Portal de Periódicos da CAPES

Size Control: The Regulation of Cell Numbers in Animal Development

1996; Cell Press; Volume: 86; Issue: 2 Linguagem: Inglês

10.1016/s0092-8674(00)80087-2

1097-4172

Martin Raff,

Cancer, Hypoxia, and Metabolism

How is the size of an organism or organ determined? Why, for example, are we larger than mice? Although growth can occur by cell enlargement and accumulation of extracellular matrix, the size of an animal generally reflects cell numbers. We are bigger than mice mainly because we have more cells than a mouse, not because our cells are larger. We have more cells because cells in a developing human divide more times, on average, than do cells in a developing mouse. What, then, controls how many times a cell divides? The answer is not known, and, despite its importance, the question has received little attention. Several recent papers seem likely to change that. Before considering them, however, I shall briefly review some general aspects of cell number control. Cell number depends on more than just cell proliferation. It also depends on cell death and, in some organs, on cell immigration, emigration, or both. Division, death, and migration all depend on intracellular mechanisms that are regulated by extracellular signaling molecules produced by other cells. The signaling molecules can either activate or inhibit the intracellular mechanisms; they can operate locally or systemically and can be soluble, bound to cell surfaces, or associated with the extracellular matrix. The challenge is to understand how the extracellular signals and intracellular mechanisms interact to ensure that each organ contains the right numbers of each cell type. The right number of cells depends on cell size, which in turn depends on the amount of DNA in the nucleus. The cells in a tetraploid salamander, for example, are four times the volume of those in a haploid salamander, yet the corresponding organs in the two animals are the same size because the tetraploid organs contain a quarter as many cells as the haploid (6Frankhauser G Int. Rev. Cytol. 1952; 1: 165-193Crossref Scopus (33) Google Scholar). Clearly, cell number control is not merely a matter of counting cells or cell divisions. Systemic controls help coordinate the growth of different organs during development. In vertebrates, for example, the anterior pituitary gland seceretes growth hormone, which stimulates growth principally by inducing the production of insulin-like growth factor 1 (IGF-1) by the liver and other organs. IGF-1 in turn promotes cell survival in many tissues and cell proliferation in some. The importance of systemic controls on growth can vary greatly between organs. If one transplants multiple fetal thymus glands into a developing mouse, each grows to its normal adult size, suggesting that their growth is mainly controlled locally within the thymus (13Metcalf D Aust. J. Exp. Med. Sci. 1963; 41: 437-448Crossref Scopus (65) Google Scholar). If one performs the same experiment with fetal spleens, the opposite result is obtained: the total mass of the transplanted spleens attains the mass of one normal adult spleen, suggesting that spleen growth is mainly controlled by factors outside the spleen (14Metcalf D Transplantation. 1964; 2: 387-392Crossref Scopus (44) Google Scholar). In neither case are the control mechanisms understood. Cyclin-dependent protein kinases (Cdks) directly control the eukaryotic cell-division cycle. They are cyclically activated to trigger the different phases of the cell cycle at the right time and in the right sequence. They themselves are controlled by a variety of regulatory proteins: cyclins activate them and help direct them to their substrates; kinases and phosphatases phosphorylate and dephosphorylate them, respectively; and Cdk inhibitors block their activity or their assembly with cyclins (11Lees E Curr. Opin. Cell Biol. 1995; 7: 773-780Crossref Scopus (262) Google Scholar). Except for the initial cleavages of the zygote and blastomeres, which are apparently cell-autonomous, most animal cell divisions depend on extracellular growth factors, which are mainly produced by neighbouring cells. The growth factors are required to maintain the cell-cycle control system: if a cell in culture is deprived of such factors, for example, the cell cycle arrests at a checkpoint in G1 called the restriction (R) point, and the cell enters a modified G1 state (G0), in which some of the Cdks and cyclins disappear. The dependence of animal cell division on signals from other cells helps to ensure that a cell divides only when another cell is needed. When part of the adult liver is removed, for example, cell proliferation returns the liver to normal size. In regenerating limbs, and probably in other regenerating and developing tissues also, if a disparity occurs in the positional values of neighbouring cells, proliferation produces cells with intermediate positional values, thereby creating a continuous and approriate cell pattern. It remains a mystery how such disparities are detected by cells and how they stimulate cell proliferation. A large effort has been devoted to identifying both the extracellular signaling molecules that stimulate cell proliferation and the intracellular signaling pathways that they activate. By contrast, there has been relatively little exploration of the mechanisms responsible for stopping cell proliferation at the appropriate point during regeneration or development. The stopping mechanisms are important, as they can influence both cell numbers and the timing of differentiation. In part at least, the stopping mechanisms are cell-intrinsic and depend on the cell's history. In principle, such an intrinsic mechanism could depend on a decrease in a positive intracellular regulator that is required to keep the cell dividing, an increase in a negative intracellular regulator that inhibits cell-cycle progression, or both. An example of the first mechanism is seen in the first cells that stop dividing at the cellular blastoderm stage of Drosophila development: these cells arrest in the G2 phase of the cell cycle because the supply of maternal String protein—the phosphatase required to activate the mitotic Cdk (Cdc-2)—runs out, so that the cells cannot progress into M phase (4Edgar B.A O'Farrell P.A Cell. 1990; 62: 469-480Abstract Full Text PDF PubMed Scopus (344) Google Scholar). A possible example of the second mechanism is seen in fibroblast senescence, where accumulation of the Cdk inhibitor p16Ink4 (p16) may be in part responsible for limiting the number of times normal fibroblasts divide in culture when stimulated by serum: early-passage human fibroblasts express low levels of p16, while late-passage, senescent ones express high levels (7Hara E Smith R Parry D Tahara H Stone S Peters G Molec. Cell Biol. 1996; 16: 859-867Google Scholar), and embryo fibroblasts from p16−/− mice readily escape such senescence (18Serrano M Lee H-W Cordon-Cardo C Beach D DePinho R.A Cell. 1996; 85: 27-37Abstract Full Text Full Text PDF PubMed Scopus (1356) Google Scholar). Just as animal cells need signals from other cells to divide, so most seem to need signals from other cells to survive. If deprived of such signals, the cells activate an intrinsic death program and kill themselves—a process called programmed cell death (PCD). Dependence on survival signals may help ensure that a cell survives only where and when it is needed. PCD is mediated by a proteolytic cascade, involving a family of cysteine proteases: when activated by proteolytic cleavage, these enzymes cleave key intracellular proteins, including other family members, at specific aspartic acid residues to kill the cell quickly and neatly (12Martin S Green D Cell. 1995; 82: 349-352Abstract Full Text PDF PubMed Scopus (1241) Google Scholar). In mammals, extensive PCD begins in the blastula and continues throughout life. In an adult human, for example, billions of cells undergo PCD every hour. Thus cell death can be just as important in controlling cell numbers as cell proliferation. In the developing vertebrate nervous system, for instance, many types of neurons are overproduced and then compete with one another for limiting amounts of survival signals (neurotrophic factors) secreted by the target cells they innervate. Only about half get enough signal to survive, while the others undergo PCD. In this way the numbers of neurons are matched to the numbers of target cells. A similar mechanism may operate in many developing organs, allowing different cell types to adjust their numbers automatically when the number of one cell type is perturbed, thereby facilitating both development and evolution. It may also operate in adult organs: if hepatocyte proliferation is transiently stimulated in adult rats by phenobarbital treatment, for example, increased PCD rapidly returns the liver to its normal size. p27Kip1 (p27), p21Cip1 (p21), and p57Kip2 are members of the Cip/Kip family of Cdk inhibitors (19Sherr C.J Roberts J.M Genes Dev. 1995; 9: 1149-1163Crossref PubMed Scopus (3155) Google Scholar). In vitro, each can inhibit the various cyclin–Cdk complexes required for G1 progression and entry into S phase. All of these complexes can phosphorylate Rb, causing it to release E2F (or a related transcription factor), which can then activate the transcription of genes needed for progression into S phase. The two known classes of vertebrate Cdk inhibitors and some other cell-cycle control proteins are illustrated in Figure 1. Several lines of evidence suggest that p27 may have a role in limiting cell proliferation, especially in response to antiproliferative extracellular signals such as TGF-β (19Sherr C.J Roberts J.M Genes Dev. 1995; 9: 1149-1163Crossref PubMed Scopus (3155) Google Scholar): as for other Cip/Kip family members, overexpression of p27 leads to cell-cycle arrest in G1 in all cultured cell lines that have been tested; p27 levels are high in many types of terminally differentiated, postmitotic cells in tissues, as well as in cells in culture that have stopped dividing in response to TGF-β, contact inhibition, or serum deprivation; and treatment with p27 antisense oligonucleotides prolongs cell proliferation in serum-deprived cultured fibroblasts (1Coats S Flannagan W.M Nourse J Roberts J Science. 1996; 272: 877-880Crossref PubMed Scopus (640) Google Scholar). Recently, three laboratories (5Fero M.L Rivkin M Tasch M Porter P Carow C.E Firpo E Polyak K Tsai L.-H Broudy V Perlmutter R.M Kaushansky K Roberts J.M Cell. 1996; 85: 733-744Abstract Full Text Full Text PDF PubMed Scopus (1306) Google Scholar, 10Kiyokawa H Kineman R.D Manova-Todorova K.O Soares V.C Hoffman E.S Ono M Khanam D Hayday A.C Frohman L.A Koff A Cell. 1996; 85: 721-732Abstract Full Text Full Text PDF PubMed Scopus (1117) Google Scholar, 15Nakayama K Ishida N Shirane M Inomata A Inoue T Shishido N Horii I Loh D.Y Nakayama K-i Cell. 1996; 85: 707-720Abstract Full Text Full Text PDF PubMed Scopus (1437) Google Scholar) independently demonstrated that mice in which both copies of the p27 gene have been inactivated by targeted gene disruption grow more rapidly and are about one third larger than normal, despite normal levels of serum growth hormone and IGF-1; interestingly, mice with only one p27 gene inactivated are about one sixth larger than normal. All of the organs examined in the p27−/− mice are increased in size and contain more cells, apparently as the result of increased cell division, rather than decreased cell death. With time, the mice develop adenomas in the intermediate lobe of the pituitary gland (as do Rb-deficient mice), but so far they do not have an increased incidence of other tumors. Whereas male p27−/− mice are fertile, females are sterile. The striking phenotype of p27−/− mice contrasts with p21−/− mice, which seem to develop normally (2Deng C Zhang P Harper J.W Elledge S Leder P Cell. 1995; 82: 675-684Abstract Full Text PDF PubMed Scopus (1890) Google Scholar). These results strongly suggest that p27 normally plays a part in limiting cell proliferation in many lineages, probably by inhibiting cyclin–Cdk complexes that operate in G1, thereby helping cells exit from the cell cycle. The p27−/− mice have more cells per unit volume, at least in the liver and brain (5Fero M.L Rivkin M Tasch M Porter P Carow C.E Firpo E Polyak K Tsai L.-H Broudy V Perlmutter R.M Kaushansky K Roberts J.M Cell. 1996; 85: 733-744Abstract Full Text Full Text PDF PubMed Scopus (1306) Google Scholar), suggesting that cell size is reduced, presumably because the increased activity of the G1 cyclin–Cdk complexes shortens the cell cycle, allowing less time for cell growth. This is consistent with previous findings that cells overexpressing cyclins D or E (19Sherr C.J Roberts J.M Genes Dev. 1995; 9: 1149-1163Crossref PubMed Scopus (3155) Google Scholar), as well as fibroblasts from Rb−/− mice (8Herrera R.E Sah V.P Williams B.O Mäkelä T.P Weinberg R.A Jacks T Mol. Cell. Biol. 1996; 16: 2402-2407Crossref PubMed Scopus (269) Google Scholar), also have a shortened cell cycle and are smaller than wildtype cells. The results also indicate that, even in the affected cell lineages, p27 is not required for cells to exit from the cell cycle, as most cells in p27−/− mice still manage to stop dividing and terminally differentiate, perhaps through the action of other Cdk inhibitors. Members of the Cip/Kip family have been shown to cooperate with members of the Ink family of Cdk inhibitors to arrest the cell cycle in a TGF-β-treated cell line (17Reynisdóttir I Polyak K Iavarone A Massagué J Genes Dev. 1995; 9: 1831-1845Crossref PubMed Scopus (865) Google Scholar), and it would not be surprising if similar collaborations operate during development. Why does cell death not increase in p27−/− mice to bring cell numbers back to normal? This may be because many cell types are increased in each of the enlarged organs, thereby increasing the levels of many kinds of survival factors. Similarly, some of the increased cell proliferation in these mice may be secondary to an increase in growth factors produced by those cells that have proliferated excessively as a direct consequence of the p27 deficiency. How are p27 levels normally controlled so that the protein helps stop the cell cycle at the appropriate time? Recent studies in worms and yeasts may provide clues to the answer. Regulated protein degradation plays a crucial part in controlling the cell cycle. In particular, the concentrations of a number of cell-cycle control proteins are regulated by ubiquitin-dependent protein degradation in proteasomes (3Deshaies R.J Curr. Opin. Cell Biol. 1995; 7: 781-789Crossref Scopus (54) Google Scholar). The cyclins were the first and still most striking examples, being rapidly degraded after they have functioned, but other cell-cycle proteins can also be regulated by changes in their degradation. The increase in p27 protein seen in some cultured cells in response to growth factor withdrawal, for example, seems to result mainly from a decrease in ubiquitin-dependent degradation (16Pagano M Tam S.W Theodoras A.M Beer-Romero P Del Sal G Chau V Yew P Draetta G.F Rolfe M Science. 1995; 269: 682-685Crossref PubMed Scopus (1706) Google Scholar). A study reported recently by Hedgecock and his colleagues (9Kipreos E.T Lauder L.E Wing J.P He W.W Hedgecock E.M Cell. 1996; 85: 829-839Abstract Full Text Full Text PDF PubMed Scopus (379) Google Scholar) directly links the regulation of protein degradation to the control of cell numbers in development. Null mutations in the C. elegans gene cul-1 result in a phenotype remarkably similar to that of the p27−/− mice: cell proliferation is increased by 1 to 3 divisions in all larval cell lineages, resulting in an increase in cell numbers and a decrease in cell size. As in p27−/− mice, cell determination and differentiation mostly occur normally, but in this case there are a number of ectopic cell deaths. Molecular cloning and DNA-sequencing analyses suggest that cul-1 encodes a large cytosolic protein. An indication of how the CUL-1 protein may normally act to regulate cell numbers comes from genetic and biochemical studies in the budding yeast S. cerevisiae on a CUL-1 homolog called Cdc53. Cdc53, which is required for cell cycle progression from G1 into S phase, is thought to act as an E3 ubiquitin-protein ligase that helps direct the ubiquitination enzymatic cascade to G1 cyclins (and probably some other cell-cycle control proteins), thereby targeting them for destruction (20Willems A.R Lanker S Patton E.F Craig K.L Nason T.F Kobayashi R Wittenberg C Tyers M Cell. 1996; 86 (in press)Abstract Full Text Full Text PDF Scopus (267) Google Scholar). Presumably, other members of the cullin family, including CUL-1, perform similar functions in other eukaryotes, and Hedgecock and colleagues have identified five cullins in the nematode , six in humans, and three in budding yeast. The finding that inactivation of cul-1 prolongs cell proliferation is consistent with the notion that the CUL-1 protein is normally involved in the degradation of a positive cell-cycle regulator such as a G1 cyclin. Failure to degrade the regulator would be functionally equivalent to inactivating a G1 Cdk inhibitor such as p27, which would explain the similar phenotypes of p27-deficient mice and CUL-1-deficient worms. Taken together, these studies in mice, worms, and yeasts suggest that both the accumulation of negative regulators and the degradation of positive regulators may be required for developing animal cells to exit normally from the cell cycle. With these clues in hand, the challenge now is to identify the relevant cell-cycle regulators in different cell lineages and to determine how their levels and activities are controlled and how they collaborate to stop cell division at the right time. When we have this information for some representative human and mouse cell lineages, we may begin to understand why we are larger than mice.