- Email: [email protected]
Organizing microtubules in the cytoplasm
Cell,
Vol.
22, 331-332,
November
1980
(Part
2). Copyright
0 1980
by MIT
Organizing Microtubules in the Cytoplasm
Frank Solomon Center for Cancer Research Massachusetts Institute of Technology Cambridge, Massachusetts 02139
The pattern of cytoplasmic microtubules in several sorts of animal cells is now familiar to us. It has been made accessible by immunofluorescence staining with antibodies directed against tubulin, the major microtubule component. The typical pattern we see is a lacy net of fibers, generally but not exclusively radial, extending to the edges of the cell but filling little of its space. There is nothing in these complex meshworks which is reminiscent of the regular geometry and dense packing characteristic of other microtubule-containing organelles such as the centrioles or cilia. It has long been understood that the distribution of cytoplasmic microtubules is an essential element in establishing cellular asymmetry. Of late, attempts to understand the structural basis for that distribution have become an important part of studies of microtubule function. In a number of other situations such as formation of mitotic spindles, specific structures have been identified that apparently nucleate the assembly of mtcrotubules. Reasoning by analogy to those situations, considerable attention has focused on attempts to demonstrate discrete centers in the cytoplasm from which microtubules might grow. Experiments have been performed to determine how many such microtubule-organizing centers (MTOCs) there are in a cell, what they are made of and precisely what they do. As we shall see, the MTOCs in the cytoplasm are frequently associated with the “centrosome,” the centriole and the electron-dense material which surrounds it. It is important to point out that there is considerable diversity among the structures which organize microtubules. In particular, the concept of an MTOC was formulated by Picket&Heaps, from studies of plant cell mitotic spindles (Ann. N.Y. Acad. Sci. 253, 352-361, 1975). The poles of those spindles contain dense material but no centrioles. Even in the poles of the animal cell spindles, which do contain centrioles, the microtubules associate more closely with the dense pericentriolar material. The work on mitotic MTOCs, which requires an extensive review of its own, shows that the significance of the relationship between centrioles and MTOCs is not yet clear. In particularly favorable situations, like the fish melanophore, putative cytoplasmic MTOCs can readily be visualized. The pigment granules of melanophores can be aggregated close to the center of the cell or dispersed in the peripheral cytoplasm. In the dispersed state, many microtubules radiate from a clearly demarcated central apparatus-tufts of dense, amor-
phous material surrounding the centriole. When the granules aggregate, both the number of microtubules and the number of tufts are greatly reduced (Schliwa et al., JCB 83, 623-632, 1979). As in the mitotic spindle, the ends of the microtubules are closer to the pericentriolar material than to the centriole. Evidence of MTOCs in interphase fibroblasts comes from immunofluorescent experiments. Cells are examined at early stages of microtubule assembly: in newly formed daughter cells (Brinkley et al., PNAS 72, 4981-4985, 1975) or during recovery from microtubule depolymerizing drugs (Osborn and Weber. PNAS 73, 867-871, 1976). In general, one point of origin near the nucleus, and occasionally near the primary cilium of one centriole, is seen. These characteristics of the origin lead to the suggestion that it is a centrosome, but high fluorescent backgrounds in those preparations could obscure other MTOCs. That problem has been dealt with by extracting cells with nonionic detergents before fixation, under conditions that preserve microtubules but release the unassembled tubulin and most other cellular protein (Osborn and Weber, Cell 72, 561-571, 1977). When this procedure is applied to cells recovering from microtubule-depolymerizing drugs, clear images with greatly reduced background are produced. Spiegelman et al. (Cell 16, 239-252, 1979) survey several cell lines, and in many cases they find microtubules growing from multiple sites in each cell. For example, in mouse fibroblasts they describe an average of 8 MTOCs per cell, and as many as 20 in one cell. These MTOCs are clearly the primary origin of the forming microtubules. However, microtubules are seen which have no connection with any center. It is not known whether these microtubules in fact arise independently of any MTOC or are fragments produced by the extraction procedure. On its face, the finding of multiple MTOCs in cells suggests that they are not centrosomes, since there should not be multiple centrioles in a normal cell. But subsequent studies raise doubts about the exact number of MTOCs. Brinkley et al. [In Microtubules and Microtubule Inhibitors, DeBrabender and DeMey, eds., (Amsterdam: Elsevier-North Holland), in press, 19801 using a similar procedure, find only l-2 sites per cell. They have also determined the number of MTOCs as a function of cell cycle stage, using synchronized populations of CHO cells. The MTOCs per cell increase from 1 to 2 by the same time course as centriole replication occurs. This correlation is strengthened by the work of Sharp et al. (J. Cell Sci., in press, 1980) who fix cells under conditions compatible with both immunofluorescence and electron microscopy. They demonstrate that each MTOC detected by fluorescence in a cell with several MTOCs is associated with a centriole. The implication is that
Cd 332
cells with multiple MTOCs have multiple centrioles. As noted above, the equation of an MTOC with a centriole is not obligatory, and whether it explains all the findings of multiple MTOCs (Spiegelman et al., op. cit.; Marchisio et al., Eur. J. Cell Biol. 20, 45-50, 1979; Watt and Harris, J. Cell Sci. 44, 103-121, 1980) remains to be seen. Attempts to identify the molecular components of the MTOC have begun. That investigation is facilitated by the finding that cytoplasmic MTOCs can function in vitro (Brinkley et al., op. cit.). The MTOCs are exposed by gentle detergent extraction of cells pretreated with microtubule-depolymerizing drugs. When these preparations are incubated with tubulin, under conditions where self-assembly does not occur, microtubules form from discrete centers. Using the ability to seed assembly in vitro as an assay, Brinkley et al. (op. cit.) and Ring et al. [In Microtubules and Microtubule Inhibitors, DeBrabender and DeMey, eds, (Amsterdam: Elsevier-North Holland) in press, 19801 have begun to purify MTOCs and to identify their components biochemically and immunologically. A major question that can be addressed in these experiments is the precise nature of the in vitro activity. It could represent true nucleation of microtubule assembly; alternatively it could represent the addition of tubulin onto microtubule remnants that survived the preincubation with drug and are not detectable by microscopic techniques. Along with this work of the nature of MTOCs have come interesting suggestions for how they might control microtubule distribution. One possibility is that the MTOCs not only organize assembly but also specify the number and length of cytoplasmic microtubules. For example, the ability of the fish melanophore central apparatus described above (Schliwa et al., op. cit.) to induce microtubule assembly in vitro depends on the state of the cell from which it was isolated. The apparatus from cells with dispersed pigment and more microtubules nucleates more assembly than does the apparatus from cells with aggregated pigment. Brinkley et al. (op. cit.) now report that MTOCs from 3T3 cells and SV3T3 cells nucleate different numbers of microtubules, both in vivo during recovery from depolymerizing drugs and in the in vitro assay using exogenous tubulin. In addition, the average length of the microtubules generated in vitro is different from MTOCs from the two cell types. There are, of course, many examples of cellular specification of unique and exotic microtubule arrays which require control of microtubule number and length, and even arrange-
ment. But it is not clear that all that control comes from a structure at one end. Similarly in the two in vitro systems cited here, the MTOCs are not purified and the observed activities may be determined by other components of the cellular preparations. An alternative proposal is advanced by Kirschner (JCB. 86, 330-334, 19801, based on recent studies of the kinetic behavior in vitro of the microtubule polymer. Microtubules are polar structures, and the two dissimilar ends have different equilibrium constants; at steady state, the values of those equilibrium constants define a primary assembly end and a primary disassembly end (Margolis and Wilson, Cell 13, l-8, 1978). The absolute values of the rate constants for the on and off sections have also been determined, and they are both greater at the primary assembly end (Bergen and Borisy, JCB 84, 141-l 50, 1980). A mathematical treatment of such a situation, and an explanation of this apparent violation of microscopic reversibility, are provided in a previous paper by Wegner (JMB 708, 139-l 50, 1976). In his model, Kirschner points out the consequences of anchoring the primary disassembly end of a cellular microtubule, possibly in an MTOC, and thus sequestering it from reaction with the pool of tubulin subunit. Under these conditions, the specific rate constants define a concentration range high enough to permit growth of the microtubules at the primary assembly end, but low enough to cause depolymerization at any primary disassembly end which is unanchored. In effect, the spontaneous formation of a microtubule free in the cytoplasm would be suppressed; all would be constrained to grow from an organizing center which would in effect determine their distribution. One prediction of this model is that unanchored cytoplasmic microtubules should be extremely unstable, and therefore rare. That prediction is testable. There are situations where cytoplasmic microtubules apparently do not insert into a common, conspicuously dense structurein the marginal band of avian and amphibian red blood cells, in the axon of a nematode neuron (Chalfie and Thompson, JCB 82, 278289, 1979) and in fragments of fibroblast cytoplasm (Albrecht-Buhler, PNAS in press, 1980). In cases like these, each microtubule may have its own organizing center or capping structure to satisfy the requirements of thermodynamics. Its distribution might be specified by other elements that interact with the microtubule along its length. Based on recent progress, dissecting out these various influences now seems possible.
Vol.
22, 331-332,
November
1980
(Part
2). Copyright
0 1980
by MIT
Organizing Microtubules in the Cytoplasm
Frank Solomon Center for Cancer Research Massachusetts Institute of Technology Cambridge, Massachusetts 02139
The pattern of cytoplasmic microtubules in several sorts of animal cells is now familiar to us. It has been made accessible by immunofluorescence staining with antibodies directed against tubulin, the major microtubule component. The typical pattern we see is a lacy net of fibers, generally but not exclusively radial, extending to the edges of the cell but filling little of its space. There is nothing in these complex meshworks which is reminiscent of the regular geometry and dense packing characteristic of other microtubule-containing organelles such as the centrioles or cilia. It has long been understood that the distribution of cytoplasmic microtubules is an essential element in establishing cellular asymmetry. Of late, attempts to understand the structural basis for that distribution have become an important part of studies of microtubule function. In a number of other situations such as formation of mitotic spindles, specific structures have been identified that apparently nucleate the assembly of mtcrotubules. Reasoning by analogy to those situations, considerable attention has focused on attempts to demonstrate discrete centers in the cytoplasm from which microtubules might grow. Experiments have been performed to determine how many such microtubule-organizing centers (MTOCs) there are in a cell, what they are made of and precisely what they do. As we shall see, the MTOCs in the cytoplasm are frequently associated with the “centrosome,” the centriole and the electron-dense material which surrounds it. It is important to point out that there is considerable diversity among the structures which organize microtubules. In particular, the concept of an MTOC was formulated by Picket&Heaps, from studies of plant cell mitotic spindles (Ann. N.Y. Acad. Sci. 253, 352-361, 1975). The poles of those spindles contain dense material but no centrioles. Even in the poles of the animal cell spindles, which do contain centrioles, the microtubules associate more closely with the dense pericentriolar material. The work on mitotic MTOCs, which requires an extensive review of its own, shows that the significance of the relationship between centrioles and MTOCs is not yet clear. In particularly favorable situations, like the fish melanophore, putative cytoplasmic MTOCs can readily be visualized. The pigment granules of melanophores can be aggregated close to the center of the cell or dispersed in the peripheral cytoplasm. In the dispersed state, many microtubules radiate from a clearly demarcated central apparatus-tufts of dense, amor-
phous material surrounding the centriole. When the granules aggregate, both the number of microtubules and the number of tufts are greatly reduced (Schliwa et al., JCB 83, 623-632, 1979). As in the mitotic spindle, the ends of the microtubules are closer to the pericentriolar material than to the centriole. Evidence of MTOCs in interphase fibroblasts comes from immunofluorescent experiments. Cells are examined at early stages of microtubule assembly: in newly formed daughter cells (Brinkley et al., PNAS 72, 4981-4985, 1975) or during recovery from microtubule depolymerizing drugs (Osborn and Weber. PNAS 73, 867-871, 1976). In general, one point of origin near the nucleus, and occasionally near the primary cilium of one centriole, is seen. These characteristics of the origin lead to the suggestion that it is a centrosome, but high fluorescent backgrounds in those preparations could obscure other MTOCs. That problem has been dealt with by extracting cells with nonionic detergents before fixation, under conditions that preserve microtubules but release the unassembled tubulin and most other cellular protein (Osborn and Weber, Cell 72, 561-571, 1977). When this procedure is applied to cells recovering from microtubule-depolymerizing drugs, clear images with greatly reduced background are produced. Spiegelman et al. (Cell 16, 239-252, 1979) survey several cell lines, and in many cases they find microtubules growing from multiple sites in each cell. For example, in mouse fibroblasts they describe an average of 8 MTOCs per cell, and as many as 20 in one cell. These MTOCs are clearly the primary origin of the forming microtubules. However, microtubules are seen which have no connection with any center. It is not known whether these microtubules in fact arise independently of any MTOC or are fragments produced by the extraction procedure. On its face, the finding of multiple MTOCs in cells suggests that they are not centrosomes, since there should not be multiple centrioles in a normal cell. But subsequent studies raise doubts about the exact number of MTOCs. Brinkley et al. [In Microtubules and Microtubule Inhibitors, DeBrabender and DeMey, eds., (Amsterdam: Elsevier-North Holland), in press, 19801 using a similar procedure, find only l-2 sites per cell. They have also determined the number of MTOCs as a function of cell cycle stage, using synchronized populations of CHO cells. The MTOCs per cell increase from 1 to 2 by the same time course as centriole replication occurs. This correlation is strengthened by the work of Sharp et al. (J. Cell Sci., in press, 1980) who fix cells under conditions compatible with both immunofluorescence and electron microscopy. They demonstrate that each MTOC detected by fluorescence in a cell with several MTOCs is associated with a centriole. The implication is that
Cd 332
cells with multiple MTOCs have multiple centrioles. As noted above, the equation of an MTOC with a centriole is not obligatory, and whether it explains all the findings of multiple MTOCs (Spiegelman et al., op. cit.; Marchisio et al., Eur. J. Cell Biol. 20, 45-50, 1979; Watt and Harris, J. Cell Sci. 44, 103-121, 1980) remains to be seen. Attempts to identify the molecular components of the MTOC have begun. That investigation is facilitated by the finding that cytoplasmic MTOCs can function in vitro (Brinkley et al., op. cit.). The MTOCs are exposed by gentle detergent extraction of cells pretreated with microtubule-depolymerizing drugs. When these preparations are incubated with tubulin, under conditions where self-assembly does not occur, microtubules form from discrete centers. Using the ability to seed assembly in vitro as an assay, Brinkley et al. (op. cit.) and Ring et al. [In Microtubules and Microtubule Inhibitors, DeBrabender and DeMey, eds, (Amsterdam: Elsevier-North Holland) in press, 19801 have begun to purify MTOCs and to identify their components biochemically and immunologically. A major question that can be addressed in these experiments is the precise nature of the in vitro activity. It could represent true nucleation of microtubule assembly; alternatively it could represent the addition of tubulin onto microtubule remnants that survived the preincubation with drug and are not detectable by microscopic techniques. Along with this work of the nature of MTOCs have come interesting suggestions for how they might control microtubule distribution. One possibility is that the MTOCs not only organize assembly but also specify the number and length of cytoplasmic microtubules. For example, the ability of the fish melanophore central apparatus described above (Schliwa et al., op. cit.) to induce microtubule assembly in vitro depends on the state of the cell from which it was isolated. The apparatus from cells with dispersed pigment and more microtubules nucleates more assembly than does the apparatus from cells with aggregated pigment. Brinkley et al. (op. cit.) now report that MTOCs from 3T3 cells and SV3T3 cells nucleate different numbers of microtubules, both in vivo during recovery from depolymerizing drugs and in the in vitro assay using exogenous tubulin. In addition, the average length of the microtubules generated in vitro is different from MTOCs from the two cell types. There are, of course, many examples of cellular specification of unique and exotic microtubule arrays which require control of microtubule number and length, and even arrange-
ment. But it is not clear that all that control comes from a structure at one end. Similarly in the two in vitro systems cited here, the MTOCs are not purified and the observed activities may be determined by other components of the cellular preparations. An alternative proposal is advanced by Kirschner (JCB. 86, 330-334, 19801, based on recent studies of the kinetic behavior in vitro of the microtubule polymer. Microtubules are polar structures, and the two dissimilar ends have different equilibrium constants; at steady state, the values of those equilibrium constants define a primary assembly end and a primary disassembly end (Margolis and Wilson, Cell 13, l-8, 1978). The absolute values of the rate constants for the on and off sections have also been determined, and they are both greater at the primary assembly end (Bergen and Borisy, JCB 84, 141-l 50, 1980). A mathematical treatment of such a situation, and an explanation of this apparent violation of microscopic reversibility, are provided in a previous paper by Wegner (JMB 708, 139-l 50, 1976). In his model, Kirschner points out the consequences of anchoring the primary disassembly end of a cellular microtubule, possibly in an MTOC, and thus sequestering it from reaction with the pool of tubulin subunit. Under these conditions, the specific rate constants define a concentration range high enough to permit growth of the microtubules at the primary assembly end, but low enough to cause depolymerization at any primary disassembly end which is unanchored. In effect, the spontaneous formation of a microtubule free in the cytoplasm would be suppressed; all would be constrained to grow from an organizing center which would in effect determine their distribution. One prediction of this model is that unanchored cytoplasmic microtubules should be extremely unstable, and therefore rare. That prediction is testable. There are situations where cytoplasmic microtubules apparently do not insert into a common, conspicuously dense structurein the marginal band of avian and amphibian red blood cells, in the axon of a nematode neuron (Chalfie and Thompson, JCB 82, 278289, 1979) and in fragments of fibroblast cytoplasm (Albrecht-Buhler, PNAS in press, 1980). In cases like these, each microtubule may have its own organizing center or capping structure to satisfy the requirements of thermodynamics. Its distribution might be specified by other elements that interact with the microtubule along its length. Based on recent progress, dissecting out these various influences now seems possible.