Nurse: Martin Raff has set up two general questions that I think we should try to address in this discussion. One concerns the nature of this timer (and we could add to this, why is the timer needed). The second is this relationship between stem cells and progenitors, and what we mean by this. But I'll start with a specific question. If it is only 6—7% of the cells that undergo one or two more extra divisions, is that sufficient to account for the p27~ l~ mutant phenotype with respect to cell number?

Raff: In the population of postnatal day 7 (P7) precursor cells, there are some that will only go through one division before they stop, and others that go through up to eight cell divisions. The ones that will go through up to eight divisions are in a minority. In the p27-deficient population, the whole curve is shifted to the right, but the ones at the extreme right hand side of the curve, which we suspect are the least mature, are easiest to detect as abnormal because they go through more divisions than any normal P7 precursor cell. In normal mice, the cardiac myocytes have all dropped out of division by P1 or P2. In the p27 knockout, there are still cells dividing at P4 and you end up with many more cells, but the cells are much smaller.

Reik: Is it true that the phenotype of that knockout is not present at birth, so at birth they are normal size?

Raff: They are normal size if you weigh the mice. But if you look carefully at cell number, there is a phenotype at birth.

Leevers: What happens to the p27 levels when you shift the temperature?

Leevers: In the p27 knockout mice, is the expression of the other proteins you showed us affected?

Raff: We don't know.

Mailer: What sort of process could be going faster at a lower temperature?

Raff: It could simply be that the degradation of p27 is decreased more by low temperature than p27 synthesis.

Mailer: Have you used these new phospho-specific antibodies to Cdc2 to examine the cycles of tyrosine phosphorylation and dephosphorylation?

Raff: No, it would be interesting to do that.

Mailer: In Xenopus, raising p27 also raises tyrosine phosphorylation of Cdc2, possibly through indirect effects on Wee1.

Nasmyth: The key underlying question is what determines how big an organ or organism gets, for which this model system might be relevant. There are two hypotheses that come to mind. One is that the bigger you get, the more of something you make and you create a concentration that is somehow related to the mass of the tissue. The other is the timer. It seems to me that one of the predictions of the timer hypothesis is that in the absence of any cell death, it is not clear how the timer can be regulated. Is that correct?

Raff: There are many things that can regulate the timer. TH is one example. If you take TH away, most of the cells just keep dividing. If you add FGF2 as well as PDGF, to stimulate proliferation, the cells also just keep proliferating and don't differentiate. As it happens, at least halfthe oligodendrocytes produced in the optic nerve undergo programmed cell death, in a process that adjusts their numbers to the number and length of axons.

Nasmyth: In this particular case I was getting the impression that timers would be one way of determining final organ size. In this particular case you clearly get regulation through death.

Raff: It would surprise me if any system depended exclusively on such timers to get the final cell numbers right.

Nasmyth: Can you use it at all?

Raff: The phenotype of the p27 knockout mice suggests such timers do play a part in size control. The simplest explanation for why you get more cells in these mice is that p27 normally plays a role in taking cells out of division at the right time. Cdk inhibitor mutations seem to play similar roles in flies and worms. p27 is clearly not the only component of the stopping mechanism. If you take it out of action, the cells still stop, just not at the right time.

Nurse: I wonder whether the word 'timer' is perhaps the most useful one. You probably came up with the term 'timer' to contrast it with counting cell divisions. It is a timer that is influenced enormously by external factors, and in vivo it will be very complex, whereas generally when we think of timers we tend to think of something that is measuring time. What you really mean is that it is something that is cell autonomous which in a defined set of conditions measures time rather than cell divisions, but it is a timer that is so influenced by external factors that it is not like a little clock ticking in the cell.

Raff: What is being affected by these manipulations may not be the timing component. p27 goes up and plateaus at the same time even if you remove thyroid hormone or add FGF to the PDGF-containing culture medium. The appropriate signals are required, however, for the timer to stop the cell cycle and initiate division at the right time. Thus, there is an intrinsic timing mechanism that operates without thyroid hormone or in the presence of FGF; it is just that you can't see that it is working unless you measure something like p27 levels.

Nurse: Perhaps it would be more helpful to think of this in a negative sense, in that whatever it is doing it is not counting cell divisions.

Raff: Yes, it is very unlikely that it is counting divisions. I don't think that cell-intrinsic timers are responsible for all of timing in development. It is quite clear that they are not. But cells do change as they develop, and some of the changes seem to reflect intrinsic programmes operating within the cells.

Reik: If you consider an organ that is already fully differentiated at birth, and then it simply grows, is the timer off in that situation?

Raff: It depends what timer you are talking about. I have been talking about only part of the oligodendrocyte development process, where the precursor cell withdraws from the cell cycle and expresses one differentiation antigen. But the oligodendrocytes then continue to differentiate over many days, turning on myelin-specific genes and wrapping axons. This is a very complex process and I have no idea how these events are controlled.

Hunt: I want to ask about PDGF. It seems a little unlikely that these are really responding to PDGF in the context of the optic nerve.

Raff: The evidence is very strong that PDGF is a major mitogen for these precursor cells in vivo. PDGF is made by astrocytes in the optic nerve. If the PDGF a gene is inactivated in mice, oligodendrocyte precursors don't proliferate and very few oligodendrocytes develop.

Hunt: It seems to me that this is a system par excellence where cell—cell interactions are very important. These things are ultimately designed to wrap neurons. There have to be enough of them so that they will totally insulate the neurons. When they put them into these pure cultures, I wonder how that is replicating the real world situation.

Raff: The remarkable thing is that in the presence of the right combination of signals, the precursors seem to stop dividing and differentiate on much the same schedule as they do in vivo. Axons are essential, however, for newly formed oligodendrocytes to live in vivo.

Hunt: Do you know what the axon normally gives them?

Raff: We know that one of the signals is a neuregulin, glial growth factor; there are almost certainly others.

Nurse: You see, naive frog people think that PDGF only comes from platelets!

Edgar: I have two simple questions about the timer. First, what starts it ticking and, second, does it have any influence on cell growth?

Raff: I don't know the answer to the second question. I suspect that the timer starts ticking when the precursors first arise from multipotential CNS stem cells, but we don't know.

Edgar: Is that determined by a cell—cell interaction?

Raff: The ventralizing signals such as Sonic hedgehog that are required for motor neuron development are also required for oligodendrocyte precursors to develop from CNS stem cells, at least in the spinal cord.

Schmidt: What happens to the six cortical divisions in the p27-deficient mice? Likewise, has anyone ever done the heroic thing of taking the basal layer that gives rise to the six cortical divisions and transplanting it down one day to see whether it still has five divisions left?

Raff: I'm not aware that anyone has done it in that way, but it is an important question. Cortical neuron precursors in the mouse, for example, go through many fewer divisions than do cortical precursors in primates, which is why mice have a smaller cortex than primates. The question is why is the behaviour ofthe two types of precursors so different? Is it because intrinsic timers are set differently? Is it because the mitogens and growth factors that drive proliferation are around for longer in primates? These are questions that could be addressed without a need for new technology.

Schmidt: There is an even simpler one, which is what about hypothyroidism, which exists in all species and can be induced?

Raff: You end up with pretty normal cell numbers in hypothyroid animals, although it takes longer to get there. Thyroid hormone seems to play a coordinating role in development.

Schmidt: Do the p27-deficient mice get more cortical divisions?

Raff: I assume so, as there seem to be more cells in the cortex.

Ko^ma: Are you referring to a problem intrinsic to all cells, or something specific to these oligodenrocytes? Because in the development of an organism, the size of an organ is also controlled by the time when puberty is reached: hormones are stopping or inducing organ growth. Another type of growth regulation is tied to the organ pattern. In the case of the oligodenrocytes you are dealing with an isolated cell. Thus, is this representative of all cell types?

Raff: I think the timing mechanism may be similar in many cell lineages where precursors divide a limited number of times and then stop and terminally differentiate. The best evidence for this is that if you inactivate p27 there are more cells in every organ. I think it is unlikely that these timers play much ofa role in the fine adjustments in cell proliferation that determine patterning and organ shape.

One thing that is interesting about the p27 knockout mouse is that cell death does not bring cell numbers down to normal levels. Why not? It may be because there are more cells in multiple lineages in each organ, and they support the survival of one another. If only one cell type is increased within an organ, then cell death would presumably bring the number back down to normal.

Edgar: Has that been done with p27 to make a mosaic?

Raff: Not as far as I know.

Nasmyth: It would be nice to make a p27_ heart. If one wants to analyse the molecular nature of this phenomenon, one of the things one would like to know is what is the variability? How well does it keep time? What is the molecular mechanism behind this? How do clocks work in the absence of a mechanical system? At the moment we use atomic decay. The same sort of principle could be occurring here. What you have is cells that started off at various epigenetic states. Genes are on or genes are off. There is a huge literature showing that these epigenetic states are unstable. They decay at a certain frequency. What we may be looking at here is the stochastic decay of epigenetic states. There are ways of making this less variable. If just one gene was involved there would be a huge standard deviation, but if a combination of genes were involved this may produce quite an accurate clock.

Nurse: This is possible, but it would depend on what was the rate-limiting step. You would only get such a state if they were all decaying with rather similar dynamics.

Nasmyth: We know that when these states do decay, it is completely gene autonomous. This is the key thing.

Nurse: Even if it was one gene you could set up the system so that it would have limited variability. It would depend on the numbers of molecules involved in that circuit. You can devise limited variability in different ways. One is by having many different elements in the way that you have described, but you can also do it by having lots of molecules involved in a single regulatory circuit, which would reduce variability.

Nasmyth: Then it is not stochastic.

Nurse: The amount of variability will be determined by the number of components in that system. If you are saying it is stochastic, if it is really stochastic you need some factor in there which is limiting and decaying. This you would have to produce by having one component that is rate limiting and stochastic.

Nasmyth: My understanding is that the gene itself, whether it is on or off, is completely stochastic.

Nurse: The regulation of it is not stochastic.

Nasmyth: No, it is epigenetically controlled; it is not affected by the environment— that is the whole point.

Nurse: It will be on or off, but the dynamics of it are not stochastic.

Nasmyth: No, that is what we know about these epigenetic states: they are either on or they are off.

Nurse: That is a description of the final state: it is not a description of how you achieve that final state.

Reik: There is another way of doing it. This can be determined quite precisely. For example, in Schi^osaccharomyces pombe the grand parental DNA strands are marked epigenetically to affect mating type switching.

Nasmyth: This doesn't give you the clock.

Reik: It can, because if you switch one system on like that, you have something two cell divisions later.

Nasmyth: The point is, this doesn't depend on cell divisions.

Raff: Remember that there is an accurate timer in Xenopus embryos that doesn't depend on DNA, RNA, or protein synthesis. It is still a mystery how it works.

Novartis 237: The Cell Cycle and Development. Copyright © 2001 John Wiley & Sons Ltd Print ISBN 0-471-49662-6 elSBN 0-470-8 4 6 6 6-6

0 0

Post a comment