Control of Cell Growth in Animal Development • iBiology

Part 1: Regulation of Cell Size

00:00:07.00 So I'm Martin Raff from the Medical Research Council Laboratory
00:00:11.01 for Molecular and Cell Biology at the University College London in the UK.
00:00:15.00 And the subject I want to talk about today is growth control or size control.
00:00:21.03 How is it that the size of an animal, or one of its organs,
00:00:25.00 is controlled. Why is it, for example, that we grow
00:00:27.00 to be so much larger than a mouse? The answer is that we haven't really a clue
00:00:33.00 why we grow to be larger than a mouse, and the reason for that is largely that this problem
00:00:40.28 has not captured the imagination of developmental biologists
00:00:43.00 the way some other aspects of development have. So size control has been relatively neglected
00:00:48.00 and we don't understand much about it.
00:00:51.00 But what we do understand is that the size of an animal
00:00:54.01 and one of its organs depends on two things
00:00:58.00 the total cell mass, which in turn depends on cell size and cell number,
00:01:02.00 and the extracellular materials, largely proteins in the form of extracellular matrix
00:01:08.00 that can be cuticle, or it can be a shell, or it can be bones and so on.
00:01:14.00 In the case of mammals it is largely cell numbers that matter most.
00:01:19.00 So we are about 3000 times larger than a mouse,
00:01:24.02 and we have about 3000 times more cells than a mouse.
00:01:28.00 So it is largely cell numbers that one needs to be concerned about in size control in a mammal.
00:01:34.00 Cell number in turn depends on two processes, one is cell division,
00:01:40.01 or cell proliferation, and the other is cell death.
00:01:43.00 Both of these contribute to determining the size of an animal or one of its organs.
00:01:50.00 Now, in thinking about the control of the size of an organ, there is two kinds of processes.
00:01:58.00 There are those that operate in intrinsically in the organ to help determine its size,
00:02:02.01 and then there is this systemic control, endocrine controls for example
00:02:06.00 that help control size. Don Metcalf working in Australia in the 1960's
00:02:12.02 did a series of very interesting experiments
00:02:15.00 that gives you some sense of internal versus extrinsic control.
00:02:21.00 In the first set of experiments, he took the thymus glands from newborn mice
00:02:27.01 and put multiple thymus glands into a newborn mouse
00:02:31.03 under the skin, under the kidney capsule, let the mouse grow up
00:02:35.00 and then took the thymuses out and asked what size they had grown to.
00:02:40.00 And the result was clear cut. They grew to be the size of a normal thymus.
00:02:44.01 It didn't matter how many thymuses
00:02:45.00 were growing in the mouse, each one grew to be its normal size.
00:02:49.00 So it seems that the thymus knew how big it should grow to
00:02:53.00 and did it independently of what was going on elsewhere in the animal.
00:02:57.00 On the other hand, when Metcalf looked at the spleen,
00:03:01.01 he got the opposite result. When he put multiple
00:03:03.00 spleens into a newborn mouse and let the mouse grow up
00:03:07.02 and then take the spleens out and measure their mass,
00:03:09.00 the total mass of all those spleens equaled the mass of a normal spleen in a normal mouse.
00:03:16.00 It looks as though the control was largely outside the spleen.
00:03:21.01 And for most organs its a combination of intrinsic control and systemic control or extrinsic control.
00:03:28.00 The systemic controls on growth that are best understood are those that operate
00:03:36.02 in the growth hormone insulin-like growth factor one, the IGF-1 pathway.
00:03:42.00 The way that the pathway works is the hypothalamus produces growth hormone stimulating
00:03:51.01 or releasing hormone. That acts on the anterior pituitary
00:03:55.00 causing it to secrete growth hormone into the blood. The growth hormone then
00:03:59.00 stimulates the liver and other cells in other tissues
00:04:02.03 to produce insulin-like growth factor 1, IGF-1, which in turn induces cells to grow,
00:04:09.00 promotes their survival, and in that way promotes the grow of tissues, organs, and the animal.
00:04:14.03 In mammals such as a mouse or us, this system accounts for about 2/3 of our growth.
00:04:20.00 If you would activate growth hormone, or IGF-1, and the receptors,
00:04:24.01 then a mouse would grow to about 1/3 the normal size.
00:04:28.00 This system accounts largely for variations in size within a species.
00:04:33.03 But it's really not that problem that interests me, it's why are we so much larger than the mouse?
00:04:41.00 And no matter how much growth hormone you give a mouse, or how much IGF-1 you give a mouse,
00:04:45.03 you can increase it's size perhaps to double its size, but you cannot even make it the size of a rat
00:04:51.00 let alone the size of a human. So we still donâ€™t' have any idea why we grow to be so much larger than a mouse.
00:04:59.00 I want to talk about two cell systems that we have used to assess growth control.
00:05:07.01 Both of them are glial cells, supporting cells in the nervous system.
00:05:12.00 One of those is the Schwann cell that wraps around axons
00:05:16.03 in the peripheral nervous system to myelinate them.
00:05:19.00 The other is the oligodendrocyte that myelinates axons in the central nervous system.
00:05:26.00 It's really the cells that give rise to the differentiated myelinating cells that I'm going to talk about.
00:05:33.00 So first I'm going to talk about cell size control and we'll use Schwann cells to address that problem,
00:05:42.00 and then I will turn to cell proliferation control and I will turn to
00:05:47.03 the oligodendrocyte precursors that give rise to oligodendrocytes.
00:05:52.00 Let's begin then by looking at cell size control. Here's the nature of the problem.
00:06:01.01 It's known in all organisms from bacteria to humans that the size of a cell
00:06:09.00 is roughly proportional to the ploidy, the number of chromosomes.
00:06:13.00 So a haploid cell in the same organism is generally smaller than a diploid cell
00:06:17.03 and a tetraploid cell is larger than a diploid cell.
00:06:21.00 But if you look at a single organism such as a mouse or a human,
00:06:25.01 most of the cells are diploid, and yet they vary enormously in size.
00:06:30.01 For example, this nerve cell in the retina of the monkey is very much larger
00:06:34.02 than the lymphocyte in the same animal, yet they are both diploid.
00:06:41.00 We have no idea why a cell such as a nerve cell grows to be so much larger than the lymphocyte.
00:06:48.00 It's a very important question, and there are very few people who have been studying this problem,
00:06:54.02 which is why we know so little about it.
00:06:57.00 But the problem that I want to talk about is a different one,
00:07:00.02 It's when a proliferating population of cells divide, generally they double their mass,
00:07:07.01 double the number of organelles, and then they divide in this way
00:07:12.01 to maintain the size of this typical proliferating cell.
00:07:15.00 So the question that one wants to know the answer to is how does a
00:07:21.01 proliferating cell coordinate its growth so that when it divides it maintains the appropriate size.
00:07:30.00 Now largely from studies in yeast, there are some ideas of how cells coordinate their growth,
00:07:36.02 but I should say that there is much less attention to cell growth than their has been to cell division itself.
00:07:44.00 Yet cell growth is as fundamental to life. Any organism, bacteria, plant, animal, it doesn't matter,
00:07:53.00 could not grow unless cells grow. If all that happened was cells divide
00:07:59.00 the cells would get smaller and smaller and smaller. So cell growth is as fundamental
00:08:04.00 to biology as cell division. Yet for each lab that studies cell growth there is probably
00:08:11.00 hundreds or thousands of labs that study cell division
00:08:13.00 and it's actually easier to study cell growth, and there is no rational reason why that should be so.
00:08:19.00 One of the problems is terminology. For example, the term "cell growth" is used by most biologists
00:08:29.02 to mean cell proliferation. So there isn't even a term that we have to mean cell enlargement and increase in cell mass.
00:08:37.00 There's a perfectly good phrase for cell proliferation: cell proliferation. So one should really use cell growth
00:08:44.00 to mean precisely that: cell enlargement, an increase in cell mass. Most people do not, but I will.
00:08:51.00 The other term that is confusing is "growth factor", which is used to describe extracellular signal molecules,
00:09:01.00 usually proteins, that stimulate cells to grow, stimulate cells to divide,
00:09:06.00 stimulate cells to survive, stimulate cells to differentiate, and often very little attempt
00:09:11.02 is made to distinguish whether a growth factor is stimulating cell cycle progression to division or cell growth.
00:09:19.00 So these are terminology problems within the field that has held it back and continues to do so.
00:09:25.00 Largely from studies in yeast, it is thought that one way that proliferating cells coordinate their growth
00:09:35.00 and division to maintain an appropriate size, is that cell growth limits progress through the cell cycle.
00:09:42.00 An extreme form of that is the notion that cells may have cell size checkpoints in the cell cycle control system.
00:09:52.00 For example, there is thought to be, at least in some yeast, a cell size checkpoint in G1,
00:09:59.01 where the cell cycle control system can pause
00:10:03.00 to assess the cell size and make sure it's large enough before entering S phase and replicating the DNA.
00:10:09.00 And in some yeast and in some other cells it is thought that there may be another cell size checkpoint in G2,
00:10:16.02 where the cell pauses to make sure it is large enough before it goes into mitosis and divides.
00:10:24.00 The studies I want to talk about were done by a very talented graduate student, Ian Conlon,
00:10:30.03 who has now left science and works for the Civil Service in Britain studying cell traffic control in London.
00:10:38.00 He studied the cell growth control of Schwann cells.
00:10:44.00 These were studies done many years ago by two post-docs, Kay Fields and Jeremy Brockes,
00:10:49.01 who showed that you can use antibodies to distinguish cells that are isolated from
00:10:55.01 the newborn rat optic sciatic nerve, peripheral nerve, and you find two main types of cells in those cultures.
00:11:04.02 You see these red Schwann cells and the green fibroblasts and if are labeled with different antibodies,
00:11:11.00 you can use these antibodies to not only distinguish the different cells, but also separate them.
00:11:17.02 In that way you can purify the Schwann cells, and Ian used purified Schwann cells from the rat newborn sciatic nerve.
00:11:26.00 In order to study cell growth independently of cell cycle progression and cell division,
00:11:35.00 he blocked cell cycle progression in S phase by using aphidicolin,
00:11:41.00 a drug that blocks DNA polymerase alpha that's required for DNA replication.
00:11:47.00 Therefore the cells are arrested in the cell cycle in S phase,
00:11:52.00 and he studies cell size by measuring the volume of the cell in a Coulter counter,
00:11:59.00 which measures cell volume by the displacement of fluid, so it doesn't matter what the shape of the cell is.
00:12:05.00 In this experiment he shows that cells proliferating in serum continue to divide and there is a
00:12:12.03 spread of cell volumes as you can see here. If you block the cells in aphidicolin after one day
00:12:20.02 the cells continue to enlarge even though they are no longer going through the cell cycle
00:12:26.00 and so the average volume of the cell has increased dramatically.
00:12:31.00 And so that is his assay: aphidicolin arrested cells to measure the growth of the cells.
00:12:38.00 To know that cell volume is of interest, because what you are actually interested in is cell mass,
00:12:47.02 to know that the volume is reflecting cell mass,
00:12:50.01 he collaborated with Graham Dunn at Kings College in London
00:12:54.00 to study the dry mass of Schwann cells growing under these same conditions.
00:12:59.01 He uses microinterferometry to do that. And you can see that the result
00:13:05.00 is the same: cells proliferating in serum without aphidicolin have an average dry cell mass
00:13:11.01 that is relatively low compared to the cells that have been arrested with aphidicolin for 24 hours.
00:13:17.00 So he's confident that he gets the same result measuring cell mass as he gets for measuring
00:13:22.02 Coulter counter volume and he continues to use volume because it is much more convenient.
00:13:28.00 Now these Schwann cells that are arrested in aphidicolin, they will continue to grow for days.
00:13:36.00 They get bigger and bigger and bigger and Ian never found an upper limit to the size of a Schwann cell.
00:13:43.00 Now there must be an upper limit, but I'm not aware of anyone who knows
00:13:47.02 or has asked the question: what is the upper limit size for any cell type within an organism.
00:13:55.00 And we still don't know if there is an upper limit, there must be for the Schwann cell,
00:13:58.02 and when you got to that upper limit what it is that's limiting growth at that point.
00:14:04.00 So now he asks, does cell growth require signals in the extracellular space to simulate the growth.
00:14:14.03 By and large cell divisions, cell survival, cell migration,
00:14:17.02 most activities of cells depend on being signaled by other cells
00:14:23.00 that's the way animals work, so that cells behave in a way that is appropriate for the entire organism.
00:14:29.00 And here he shows that it is indeed the case. So here the population of cells in heterogeneous volumes
00:14:35.01 growing in serum, you block them in aphidicolin, they continue to grow and so the cells are very much larger.
00:14:42.00 But if you block in aphidicolin and there is no serum there and no growth stimulant
00:14:48.00 in the extracellular fluid then the cells do not grow.
00:14:51.00 So growth, like cell division, cell migration, depends on signals from other cells.
00:14:57.00 Now he's in a position to look at known growth factors that have been shown
00:15:03.02 to be important for stimulating Schwann cell proliferation both in culture and in the animal.
00:15:10.00 One of those is insulin-like growth factor one, IGF-1 that I mentioned earlier,
00:15:14.00 and the other is the neuregulin glial growth factor, GGF. So here he's used the aphidicolin trick again,
00:15:23.01 blocks the cells with aphidicolin, and the only signaling molecule added in this case is IGF-1,
00:15:32.01 and that indeed in this 24 hour period in the aphidicolin blocked cells induces the cells to grow, to get larger.
00:15:39.00 Whereas glial growth factor, the neuregulin at subsaturating and saturating concentrations
00:15:46.01 has no effect on cell growth. Indeed, these, although they are called growth factors,
00:15:52.01 are not really growth factors for Schwann cells, whereas IGF-1 is.
00:15:58.00 Moreover, if you look at cells that have been stimulated to grow by IGF-1,
00:16:03.01 adding glial growth factor even at saturating concentration doesn't do anything, doesn't increase the growth.
00:16:11.00 So IGF-1 is a growth factor for Schwann cells, glial growth factor is not.
00:16:17.02 But I told you that both of these factors have been shown to promote the proliferation
00:16:21.00 of Schwann cells, both in culture and in the animal. So it turns out that glial growth factor is a mitogen,
00:16:29.02 and I'm using that term for an extracellular signal that simulates cells
00:16:34.00 to go through the cell cycle and in particular go through G1 into S phase.
00:16:40.00 Here he is using an assay incorporation of BrdU, bromodeoxyuridine, into DNA as a measure of the cells
00:16:49.01 moving through G1 into the S phase. And you can see that IGF-1 is pathetic at inducing cells
00:16:57.01 to move into the S phase. It is a weak mitogen on its own, hardly any activity,
00:17:02.03 whereas glial growth factor on its own is very potent
00:17:05.02 in inducing cells to progress through G1 into the S phase.
00:17:09.00 Surprisingly, when you add IGF-1 and GGF together, now you get a much bigger response,
00:17:16.01 so that IGF-1 on its own is weak, but it can synergize with GGF
00:17:21.00 to drive the cell through the cell cycle. This discrepancy between the activity of a growth factor on cell growth,
00:17:30.03 cell enlargement, and cell cycle progression has been shown by others in other cell systems.
00:17:36.01 This is not going to be confined to Schwann cells.
00:17:39.04 So Zetterberg studying mouse 3T3 cells showed IGF-1 stimulates growth,
00:17:45.00 whereas epidermal growth factor EGF stimulates cell cycle progression. Delue et. al. showed using
00:17:50.01 dog thyrocytes, IGF-1 stimulates growth, whereas thyroid stimulating hormone, TSH,
00:17:56.02 promotes cell cycle progression. So it looks that there will be a number of extracellular signals
00:18:03.00 that have different effects on growth and cell cycle progression.
00:18:07.00 One thing you would expect of a mitogen such as GGF, is that it would stimulate the production of
00:18:15.01 a G1 cyclin that is required to activate the CDKs,
00:18:19.02 the cyclin-dependent kinases that operate in G1 of the cell cycle.
00:18:24.00 Here Ian has looked using western blotting at two cyclin Ds, cyclin D1 and cyclin D2.
00:18:31.00 As you'd expect, GGF, which is a potent mitogen, stimulates the increase, the synthesis,
00:18:39.01 of cyclin D1 and cyclin D2 proteins in this 4 hour assay in GGF. Whereas IGF-1, which is a potent growth factor
00:18:48.00 but a weak mitogen has very little effect on cyclin D1. You can't see any stimulation
00:18:54.03 where there is a weak simulation of cyclin D2. Interestingly when you add IGF-1 and GGF
00:19:03.01 you don't see any greater stimulation of cyclin D1 and D2 proteins compared to GGF alone.
00:19:11.00 So this is not the location of the synergy. We don't understand, yet,
00:19:16.00 how IGF-1 synergizes with GGF to promote cell cycle progression. Now you might ask an interesting question,
00:19:27.13 why should glial growth factor and IGF-1 both of which bind to the same class of receptors,
00:19:34.02 receptor tyrosine kinases, which activate the same types of intracellular signaling molecules,
00:19:40.01 why should be they have such different effects on the Schwann cell,
00:19:45.00 one inducing growth, one stimulating cell cycle progression. Usually when you see that discrepancy on
00:19:50.02 receptor tyrosine kinases, it has to do with the kinetics of the responses inside the cell along signaling pathways.
00:19:59.27 So Ian looked at two signaling pathways. He looked at the activation of the MAP kinases,
00:20:07.02 ERK1 and ERK2 using western blotting again, using antibodies that recognize
00:20:13.01 the phosphorylated activated forms of ERK1 and ERK2.
00:20:17.00 And then looked at signaling along the PI3 kinase Akt pathway by looking at phosphorylated and activated Akt.
00:20:27.00 And what he saw is that with GGF, which is a potent mitogen, there is sustained activation along
00:20:36.01 the MAP kinase pathway as seen here. For the full 24 hours in GGF you see activation of these two MAP kinases.
00:20:44.00 Whereas in IGF-1, you see a transient activation for only one or six hours and then it is gone.
00:20:54.00 For IGF-1 that stimulates cell growth and is a poor mitogen on its own, you see the opposite. You see a sustained
00:21:02.02 activation of the Akt kinase whereas you see a very transient one hour and then off
00:21:10.01 activation of Akt with GGF. So sustained activation along the PI3 kinase Akt pathways stimulates
00:21:20.01 growth in this situation, whereas sustained activation along the Ras MAP kinase pathway
00:21:25.01 seems to stimulate cell cycle progression, and that's been true in
00:21:28.03 other cell systems although not all cell systems. So now Ian is in a position to do the crucial experiment
00:21:37.02 that he was aiming at all along, which is what would happen if you kept cell growth constant
00:21:44.01 by keeping a constant concentration of IGF-1 at a saturating concentration, 100ng/ml.
00:21:51.00 So now growth is going to be constant, and now you vary the concentration of GGF the mitogen,
00:21:58.00 which has no effect on growth and ask what happens to cells when you increase the concentration
00:22:04.03 of GGF keeping cell growth constant. As you might expect in the high concentration in blue of GGF,
00:22:13.02 the cells go through the cell cycle faster even though they are growing at the same rate,
00:22:17.02 they go through the cycle faster. So cells accumulate faster at 48 and 72 hours
00:22:23.02 there are more cells in high GGF than there are in low GGF in red. Well if cells are going through the cycle faster
00:22:33.02 but they are growing at the same rate in high GGF compared to low GGF,
00:22:39.00 then the cells having less time to grow should be smaller. And indeed that is exactly what Ian sees.
00:22:45.03 In high GGF, at the saturating concentration, the cells are smaller on average than those cells proliferating
00:22:53.03 in low GGF, and thatâ€™s true at every phase of the cell cycle. If you look at cells that are rounding up to undergo mitosis,
00:23:01.02 you can see in high GGF the mitotic cells are smaller than in low GGF.
00:23:06.00 That is quantified in this graph here. If you do the experiment of labeling the cells with BrdU to label the S-phase cells
00:23:16.03 and then separate the S-phase cells in a fluorescence activated cell sorter and ask what about cells
00:23:23.03 that have been proliferating in high GGF compared to low GGF, you find again the S-phase cells
00:23:30.03 are smaller in the high GGF cells. So at all phases of the cell cycle, cells growing in high GGF and
00:23:40.01 proliferating in high GGF are in fact smaller as you'd expect, because they have less time to grow.
00:23:48.00 So the conclusion then from these studies of Ian's on Schwann cells is that the size of a Schwann cell, the time of division,
00:23:59.02 depends on how fast the cells are growing and how fast they are going into the cycle. They are going through
00:24:06.09 the cycle slowly and have more to grow, the cells will be bigger at division if they are going through faster,
00:24:11.10 they'll have less time to grow and they will be smaller. And the rates at which the cells go
00:24:15.02 through the cycle and the rate at which the cells grow depend on
00:24:20.01 extracellular signals and their concentrations.
00:24:22.00 Some of these signals stimulate growth primarily, that is enlargement,
00:24:26.03 some stimulate cell cycle progression primarily, and some stimulate both.
00:24:32.00 So in these studies, Ian saw no evidence for cell size checkpoints, but none of these experiments
00:24:42.01 exclude the possibility that there are cell size checkpoints operating in Schwann cells.
00:24:47.00 Now one reason for conceptually thinking there must be cell size checkpoints is that in a
00:24:56.00 population of animal cells or yeast cells proliferating in culture, there is a distribution of sizes
00:25:03.00 and that distribution remains constant if the conditions in the culture is constant
00:25:07.00 over weeks, months, and years as the cells proliferate. But if big cells grow faster than little cells,
00:25:14.01 and intuitively thatâ€™s what you would think because a big cell would have more ribosomes and so on,
00:25:20.00 then you would expect in a population of a distribution of cell sizes
00:25:26.00 the big cells growing faster than the little cells would get bigger and bigger
00:25:29.02 and separate from the little cells, and you would see this size distribution spreading over time and you do not.
00:25:37.00 That would argue cells must have some kind of cell size checkpoint. So the question then becomes, do big cells
00:25:46.06 grow faster than little cells and the reason for thinking they do other than intuitively is a very important experiment
00:25:53.00 done by Paul Nurse and Kim Nasmyth years ago when they were post-docs in Scotland.
00:25:57.00 They studied a mutant fission yeast that is mutant and blocked in S-phase and not surprisingly
00:26:07.03 therefore as these cells proliferate cell number remains constant because
00:26:11.01 they are blocked in S-phase, DNA levels stay constant because they are blocked in S-phase,
00:26:15.02 but if they look at the amount of protein per cell,
00:26:19.00 it gets bigger and bigger and bigger, and big cells, these cells that are growing and getting bigger and bigger and bigger
00:26:27.00 are adding more protein per cell as they get bigger. Note this is an exponential Y-axis and
00:26:35.03 this looks like an exponential increase in the growth rate. Whether it is or not is uncertain, but what is certain is that
00:26:44.03 big yeast cells are growing faster than little yeast cells when they are blocked in S-phase by a mutation.
00:26:52.00 So Ian studied Schwann cells in this same sort of way to ask: do big cells grow faster than little cells?
00:27:01.00 And the surprising result is the answer is no. Unlike yeast cells, large Schwann cells and
00:27:07.09 small Schwann cells seem to grow at the same rate.
00:27:09.00 So here are cells that have been in aphidicolin now for 9 days,
00:27:13.02 so these cells are very much larger in volume as shown on the Y-axis here,
00:27:20.01 than these cells. But if you look at the amount of volume they are adding each day,
00:27:25.02 it's the same, whether the cell is large or the cell is small. So the cell must care to keep its growth rate,
00:27:34.02 the amount of volume it's adding per day, independent of how big the cell is. So one explanation for Schwann cells
00:27:44.02 growing at a linear rate independent of their size would be that maybe Schwann cells have an intrinsic growth rate
00:27:50.00 and it doesn't matter what their size is, that is their growth rate. Well it turns out that it doesn't work like that
00:27:57.00 because cells in 1% serum are growing at a particular rate but if you stimulate growth more with 3% fetal calf serum
00:28:04.01 the slope increases, it remains linear, big cells grow the same rate as little cells,
00:28:09.01 but the rate of growth is greater. At 10%, the rate is even greater and Ian's gone up to 40 or 50%
00:28:15.01 and the growth rate is even greater. So the growth rate is variable depending on the stimulus
00:28:22.00 but in each case, big cells and little cells are growing at the same rate. What that means is if big cells and
00:28:30.05 little cells grow at the same rate, and if they go through the cell cycle at the same rate,
00:28:34.00 then you don't need a cell size checkpoint to explain why the distribution of sizes remains fixed as cells proliferate.
00:28:42.00 But a question would be: why is it that yeast seem so different from these Schwann cells? Because there,
00:28:50.11 when you block in S-phase with a mutant, you see big cells growing at a much greater rate than the little cells.
00:28:57.00 So one possibility, because we are used to thinking that these basic phenomenon like cell division and cell growth are conserved
00:29:06.02 in their mechanisms from yeast through humans in evolution. So it was surprisingly to find this difference.
00:29:14.00 One possibility is the assays are different, so Ian is measuring growth by looking at the increase in cell volume,
00:29:21.02 and Nasmyth and Nurse were looking at proteins per cell. Another possibility is the use of aphidicolin
00:29:29.03 so Ian went back and did these experiments again now measuring, instead of volume increases. looking at protein per cell.
00:29:38.00 He found the same thing as when he measured with volume, that here cells growing in 3% grow at a rate
00:29:45.02 that is slower than when they are in 10%, but still the big cells and the little cells are growing
00:29:52.01 at the same rate, it's linear. So it seems very unlikely that it's that assay.
00:29:57.00 Another possibility though was aphidicolin. You always worry when you use drugs
00:30:01.01 that it's doing something you don't know, and could be disturbing and confounding the analysis.
00:30:06.00 So we looked through the literature to try to find an example where people looked in vivo
00:30:11.01 at the growth rate of mammalian cells, and it was surprising how hard it was to find, but we found one example of a paper
00:30:19.02 published but Hutson and Mortimore in 1982 where they used a model that has been used widely,
00:30:25.16 that if you starve the mouse or a rat the liver rapidly shrinks and over a two day period
00:30:32.01 it drops dramatically in size and that's because the hepatocytes,
00:30:36.03 the liver cells, shrink in size, not because there is cell death.
00:30:40.00 Now after 2 days of starvation, you refeed these starved animals and very rapidly
00:30:48.01 the liver regains its size and in fact it overshoots and becomes larger than normal.
00:30:54.00 And all of that growth is because the liver cells get bigger and bigger and bigger when you refeed them.
00:31:01.00 Notice this is a linear scale and this growth is linear. The big cells are growing at the same rate
00:31:09.02 as the little cells, so there is no aphidicolin these are liver cells not Schwann cells, and they are in vivo,
00:31:14.28 not in a culture system. So my suspicion is that for mammalian cells at least, cell growth is independent
00:31:22.24 of size at least for these types of cells. Now as I said a moment ago, if big cells and little cells grow
00:31:32.27 at the same rate and they go through the cycle at the same rate, then you donâ€™t actually need
00:31:37.01 a cell size checkpoint to maintain the size distribution within a dividing population.
00:31:43.00 Robert Brooks pointed this out a number of years ago in 1981 where he pointed out if you just plot a model
00:31:52.00 take a cell that is 10 units in size, compared to a cell that is at 1 unit in size,
00:31:59.02 and suppose they are adding 5.5 units before they divide,
00:32:04.02 then the big cell at 10 will become 15.5 units and then divide to give you something a little over about 8 units
00:32:13.00 whereas the small cell at 1 unit adds 5.5, gets to 6 or 7,
00:32:17.01 and then divides into about 3 so you can see that the little cell is more than doubling its size
00:32:23.00 in a single cycle, whereas the large cell is not even doubling its size, and eventually with
00:32:29.03 more and more divisions, these cells will reach a common cell size over a number of days and a number of divisions
00:32:37.00 Even if there is no cell size checkpoint, the cells will hunt to a common cell volume.
00:32:45.00 Another reason for thinking that cells might have a cell size checkpoint in yeast is that
00:32:55.03 when you switch yeast from a rich nutrient environment to a less rich nutrient environment,
00:33:03.02 they very quickly change their growth rate and divide at a new cell size appropriate for the less good environment.
00:33:13.00 And if you do the opposite, growth them in a nutrient poor medium and then switch them to a rich medium,
00:33:19.02 they very quickly within one division start dividing at a new larger cell size.
00:33:24.00 So Ian did this experiment in Schwann cells. If cells are growing in no serum,
00:33:30.01 just with IGF-1 and with GGF, they will divide indefinitely in this culture system. But they divide at a relatively small size
00:33:39.00 because they are growing at a slower rate than when cells are dividing in such cultures in 3% serum.
00:33:46.00 In both cases they will continue to divide but the cells will be larger in the higher concentration of serum,
00:33:54.01 because they grow more in a single cycle than they do when they are not in serum.
00:33:58.00 So now he switches them from no serum conditions to 3% serum,
00:34:03.02 and unlike yeast cells which would switch to the new cell size immediately within a division,
00:34:09.02 these cells gradually over 4, 5, 6 days, and many cell cycles,
00:34:16.00 gradually come to the new cell size appropriate for the 3% fetal calf serum. So if they do have a cell size
00:34:23.15 checkpoint, it doesn't operate in the way that the yeast cell size checkpoint is thought to operate.
00:34:32.00 So we would argue that it is unlikely that they have cell size checkpoints. So then the question becomes:
00:34:37.02 why are yeast cells behaving differently in this respect in terms of cell size growth?
00:34:43.24 So growth control compared to the Schwann cells. Well one reason is technical
00:34:50.20 that the yeast experiments are done where the cells are growing in optimal nutrient
00:34:57.14 and yeast cells will divide and grow as fast as they can, they care only about nutrients they don't care about
00:35:02.27 other cells, and if you are in optimal nutrients, then the thing limiting your growth rate is inside the cell,
00:35:10.01 we don't know what it is, but it could be ribosomes, it could be genes you don't know.
00:35:13.00 In the Schwann cell situation we know it is not anything inside the cell because as you add higher and higher
00:35:20.14 concentrations of serum, which contains growth signals, growth factors, then the cells grow at a faster and faster rate.
00:35:29.00 So the control here is extracellular rather than intracellular,
00:35:32.02 and perhaps that is the fundamental different here. Of course yeast cells
00:35:38.02 are unicellular organisms and they grow and divide as fast as they can, they don't care about other cells,
00:35:44.02 but animal cells don't work that way. It would be a disaster if they worked that way.
00:35:48.01 In fact, they only grow and they only divide when other cells signal them to do so for the good of the organism as a whole,
00:35:56.01 so it isn't perhaps surprising that they behave so differently in this way.
00:36:01.00 Now what does it mean for a cell to actually be growing and increasing its protein content over time.
00:36:10.00 What it means is that it's making more protein than it's degrading and secreting.
00:36:16.00 Otherwise, it wouldn't be growing. So one possibility is for Schwann cells growing independent of size,
00:36:25.02 the big cells grow at the same rate as little cells, is that they make protein at the same rate
00:36:30.01 independent of their size. Well it turns out that's not right.
00:36:33.01 So Ian looked using a 2 hour pulse of 35S-methionine to look at protein synthesis rate
00:36:42.00 and when you look at cells that have been in aphidicolin for 1 day,
00:36:45.01 which are small compared to cells that have been in aphidicolin growing for 2 days or even 3 days
00:36:50.25 the amount of protein made in this two hour pulse is very much greater the bigger the cell.
00:36:57.00 So the cells are making protein at a greater rate if they are larger than do smaller cells.
00:37:05.00 But if big cells and little Schwann cells are growing at the same rate, they are adding the same amount of protein per day
00:37:14.00 it must be that they are degrading protein at a faster rate too, when they are larger.
00:37:20.00 And indeed that is true, so Ian pulses them with the 35S-methionine and chases it for a number of hours,
00:37:27.01 6.5 hours here, and if you look at the small cells that have been in aphidicolin for 2 days,
00:37:33.03 they are degrading the protein the loss is slower than in the large cells.
00:37:40.11 So the large cells are making protein faster, and they are degrading protein faster.
00:37:45.00 So how do cells control their rate of protein synthesis and their rate of protein degradation so that big cells
00:37:54.00 and little cells grow at the same rate. It is a remarkable thing and
00:37:56.02 there must be communication between the synthetic apparatus and the degradation apparatus.
00:38:03.00 So we know very little about this, but here's a bit that is known and
00:38:08.01 this is an experiment that was done a number of years ago by Franklin and Johnson.
00:38:12.03 They looked at sympathetic neurons and I should say these are the only cells I'm aware of in a mammal,
00:38:19.02 where we know something about what determines their size.
00:38:23.00 Sympathetic neurons depend on signals that they get from the target neurons
00:38:28.01 that they innervate and that signal is nerve growth factor, NGF.
00:38:33.00 In an adult rodent, if you give sympathetic neurons more NGF they become bigger, everything becomes bigger,
00:38:40.02 the axon, the cell body, the dendrites. If you give them less NGF by injecting antibodies against NGF
00:38:46.01 to neutralize it, the cells get smaller. So the cell is simply reading out how much NGF
00:38:51.02 it gets from its target cells to determine its size. Now the experiment that
00:38:57.01 Johnson and Franklin did was they treated sympathetic neurons in culture with a
00:39:02.03 protein synthetic inhibitor, cycloheximide. When they did that in the presence of NGF,
00:39:10.01 the cells maintain their normal size for up to a week or ten days. So that means you inhibit protein synthesis,
00:39:19.00 the cell can detect that and immediately shuts off degradation and therefore can maintain its normal size.
00:39:25.03 But now if you do the same experiment in the absence of NGF,
00:39:30.01 now when you inhibit protein synthesis, degradation carries right on and the cell shrinks down to hardly being
00:39:37.17 able to be seen again. Now you remove the cycloheximide or you add back the NGF and this cell will regain its normal size.
00:39:45.12 So this communication between protein synthetic machinery and protein degradation machinery,
00:39:51.00 this coupling in some way depends on extracellular signals, in this case NGF.
00:39:58.00 And finally, I just want to say that there has been great progress now in the intracellular signaling pathways
00:40:04.00 that stimulate cell growth and a key player here is this kinase mTOR, target of rapamycin.
00:40:13.00 So this kinase is activated in yeast only by nutrients, but in animal cells
00:40:20.02 its activated both by nutrients and by extracellular growth factors.
00:40:27.00 And it stimulates protein synthesis and it inhibits protein degradation. In that way, it stimulates cell growth
00:40:35.19 so here is one intracellular protein kinase that helps to coordinate protein synthesis and degradation,
00:40:45.11 and it turns up protein synthesis and turns down protein degradation to promote cell growth.

Part 2: Cell Number Control

00:00:03.09 So we've talked about control of cell size and cell growth, and using Schwann cells,
00:00:10.17 the myelinating glial cells of the peripheral nervous system, to study the problem,
00:00:14.27 and now I'm going to switch gears and talk about the other aspect of growth control,
00:00:20.19 which is cell number control. And I'm going to switch from talking about the
00:00:26.19 myelinating glial cell of the peripheral nervous system, the Schwann cells, and talk about
00:00:30.14 the myelinating glial cells of the central nervous system, the oligodendrocytes,
00:00:34.05 and in particular, the precursor cells that give rise to them.
00:00:39.01 As I said, cell number control is much more important than cell size control in
00:00:47.14 determining the size of an organ or an animal, at least when dealing with mammals, as I said,
00:00:54.01 humans are about 3,000 times the size of a mouse and they have about 3,000 times more cells.
00:00:59.02 So what one really needs to understand, is how do you control cell number, and
00:01:04.14 there is two processes that are important here, one is cell division, which creates more cells,
00:01:10.00 and the other is cell death, that gets rid of cells.
00:01:13.02 I'm going to start by talking about cell death, but first let me introduce the oligodendrocyte and its precursors.
00:01:24.01 So oligodendrocytes make myelin in the central nervous system and they derive from precursor cells,
00:01:31.17 appropriately called oligodendrocyte precursor cells, or OPCs,
00:01:37.00 and the oligodendrocyte precursors divide rapidly then they stop dividing
00:01:42.00 and differentiate into oligodendrocytes which then myelinate axons.
00:01:47.02 We study these cells in the optic nerve, and we do that because the optic nerve is
00:01:53.22 one of the simplest bits of the mammalian central nervous system, which is enormously complex,
00:01:58.05 and it's simple because it doesn't contain any nerve cells.
00:02:02.01 It contains the axons shown here in blue from the retinal ganglion cells
00:02:07.23 that project from the eye back to the brain carrying visual information.
00:02:12.01 But in the nerve itself there are no nerve cells but there are glial or supporting cells:
00:02:17.19 there are astrocytes and there are oligodendrocytes.
00:02:20.01 The astrocytes have many functions, many of which are not known,
00:02:24.12 the oligodendrocytes mainly myelinate axons in the nerve and as I said they derive from dividing oligodendrocytes precursors,
00:02:32.14 or OPCs, and these cells in the optic nerve migrate into the nerve
00:02:37.06 early in development. They divide a limited number of times, and then they stop and terminally differentiate.
00:02:44.13 They don't divide again once they differentiate, and they myelinate axons.
00:02:48.02 So the number of oligodendrocytes in the optic nerve of a rat, of an adult rat, is about 350,000
00:02:59.05 and we want to know why. How is that number determined, why isn't it 100,000, why isn't it 500,000.
00:03:07.00 From what I've already told you, you will see that that number is going to depend on at least three things.
00:03:13.00 One is, how many precursors migrate into the nerve during development,
00:03:17.22 and we don't know that number, but I suspect its small.
00:03:20.01 Secondly, how much cell death occurs in the lineage, either in the precursors or in the oligodendrocytes,
00:03:26.07 and third, how many times the precursors divide once they are in the nerve
00:03:30.03 before they stop, differentiate, and start myelinating axons.
00:03:36.01 So I'm going to start by talking about the contribution of cell death to controlling cell number,
00:03:43.00 so this is the control of cell survival and death of oligodendrocytes, and you'll see it makes a major contribution
00:03:49.26 to getting the numbers of oligodendrocytes in the adult nerve right.
00:03:54.02 So here's an experiment done by a graduate student, Ian Hart, many years ago.
00:04:00.01 People in the United States had argued, that the oligodendrocyte precursor is
00:04:05.26 induced to differentiate into an oligodendrocyte by IGF-1, our old friend, Insulin-like Growth Factor-1,
00:04:12.03 because they found in culture, it increased the number of oligodendrocytes that developed.
00:04:17.02 So Ian, we had a different view, our view was if you remove the mitogen for oligodendrocytes precursors,
00:04:24.26 which is mainly platelet derived growth factor PDGF, that the cells
00:04:28.24 automatically differentiated into oligodendrocytes by default.
00:04:32.02 So to try to address this controversy, Ian took single oligodendrocyte precursors,
00:04:39.12 put them in a microwell, either without any signaling molecules or with IGF-1
00:04:45.29 or supernatants of cultures of its neighboring cells such as astrocytes.
00:04:50.01 What he found was, when there were no signals at all, the cell very quickly died. It died within a day,
00:04:57.10 and the morphology of the death was apoptosis, which was already known
00:05:02.24 to be a suicide program that the cell can activate to kill itself neatly and quickly.
00:05:07.02 Whereas when IGF-1 was present with a single cell, that was sufficient to allow the
00:05:13.24 cell to survive and not undergo apoptosis and signals from neighbors including IGF-1,
00:05:19.10 but others could also promote the survival of these cells.
00:05:23.02 And so we asked ourselves, why should it be, so we by the way concluded that IGF-1 isn't a differentiation
00:05:32.16 inducer in this system, it's simply allowing the cell to survive so that it can differentiate.
00:05:38.01 But the question is, why should the oligodendrocyte precursor die when there are no signals
00:05:44.02 there because we're giving it the nutrients and the vitamins and all the other things that cells need to live,
00:05:49.06 and so we propose that maybe all cells in an animal depend on
00:05:55.26 signals from other cells in order to keep this suicide apoptotic program off,
00:06:02.02 and this was a so-called "death by default" mechanism.
00:06:07.01 Now this would have advantages in building a complex animal because it would
00:06:11.11 ensure that any cell that didn't develop appropriately on the right time schedule or ended up in the wrong place,
00:06:18.07 would automatically kill itself because it wouldn't get the signals it needs to survive.
00:06:22.28 So it could be a very useful way of ensuring that cells only survive where and when they're needed.
00:06:29.02 But in addition, it provides a very powerful mechanism for regulating cell number.
00:06:34.21 As had already been shown in the nervous system, in the so-called neurotrophic hypothesis.
00:06:40.03 There it was shown years earlier that neurons are overproduced, many classes of neurons are overproduced,
00:06:50.09 they send out their axons, there long processes, contact the target cells
00:06:55.04 that they are going to innervate, there they compete for limiting amounts of the survival signal, the neurotrophic factors
00:07:02.00 released by the target cells only about half of them get enough to survive, and the other half kill themselves
00:07:08.26 by undergoing apoptosis and that is a very powerful mechanism
00:07:13.15 of adjusting the number of nerve cells to the number of target cells they innervate.
00:07:18.00 To carry this kind of hypothesis through to oligodendrocyte development,
00:07:24.09 Ben Barres, who was a post-doctoral fellow at the time, now in his own laboratory at Stanford,
00:07:29.29 showed that during normal development in the rat optic nerve, at least 50% of the newly formed oligodendrocytes
00:07:37.29 died by apoptosis, they killed themselves. So that led him to propose
00:07:42.11 this model that oligodendrocyte precursor cells divide rapidly, they stop dividing,
00:07:49.15 they differentiate into oligodendrocytes and at this point, the survival signals
00:07:54.16 that the newly formed oligodendrocytes need changes.
00:07:58.02 They now need to find an axon in order to live, and about half of those cells find an axon and live,
00:08:05.13 and myelinate the axon, the other half fail to find an axon and kill themselves,
00:08:10.19 and that would be a way of matching the number of oligodendrocytes to the number
00:08:15.01 and length of axons that needed myelination. So Ben and others then tested this hypothesis and
00:08:21.13 overwhelming evidence was rapidly provided that this was actually the way it works in
00:08:27.03 controlling oligodendrocyte survival and controlling their number.
00:08:30.02 Ben first showed that if you cut the optic nerve just behind the eye,
00:08:35.06 the axons distally along the nerve very rapidly degenerate in a day or two and what Ben showed is that
00:08:41.27 virtually all the oligodendrocytes then undergo apoptosis, they kill themselves,
00:08:46.21 consistent with the idea that they need a signal from axons to avoid apoptosis.
00:08:51.03 Then Julia Burne when she was a graduate student in the lab, made use of a transgenic mouse
00:08:57.05 that Jean-Claude Martinou had made which expresses a Bcl-2 gene,
00:09:03.08 which is anti-apoptotic, it produces a protein that blocks apoptosis, in neurons in the retina, and because retinal neurons
00:09:12.18 die during neuronal development, when you overexpress this transgene that death is markedly decreased
00:09:18.01 and you end up with double the number of axons in the optic nerve.
00:09:21.01 As a result, there's double the amount of survival signal, fewer oligodendrocytes die, and their number increases
00:09:28.29 to match the increased number of axons. You could see why evolution would have chosen to do it this way.
00:09:35.06 Overproduce the cells and then death sculpts the number down to the number that are appropriate.
00:09:41.02 Bill Richardson, a colleague at University College, showed that if you overexpress platelet derived growth factor
00:09:48.15 in the optic nerve, then you greatly stimulate the proliferation in the oligo precursors,
00:09:53.14 and you get many more oligodendrocytes produced than normal.
00:09:57.00 But because the number of axons hasn't changed, all those extra oligodendrocytes kill themselves,
00:10:02.25 and you end up with a normal number of oligodendrocytes as you would predict
00:10:06.14 from the model that Barres had proposed. Charles Ffrench-Constant who was a post-doc at the time,
00:10:13.10 showed in his own laboratory in Cambridge that if you decrease oligodendrocyte
00:10:19.11 production in a way that I'm not going to describe, then again,
00:10:23.26 fewer oligodendrocytes die because there is less competition for the survival signals that the axon provides,
00:10:33.01 and so the number again adjusts to be normal. So what is the survival signal,
00:10:40.19 or what are the survival signals that the axon provides? Pierre-Alain Fernandez,
00:10:47.26 who was then a postdoctoral fellow, provided strong evidence that our old friend glial growth factor,
00:10:53.26 the neuregulin, is an important survival signal that the axon provides for the newly formed axon.
00:11:01.01 First he showed as shown here that when you take oligodendrocytes in culture without signals,
00:11:06.11 they very quickly undergo apoptosis, whereas if you add glial growth factor, you promote their survival potently.
00:11:14.02 Then he showed that neurons from the dorsal root ganglion, a sensory ganglion beside the spinal cord,
00:11:22.08 will also promote the survival of oligodendrocytes in the absence of other signals.
00:11:27.03 And if you add the dorsal root ganglion neurons and additionally add antibody against glial growth factor,
00:11:35.07 that decreases the survival promoting effect of the axons, suggestion the axons
00:11:42.00 are promoting survival of oligodendrocytes through GGF.
00:11:46.02 And finally, if you add GGF with the antibody, which neutralizes the antibody,
00:11:51.15 you bring back the survival promoting effects to the normal level.
00:11:55.00 The critical experiments that Pierre-Alain did was to show that neuregulin,
00:12:00.11 the glial growth factor, operates in the animal as well to promote survival of oligodendrocytes.
00:12:07.01 So here what he's done is he takes a cell line COS cells, a monkey cell line,
00:12:14.00 and puts in various genes that enables those cells to secrete various types of signaling proteins.
00:12:21.01 So if he takes COS cells and he transfects them with a gene that encodes glial growth factor
00:12:28.17 and then transplants those COS cells into the brain of a developing rat, then you see that the apoptosis of
00:12:37.11 oligodendrocytes is greatly decreased so that adding exogenous GGF
00:12:42.15 to a developing rat brain, an optic nerve, decreases the normal death of oligodendrocytes in the optic nerve.
00:12:51.02 Then in this critical experiment he shows that if you transfect the COS cells
00:12:56.26 with a construct that encodes a receptor for glial growth factor
00:13:02.01 fused to the Fc region of an immunoglobulin molecule, and this molecule is secreted by
00:13:07.08 the COS cells and now transfer those cells into the brain of a developing rat, then you greatly increase apoptosis.
00:13:15.01 So you are neutralizing the endogenous glial growth factor
00:13:19.04 presumably provided by the axons and now death of oligodendrocytes go up.
00:13:23.02 So all of that together strongly suggests that the axons in the optic nerve are promoting survival
00:13:32.06 in newly formed oligodendrocytes at least in part by glial growth factor on the surface of the axon where it is known to be.
00:13:41.02 So now I want to move from cell survival and death control as a process that helps control
00:13:49.09 the number of oligodendrocytes in the developing optic nerve, to proliferation control.
00:13:54.03 And again, looking at the optic nerve here, I remind you that early in development the oligodendrocyte
00:14:05.10 precursors, the OPCs, migrate into the nerve, they divide rapidly
00:14:10.05 a limited number of times and then they stop and differentiate into oligodendrocytes.
00:14:16.00 The number of divisions is going to greatly influence the number of oligodendrocytes that develop.
00:14:23.00 So the question we addressed is why do oligodendrocyte precursors stop dividing and differentiate when they do?
00:14:31.01 And I should say that most of the cells in your body work in that way, they are developed from precursor cells
00:14:38.07 that divide a limited number of times, and then they stop and differentiate,
00:14:42.26 that is true for the skin, for the gut, for most cells in the brain and the muscle and most blood cells and so on.
00:14:49.00 Yet there isn't a single case where we understand why the cells stop dividing and differentiate
00:14:55.03 when they do, and we wanted to try to address that question for the oligodendrocyte precursor.
00:15:01.02 So the first thing we did was to get this dropping out of division and differentiation to occur
00:15:09.16 in a dissociated cell culture of the optic nerve. So if you look at the optic nerve, in blue
00:15:15.10 you see the development of oligodendrocytes in the intact rat optic nerve.
00:15:20.00 So the precursors migrate into the nerve a few days before birth, they divide there,
00:15:25.27 the first precursors stop dividing and become oligodendrocytes on the day of birth,
00:15:30.20 which is embryonic day 21 around in the rat, and new precursors drop out of division and differentiate
00:15:37.27 for the next six weeks postnatally in the optic nerve of a rat.
00:15:41.17 If you take the cells out of the optic nerve three days before birth, dissociate them, put them in culture
00:15:48.05 in the presence of serum, these cells will divide. The first oligodendrocyte precursor stop dividing
00:15:54.27 and differentiate after 3 days, the equivalent of the day of birth, and new ones will do so for the next weeks in culture.
00:16:01.01 So we've now confined to a culture dish, the process that we want to study. That is
00:16:06.26 when the cells stop dividing and differentiate seems to happen just as it does in the nerve in this culture dish.
00:16:14.01 This is a pretty complicated culture. There is about six or seven cell types
00:16:18.16 in the optic nerve and in this culture there is fetal calf serum present.
00:16:23.01 So a big step forward was taken by Ben Barres as I said earlier now in his own lab in Stanford,
00:16:32.03 he purified the oligodendrocyte precursors from the rat optic nerve. We had antibodies that distinguished
00:16:38.27 the precursors from the differentiated oligodendrocytes and from the astrocytes and so using
00:16:43.09 sequentially immunopanning he was able to purify the oligodendrocyte precursors to homogeneity.
00:16:50.02 And he showed that you take these precursors, put them in culture
00:16:53.23 with the major mitogen PDGF, platelet derived growth factor,
00:16:58.09 that Bill Richardson and Mark Noble had previously shown as the main mitogen for the oligodendrocyte precursors.
00:17:04.02 If you do that, these cells will divide in culture, they will migrate in culture
00:17:09.07 as they do in the animal, and they will stop dividing and differentiate after a limited number of divisions.
00:17:15.02 So in this relatively simple culture system the cells seem to behave in the way they do in the animal.
00:17:22.02 And here is what the cells look like in a time-lapse video recording. So you can see that the
00:17:29.10 cells are migrating, these are purified oligodendrocyte precursors in PDGF, they are migrating,
00:17:34.25 they are dividing, and here is a cell that has stopped dividing, stopped migrating,
00:17:38.02 becoming an oligodendrocyte. Here's a cell that's doing the same, stops migrating, stops dividing,
00:17:43.25 and becoming an oligo, and after seven or eight days these cells will have all stop dividing and differentiate.
00:17:50.01 And the reason the cells do it at different times even though they come from the same optic nerve,
00:17:55.04 we think is because they are at different stages of maturation.
00:17:59.02 Cause if you go earlier, if this is a post-natal day seven optic nerve, where these cells were isolated,
00:18:04.05 if you go to the embryo then the cells go through more divisions before they stop and differentiate
00:18:09.11 so it seems likely that this different number of divisions depends on the maturation of the cells and that video
00:18:15.12 was taken by Nathalie Billon when she was a post-doctoral fellow, now in her own laboratory in Nice in France.
00:18:22.00 Okay, so the next step forward was taken by Fen-Biao Gao at the Gladstone Institute in San Francisco,
00:18:30.21 where he purified the oligodendrocyte precursors using
00:18:34.15 the sequential immunopanning procedure that Barres had developed
00:18:37.03 from embryonic day 18 rat optic nerve. Now that was no mean feat because these cells are less than 0.1%
00:18:44.16 of the cells in the nerve, but he was able to do it. And if you culture those cells in PDGF
00:18:49.04 without serum, now those cells will behave just as they do in the nerve.
00:18:53.24 They will divide, the first cells will stop dividing and differentiating at the equivalent of the day of birth,
00:18:58.02 and new cells will do so for the next days or weeks in culture. So that critical experiment says that the
00:19:08.06 timing of when the cells stop dividing and differentiate is either built into the population of the cells,
00:19:14.09 because during this experiment the environment is kept constant, there is no other cell types present,
00:19:20.10 so it is either built into the population of precursors or its built in the individual cells.
00:19:25.01 The evidence that it might be built in individual cells came earlier from a graduate student Sally Temple,
00:19:32.24 who is now in her own laboratory in Albany in New York. What she did was to micromanipulate a single oligodendrocyte precursor
00:19:40.15 and put it on a monolayer of astrocytes in a microwell. The astrocytes make
00:19:46.06 platelet derived growth factor and the survival signals that these cells need to survive
00:19:50.28 and so the cell will now divide and differentiate and form a clone.
00:19:56.09 What she found is that in blue are the number of oligodendrocyte precursors
00:20:02.21 within this typical clone, and the cell in this case divides
00:20:06.21 four times to give eight cells, and at this point,
00:20:09.28 well it gives 16 cells rather, and at this point all of the progeny of this single cell
00:20:15.28 stopped dividing and differentiate to form oligodendrocytes at the same time.
00:20:21.18 So that strongly argued that the cell has some built in mechanism to in fact control when it stops dividing.
00:20:30.06 But Sally actually proved that in this way. She put the astrocytes in the side of the microwell,
00:20:36.06 put a single precursor on the base, let the cell divide once and then transfered
00:20:41.25 the sibling cells to two separate astrocyte monolayers, and showed that if one sibling divided three times
00:20:49.17 for three days and then stopped so the other sibling tended to do as well.
00:20:55.01 So that established beyond any doubt that built into each individual precursor cell is a mechanism
00:21:02.09 thatâ€™s determining or helping to determine when the cell stops dividing and differentiate.
00:21:09.01 Could it be that the cell is simply counting divisions. In Sally's experiment using postnatal day seven optic
00:21:15.16 nerve precursors, the maximum number of divisions she saw was eight. Is it possible that the cells are
00:21:21.18 just counting and when they get to eight they just stop and differentiate.
00:21:24.01 Here's another experiment done by Fen-Biao Gao that suggests that isn't the way this thing is working.
00:21:30.09 He dropped the temperature in the culture dish to 33 degrees and compared it to cells cultured at the usual 37 degrees.
00:21:39.01 Now this is an artifact, in an animal, in a mammal, the temperature never gets to 33 degrees if the animal is alive.
00:21:45.01 However, as you'd expect, at the low temperature the cell cycle time is increased.
00:21:52.22 The cell cycle is slow, that's what you'd expect at the low temperature.
00:21:56.01 What you might not have expected is that at the low temperature the cells stop dividing,
00:22:01.03 and differentiate sooner after fewer divisions at the lower temperature so you dissociated cell division
00:22:09.13 counting and timing when start dividing and differentiate, and so we now call this thing an intracellular timer,
00:22:16.01 because it seems to measure time in some way that doesn't depend on counting cell divisions.
00:22:22.02 Although this timer is built into the individual precursor cell, it doesn't act autonomously.
00:22:31.02 Cells in animals never act autonomously, every process in the cell, almost, is regulated by signals
00:22:39.20 from other cells, and that is true too of this timer. As Ben Barres showed when he was a post-doc,
00:22:45.27 first if you take PDGF out of this culture of purified precursors, the cells immediately stop dividing and differentiate.
00:22:55.01 So for the timer to work normally, you need PDGF to keep the cells dividing.
00:23:01.01 If there is no thyroid hormone here, then the cells just continue to divide, they don't stop dividing
00:23:09.03 and differentiate when they normally would. But if you let the cells divide in PDGF without thyroid
00:23:14.11 hormone for eight days and now add back thyroid, now the cells very quickly stop dividing and differentiate,
00:23:24.05 and if the cells are in PDGF and thyroid hormone together, then they stop after seven or eight days and differentiate.
00:23:32.00 Some doing it earlier, some doing it later. So the simple interpretation of this experiment is that there is
00:23:40.00 an intracellular timer, the timer itself does not depend on thyroid hormone, but when the timer says
00:23:47.18 now, stop dividing and differentiating, if there is no thyroid hormone there the cells continue to
00:23:52.06 proliferate and don't differentiate. So the timer has at least two components, something that is
00:23:57.05 keeping time independent of thyroid hormone and an effector mechanism that depends on
00:24:02.05 thyroid hormone that allows the cell to stop and differentiate. So now we wanted to understand, what's
00:24:09.07 the molecular mechanism of this intracellular timer, and it turns out that it's quite complex.
00:24:16.02 I'm going to give you two examples of proteins that change in their concentration in the cell as the precursors
00:24:24.23 are proliferating that make sense and are components of the timer. One of these is the P27 member
00:24:34.29 of the CIP/KIP family of cyclin-dependent kinase inhibitors. These proteins act as a brake on
00:24:43.16 the cyclin E/cyclin A Cdk2 complexes that are important for driving the cell through G1 and into S phase.
00:24:51.02 So if it's part of the timer you would expect it to increase as the cells proliferate to help take the cells
00:24:58.04 out of division after a certain period of time. And that is indeed what happens, and that was shown by
00:25:04.13 Bea Durand when she was a post-doctoral fellow, now in her own laboratory at the Pasteur Institute in Paris.
00:25:10.02 So here's she's taken purified oligodendrocyte precursors from the post-natal day seven
00:25:15.13 rat optic nerve, cultured them in, in this case, without thyroid hormone, but in the presence of PDGF.
00:25:24.02 So these cells in the course of this experiment are going to continue to divide,
00:25:28.20 they are not going to stop and differentiate because there is no thyroid hormone, but if you look at what happens
00:25:33.15 to the level of P27 in the nucleus, asses quantitatively by
00:25:38.03 confocal fluorescent microscopy, early on the levels are low, late on the levels are high.
00:25:45.02 If you plot the levels in the population they start off low, increase, and then plateau right around the time
00:25:52.00 the cells would have stopped dividing, so that's exactly the behavior you'd expect if P27
00:25:57.09 was playing a role in helping to take the cells out of division. And I should say that these large bars
00:26:03.26 here are not error of standard deviations, they are the extremes within the population.
00:26:08.23 And remember, that at this point there's a heterogeneity of cells in terms of how long they will divide for
00:26:14.17 before they differentiate. Some will stop now, other will go for another four, five, six days.
00:26:19.24 If you look at a clone of cells, then the level of P27 is the same in all the cells in the clone.
00:26:27.01 Here is one explanation for why the timer works faster at the low temperature.
00:26:37.26 If you look at the level of the P27 protein in oligodendrocyte precursors dividing in culture at 33 degrees,
00:26:47.00 when the timer ran faster, P27 levels rise faster. So presumably this is one reason why the timer is accelerated
00:26:57.10 at the low temperature. It's probably not the only reason, but it's one reason. So here's a gain of function experiment
00:27:03.25 to ask if you overexpress P27 in purified oligodendrocyte precursors proliferating in culture,
00:27:10.27 in the presence of both in this case PDGF and thyroid hormone, what happens to the timing of when
00:27:19.19 the cells stop and differentiate. The answer is if you use a retrovirus to express either green
00:27:28.04 fluorescent protein or green fluorescent protein and P27, you find that green fluorescent protein
00:27:36.03 has no effect on the time. The cells will divide, stop dividing, and differentiate, most of them
00:27:40.24 within seven to eight days. But if you overexpress P27 so that the levels rise artificially fast,
00:27:47.26 now you accelerate the timer and cells stop dividing and differentiate sooner. That's what you'd
00:27:54.20 expect if P27 was part of this timer. Here's a loss of function experiment, Bea Durand collaborated
00:28:02.19 with one of the three labs that knocked out the gene encoding P27 in the mouse,
00:28:09.16 and this was Jim Roberts in Seattle. And then she got the mouse from Jim Roberts, isolated the optic nerve cells,
00:28:17.26 cultured them in the presence of PDGF and thyroid hormone so that the timer should operate normally,
00:28:25.01 and then counted the number of oligodendrocytes in clones to ask how many divisions
00:28:31.00 do the cells go through before they stopped and differentiated. And what they found was, what she found,
00:28:36.15 was in the wildtype animal the maximum number of divisions was six. Now interestingly in rats of the same age
00:28:44.02 that number was eight. And whether one reason a rat is bigger than a mouse is that precursors go
00:28:51.14 through more divisions before they stop could well be one reason the rat is bigger than the mouse in this case.
00:28:57.02 But the important experiment's this one, where you have only one copy of P27 gene or no copies,
00:29:05.10 then some of the cells proliferate for longer, and you get extra divisions before the cells stop dividing.
00:29:11.02 Cell cycle time is unchanged, the experiment argues strongly, in fact these experiments taken together
00:29:17.18 argue strongly, that P27 is a component of the timer. It's a minor component because when you take it away,
00:29:24.21 the cells still stop and differentiate, but they don't do so accurately.
00:29:29.01 They go through another division or two abnormally before they stop and differentiate.
00:29:34.01 Now the generality here is that P27 knockout mice that have no P27 at all, are 30% larger than wildtype
00:29:43.19 normal mice, and the reason they are larger is that they have more cells in every organ that's been looked at.
00:29:48.02 And they have more cells because there is extra proliferation, not because there is less cell death.
00:29:53.24 So it seemed likely that P27 plays a role in helping to take cells out of division at the right time
00:30:00.19 in many cell lineages, not just the oligodendrocyte lineage.
00:30:04.01 Even more generally, there are P27 homologs in both C. elegans worms and in Drosophila.
00:30:13.02 One gene in each of those organisms, and when you inactivate those genes you get an extra division
00:30:20.03 or two in multiple lineages, both in the worm and in the fly. So it argues that CDK inhibitor proteins
00:30:26.14 such as P27 probably play a role in timing when cells come out of division in all animals.
00:30:33.02 Now to identify a component, a protein component of the timer, doesn't actually tell you how the timer
00:30:42.20 works. You need to know why P27 protein accumulates over time as the oligodendrocyte precursors proliferate.
00:30:51.02 Well we don't know why it accumulates but we do know that the mechanisms are post-transcriptional
00:30:57.05 thanks to the work of Yasu Tokumoto when he was a post-doctoral fellow, now back in Tokyo.
00:31:03.03 He showed using RT-PCR, or real time PCR, that as the protein in OPCs increases over time as they
00:31:13.23 proliferate, messenger RNA levels for P27 remain flat. So whatever the mechanisms are, they are
00:31:22.18 post-transcriptional. Recently, a talented post-doctoral fellow in Ben Barres lab at Stanford, Jason Dugan,
00:31:32.02 has identified P57, which is another member of the CIP/KIP family of CDK inhibitors
00:31:38.00 as an important component of the timer. He's shown, just like in the case of P27,
00:31:42.23 it increases over time and plateaus, and in this case he's shown that the messenger RNA
00:31:48.01 and the protein increase together, arguing that its likely
00:31:52.29 transcriptional control. So I now want to turn to a second or third protein
00:31:58.25 that works in the opposite way of the CDK inhibitors. These are the inhibitors of differentiation
00:32:07.00 proteins. The so-called Id proteins. These proteins block the basic helix-loop-helix proteins that are
00:32:17.08 known to be required for differentiation in most lineages. They bind to them, block their activity,
00:32:23.15 inhibit differentiation and thereby promote proliferation. So if the Id proteins are going to play a part
00:32:30.29 in this timer, their concentration should fall over time as these cells proliferate.
00:32:36.02 That is indeed what happens as was shown by Toru Kondo, a very talented post-doctoral fellow
00:32:42.21 now in his own laboratory in Kobe, Japan. So he showed that four Id proteins, or at least messenger RNAs,
00:32:52.08 are expressed in oligodendrocyte precursors but only one of them decreases over time
00:32:57.24 as these cells proliferate, suggesting its a component of the timer. And in this case, both the
00:33:04.06 messenger RNA and the protein fall progressively both in vitro as shown here and in vivo
00:33:12.03 message and protein fall progressively as the precursors divide. Here's a gain of function experiment
00:33:20.07 where Toru's used a retrovirus to put Id-4 protein into oligodendrocyte precursors
00:33:28.10 together with green fluorescent protein. And now he induces differentiation by removing platelet derived
00:33:34.16 growth factor and shows that the cells that are transfected and overexpressed Id-4 fail to drop out of
00:33:41.21 division and do not differentiate, whereas overexpressing Id-1 in these cells with GFP doesn't have this effect.
00:33:49.01 So it's some specificity for Id-4 over the other Id proteins, and the effect here is quantified in this graph.
00:33:57.10 So overexpression of Id-4 blocks differentiation and promotes proliferation.
00:34:03.01 Finally here's a loss of function experiment where he's collaborated with Fred Zablitsky in Nottingham who
00:34:09.26 has made an Id-4 knockout mouse. Toru has taken neural stem cells from the brains of these knocked
00:34:17.29 out mice, puts them in culture with PDGF, thyroid hormone, and another signaling protein
00:34:23.07 sonic hedgehog, thatâ€™s required to get oligodendrocyte development on schedule.
00:34:28.02 In the Id-4 knockout mouse, you get oligodendrocytes days prematurely compared to the
00:34:36.15 wild-type neural stem cells, suggestion that Id-4 really is a component of the timer and helps oligodendrocyte
00:34:44.03 precursors to differentiate at the right time. So let me finish by going back to the initial question.
00:34:54.02 Why is it that we grow to be larger than mice?
00:34:59.15 I've already told you that the reason we are bigger than mice is that we have more cells than mice.
00:35:05.02 The reason we have more cells is that on average, human cells divide more before they stop than do mouse cells.
00:35:14.02 The question is, why do they divide more times before they stop?
00:35:19.18 Is it because the intracellular timers that I've been talking about that help determine how long the cells divide
00:35:27.09 before they stop are set differently, are programmed differently so that cells divide much longer
00:35:33.04 in humans before they stop because of an intrinsic timer mechanism set in this way.
00:35:38.03 Or is it that the signals, the mitogens, the growth factors that are responsible for driving the growth
00:35:46.25 and proliferation of human cells are around longer than they are in the mouse.
00:35:53.14 My guess is that both of these things will turn out to be true, but the fact is we don't know and
00:35:59.12 we don't know largely because this problem hasn't been studied in this way.
00:36:05.03 So what I have ignored is the fact that most of us don't look like mice. Not only are we larger
00:36:13.16 than mice but we don't look like mice. And the reason we don't look like mice is because these
00:36:19.21 local controls on growth, so that you get a nose and you get ears of this shape and so on,
00:36:25.14 are controlled differently in a mouse from in a human. And we know very little about the local controls
00:36:34.07 that pattern growth control such that humans have bigger heads than mice and
00:36:40.24 they don't have the same facial appearance and so on.
00:36:45.01 Anyway, this is a really important problem in development, growth control, and it needs more
00:36:51.27 people working on it. So it just remains for me to thank the people who I think I've mentioned
00:36:59.05 throughout this talk, who have made all this work possible, and I'm grateful to them,
00:37:05.18 they are an outstanding group of people without them I wouldn't be here today so thanks.

Videos in this Talk

Part 1: Regulation of Cell Size
Audience:
- Student
- Researcher
- Educators of H. School / Intro Undergrad
- Educators of Adv. Undergrad / Grad
Part 2: Cell Number Control
Audience:
- Researcher
- Educators of Adv. Undergrad / Grad

Speaker: Martin Raff

Total Duration: 1:18:09

Recorded: November 2006

Educator Resources for this Talk

All Talks in Development and Stem Cells

Talk Overview

The size of an organ or organism depends mainly on the sizes and numbers of the cells it contains. In the first segment of my talk, I describe our work on cell size control in cultures of purified rat Schwann cells. Most proliferating cells grow before they divide, but it is not known how growth and division are co-ordinated to ensure that cells divide at an appropriate size. We have found that extracellular signals can control cell growth and cell-cycle progression separately and that the size of Schwann cells at division depends on the signalled rates of both cell growth and cell-cycle progression, rather than on a cell-size checkpoint that monitors cell size.

In the second segment of my talk, I describe our work on cell number control in the rat oligodendrocyte cell lineage. Cell numbers depend on controls on both cell death and cell proliferation. We have found that oligodendrocytes are normally overproduced and kill themselves in large numbers in a competition for survival signals on the surface of the axons that the oligodendrocytes myelinate. Most differentiated cells, including oligodendrocytes, develop from dividing precursor cells that divide a limited number of times before they terminally differentiate, but it is not known what stops cell division and initiates differentiation. We have found that oligodendrocyte precursor cells have an intrinsic timing mechanism that helps determine when they stop dividing and differentiate.

Speaker Bio

Martin Raff

Martin Raff

Martin Raff was born and educated in Montreal. He received his BSc and MD degrees at McGill University and did a residency in medicine at the Royal Victoria Hospital in Montreal and in neurology at the Massachusetts General Hospital in Boston. He did postdoctoral training in immunology at the National Institute for Medical Research in… Continue Reading

Leave a Reply Cancel reply