gazelle. The sizes of the muscles are also regulated in step with the bones: the distal limb muscles in many species are represented mainly by tendons with little muscle mass. Muscle insertions may also be moved (OVERHEAD) in order to favour power or speed.

So function may lead to anatomical changes in bone lengths. We might like to ask ourselves just how such a change might be specified within the developing embryo.

A fast moving animal will need to reduce the mass of its distal limb segments as well as making them lighter. We can see that both these desiderata can be achieved together. If you are going to have a long limb it is sensible to achieve this by using existing structures, perhaps by going up on tiptoe. If you stand on the longest toe then the others won't reach the ground and won't contribute much to locomotion. The muscles attached to them won't do anything either. It is therefore sensible to lose both muscles and bones. We can see two good examples of this within our mammalian cousins, in horses and cattle (OVERHEAD).

Although we saw that the conventional horse story is really rather dubious we can still arrange fossil and modern horses in a series which show increased specialisation with lengthening of the second toe and progressive reduction of the others. Cattle show a similar set-up but seem to have had no clear longest toe 3 and 4 seem to have been about the same length and to have been retained. In this case we have another process chipping in: metatarsals 3 and 4 are clearly acting as a strut, an extension to the tibia. As such, separate movements are not needed, and the two tend to fuse to form a cannon bone. Quite commonly in mammals the tibia and fibula do the same at the distal end.

So we must add another pair of embryological processes to our hit list. How, and at what stage can we modify limb pattern to get rid of elements and how and when can we make elements fuse together?

This page is maintained by Steve Paxton