INTRODUCTION

Top Previous Next References

The cellular basis of life requires cytokinesis after DNA replication and the distribution of the genetic material over two new daughter cells (mitosis). To ensure that these daughter cells are genetically identical, nuclear division and cytokinesis must be well organized with respect to each other in time and space. This comprises cellular logistics in prokaryotes as well as eukaryotes, and it is this aspect of cytokinesis which is the subject of this review.

This review consists of three parts. First, I will present a brief overview of cytokinesis on the cellular and subcellular levels. At the same time, I will highlight common problems to be solved by the two groups of organisms. This will then serve as an introduction to the (macro)molecular approach, which forms the second part of this review. It will be seen that the distinction between the cellular and macromolecular levels cannot always be consequently maintained. The third part discusses the eukaryotic and prokaryotic solutions to the distribution of genetic material. In this respect it will become clear that there are essential differences between prokaryotes and eukaryotes. Of course, one can choose to emphasize or play down the differences. In this review I will make the point that these differences are important and that recent results as discussed here confirm this point of view.

Present research on cytokinesis is rapidly advancing, and several model systems are being used. Animal systems include mammals, Xenopus laevis, Drosophila melanogaster, and Caenorhabditis elegans. Somewhat resembling these systems is the slime mold Dictyostelium discoidium. Yeasts and fungi represent systems where the flexibility of cellular shape is limited by a rigid, though not static, cell wall. These organisms also are distinguished from animal systems by maintaining the nuclear envelope during mitosis. The prokaryotic model organisms are Escherichia coli, Bacillus subtilis, and Caulobacter crescentus. Higher plants will not be discussed here, because they occupy a somewhat different position.

CYTOKINESIS AT THE CELLULAR AND SUBCELLULAR LEVELS

Top Previous Next References

Organization of the Bipolar Spindle Apparatus

A precondition for the cell division process is the bipolar organization of the cell. The first step toward attaining this configuration is the duplication of the centrosome or microtubule-organizing center (MTOC). In higher eukaryotes, but also in Dictyostelium, the MTOC is located near the nuclear envelope. In yeasts like Saccharomyces cerevisiae and Schizosaccharomyces pombe the MTOC is referred to as spindle pole body (SPB). Their respective SPBs are quite different in structure and composition. In S. cerevisiae the SPB is part of the nuclear envelope, whereas in S. pombe it is located in the cytoplasm during interphase and enters the nuclear envelope at the start of mitosis. The various MTOCs all contain -tubulin as part of a microtubule-nucleating apparatus that facilitates radiation of microtubules into the cytoplasm (for a review, see reference 86).

Centrosomes and the SPB of S. pombe duplicate at the beginning of mitosis; in S. cerevisiae duplication occurs at the beginning of S phase (for reviews, see references 6 and 37). After duplication, one of the MTOCs moves to a diametrically opposite position in the cell. The early SPB duplication in S. cerevisiae with respect to S. pombe, might be required for the movement of the nucleus into the growing bud (see below).

At the onset of mitosis, microtubules become arranged between the MTOCs. Because the nuclear envelope disintegrates during mitosis in higher eukaryotes, the inter-MTOC microtubules are in direct contact with the cytoplasm. By contrast, the nuclear envelope in yeasts persists, thus preventing intracellular contacts of the inter-MTOC (SPB) microtubules beyond the nuclear envelope (unless the nuclear pores serve a communicating function [see below]). Despite minor differences, the essential elements of a bipolar spindle organization, here referred to as the bipolar spindle apparatus (BSA), are shared among all eukaryotes.

In most eukaryotes the division plane becomes located halfway between the MTOCs of the BSA. Displacement of the BSA by centrifugation in Crepidula eggs was accompanied by displacement of the cleavage plane (Fig. 1a) (reference 91 and references therein). This classic experiment by Conklin has emphasized the potential importance of the cellular envelope region halfway between the MTOCs. Subsequent research has also revealed the importance of astral microtubule-mediated interaction with the nearest cell cortex for BSA positioning in the cell (for a review, see reference 101) (see below).

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Position of the BSA relative to the site of cytokinesis. (a) Animal cell. In the right-hand cell, the BSA has been shifted to the periphery by pressing a glass bead onto the mitotic cell. (b) S. cerevisiae. The elongating nucleus positions itself relative to the neck of the growing bud (thick double arrows). (c) S. pombe. Medial ring positioning follows the predetermined site of the nucleus. MT, microtubules.

In S. cerevisiae, division site selection is a direct consequence of bud site selection (14). Thus, to place the predetermined division plane halfway along the BSA, the nucleus itself has to be moved in a proper orientation in the cell (13) (Fig. 1b). This leads to the question of how the mitotic nucleus becomes properly oriented in a budding cell. A likely mediator is the SPB because it connects the inner and outer surface of the nuclear envelope. Inside the nucleus, the mitotic spindle elongates between the SPBs. Outside the nucleus, astral microtubules radiate to the opposite plasma membrane in the growing bud. Notably, in S. cerevisiae the interaction of astral microtubules with the opposing plasma membrane has been demonstrated (107). Eventually, the BSA orients toward a preexisting septum marker, the bud site (Fig. 1b).

By contrast, in S. pombe the division site must be established in relation to the cellular position of the nucleus (Fig. 1c) (13). To do this, a mechanism can be expected which maintains the nucleus in the cell center. Presumably this is achieved, at least in interphase, through the microtubular framework which surrounds the nucleus (38). It can be imagined that this framework also interacts with the plasma membrane, thus positioning the nucleus at a defined position in the cell.

In essence, this representation of the two yeasts is not different from that of a cleaving higher eukaryotic cell. In all cases, the final spatial orientation is the same: a defined orientation of the division plane perpendicular to the BSA. What differs is the way in which this orientation is achieved.

Early light micoscopic data have shown that the E. coli division plane is positioned between segregated chromosomes and halfway between the cell poles (70, 104). DNA compaction after completion of replication, as generally occurs during eukaryotic mitosis, has so far not been demonstrated in bacteria. In fact, there is no need for such a phase because, as demonstrated previously (104) and studied in E. coli in particular, the replicating chromosome becomes distributed over the prospective daughter cells during elongation of the cell (for reviews, see references 32, 103, and 125). E. coli therefore does not have a separate M phase. This is depicted schematically in Fig. 2, where the prokaryotic and eukaryotic cell cycle phases are compared.

View larger version (36K): [in this window] [in a new window]   FIG. 2.   Cell cycle phases in a prokaryote and in a eukaryote. In a prokaryote, the bacterial chromosome does not condense prior to segregation. The chromosome separates during its replication. Therefore, the S phase coincides with the M phase. G1, cell cycle phase prior to DNA replication. In E. coli, this phase is denoted B. Its occurrence depends on the strain and on growth conditions. G2, phase between termination of DNA replication and mitosis. This phase is not present in E. coli. S, DNA replication phase.

CYTOKINESIS AT THE MACROMOLECULAR LEVEL

Top Previous Next References

Cytokinesis and the Cytoskeleton

Cytokinetic ring. Cytokinesis in eukaryotes is achieved by a contractile ring associated with the plasma membrane. The ring basically consists of actin and myosin II, and the actomyosin organization resembles that of smooth muscle (for a review, see reference 101). In nondividing cells this ring is absent, and its establishment involves extensive spatial reorganization of the actin filament-based cytoskeleton. Apart from the contractile ring and its association with the cell cortex, the microtubules and the microtubule-associated proteins in the spindle interzone have been invoked as instrumental in the cytokinetic process (for reviews, see references 28 and 101). It is not my aim to compare in detail the cytokinetic process in different eukaryotes; rather, I will attempt to provide an overall "eukaryotic picture." Doubtless this picture will be basic. However, by focusing on the main aspects, a comparison with prokaryotic cytokinesis will be facilitated. It will be appreciated that the circumference of the cytokinetic ring of E. coli is considerably smaller than its eukaryotic counterpart. Even in a small eukaryote like D. discoidium, there is a 100-fold difference (Fig. 3). In addition, the macromolecular composition of the E. coli cytokinetic ring is essentially different.

View larger version (104K): [in this window] [in a new window]   FIG. 3.   Cytokinetic rings in a eukaryote and a prokaryote. In the highly schematized eukaryote, the cytokinetic ring is represented by a vertical white bar. In the prokaryote, E. coli, a cell division protein, FtsQ, has been labeled by its fusion to green fluorescent protein (GFP). Note the difference in the sizes of the two cells. (Copyright T. den Blaauwen and J. Chen.)

Microtubules and SPBs. Of the three classes of spindle microtubules, namely, kinetochore-bound microtubules, centrosome-connecting microtubules, and astral microtubules, the last, because of their interaction with the cortex, have received the most attention recently. This interaction makes the cortex perhaps a prime determinant of the bipolar organization of the cell. Two examples, in S. cerevisiae and S. pombe, underline this.

In S. cerevisiae, elegant in vivo studies have been carried out with microtubules containing a green fluorescent protein GFP--tubulin fusion protein (

View larger version (23K): [in this window] [in a new window]   FIG. 4.   Interaction between microtubules and growing poles in S. cerevisiae and S. pombe. (a) In S. cerevisiae, astral microtubles explore the poles while the nucleus moves into the bud. The motor protein dynein travels along microtubules and pulls at the dynactin molecules at the cell cortex. Reprinted from reference 12 with permission of the publisher. (b) Growing microtubules deposit Tea1p at the poles. Adapted from reference 69, with permission of the publisher, MT, microtubules.

Microtubule-associated motor proteins. It is well known that microtubules serve as roadways for plus-end-directed (kinesins) and minus-end-directed (dyneins) motor proteins, thus allowing a polarized arrangement and transport of cellular components. Astral microtubules presumably interact with the cortex in animal cells with the aid of minus-end-directed proteins (for reviews, see references 2 and 61). As mentioned above, such interaction has also been implied for the astral microtubules in the the growing bud of S. cerevisiae (Fig. 4) (107).

Small GTPases and the cellular envelope. Small GTP-binding proteins and heterotrimeric G proteins represent molecular switches that influence numerous processes related to the dynamic cellular organization (2, 61). The monomeric small GTPases include subfamilies such as Ras and Rho (41). Their functioning as switches is based on general principles. Basically, the switch involves an alternation between two conformational states which are defined by the binding of either GTP or GDP. Nucleotide binding is affected by activators, effectors, and inhibitors, which in turn are influenced by additional components. In the example shown in Fig. 5 ("Ras protein paradigma" [41]), the GTP-bound form activates an effector. Inactivation occurs through hydrolysis of GTP, resulting in the inactive GDP-bound form. In vivo hydrolysis is controlled by GTPase-activating proteins (GAPs). The GTP-bound form arises through a GDP-GTP exchange reaction, which is catalyzed by a guanine nucleotide exchange factor.

View larger version (12K): [in this window] [in a new window]   FIG. 5.   The Rho-GTPase molecular switch. The conformation of the GTPase depends on whether GTP or GDP is bound. The GTP-bound form functions through an effector. This interaction is abolished when GTP is hydrolyzed with the aid of GAP. A guanine nucleotide exchange factor (GEF) catalyzes the production of active Rho-GTP. Adapted from reference 41.

IQGAPs. A cytokinetic role of IQGAP-like proteins has been demonstrated so far for D. discoideum, S. cerevisiae, and S. pombe. These proteins were first discovered in humans (82, 120) and were first considered to be GAPs (for a recent review, see reference 65). IQGAPs contain a number of domains (Fig. 6), which include CH (calponin homology domain which putatively binds to actin), IQ (calmodulin-binding motifs) and GRD (GAP-related domain) domains. IQGAP-like proteins have been found in other organisms, but the extent to which the different domains have been retained varies considerably (65). Whereas the D. discoideum IQGAP possesses only the IQ and GRD domains, the S. cerevisiae and S. pombe IQGAP-like proteins resemble the human protein in organization.

View larger version (44K): [in this window] [in a new window]   FIG. 6.   Domain composition and homology of IQGAP-related proteins from various organisms. The percentages refer to the conservation of domain composition compared to mammalian IQGAP1. CH, calponin homology domain; IR, internal repeats; WW, SH3-mimicking domain; IQ, calmodulin binding motifs; GRD, RasGAP-homology domain. For further details, see reference 65. Adapted from reference 65 with permission of the publisher.

Cytokinesis in E. coli. Traditionally, cytokinesis has been termed cell division in prokaryotes. In E. coli, division is basically a constriction process, whereas in B. subtilis, there is an ingrowing septum like that in S. pombe. Constriction in E. coli involves three envelope layers, which are, from outside to inside, the outer membrane, the peptidoglycan layer, and the cell membrane. The peptidoglycan layer is embedded in the so-called periplasm and serves as an exoskeleton to the bacterial cell. Its local synthesis at the site of constriction is indispensable for the division process, as are cytoplasmic polymers (for reviews, see references 43 and 76) (see below).

View larger version (46K): [in this window] [in a new window]   FIG. 7.   Topology of E. coli cell division proteins relative to the plasma membrane. The numbers in the molecules reflect the molecular masses in kilodaltons. With the exception of FtsK and FtsW, the membrane-bound proteins contain a single membrane anchor. The way in which the various proteins interact with each other is not known. pm, plasma membrane; pg, peptidoglycan. Reprinted from reference 76a with permission of the publisher.

View larger version (42K): [in this window] [in a new window]   FIG. 8.   Schematic representation of the FtsZ ring and its associated subassemblies. The subassemblies are supposed to contain proteins involved in peptidoglycan synthesis in addition to the cell division proteins depicted in Fig. 7. Reprinted from reference 76 with permission of the publisher.

Bipolar organization of the bacterial replicating chromosome. Mitosis in eukaryotes is an active process, encompassing dynamic cytoskeletal structures and various motor proteins. It even can proceed in a cell-free system (2, 61). By contrast, the small size of a bacterium (Fig. 3) has precluded a description of its partitioning apparatus in macromolecular terms. It also explains why progress in this field has been made only very recently (see below). It should be remembered that E. coli as well as B. subtilis has a circular genome which replicates bidirectionally from a fixed origin of replication toward the so-called terminus diametrically opposite the origin. This situation is quite different from that of eukaryotes, where there are many origins per chromosome and where defined termini are probably not present (Fig. 9). So far, for prokaryotes a dominant view has been that the interaction of the bacterial chromosome with the elongating cell envelope is instrumental in separating newly replicated chromosomal domains.

View larger version (28K): [in this window] [in a new window]   FIG. 9.   Replicating eukaryotic and prokaryotic (E. coli) chromosomes. A eukaryotic chromosome contains numerous bidirectional replicating units, each containing its own origin of replication. E. coli replication starts at a defined origin and proceeds bidirectionally. An E. coli chromosome does not contain a centromere or telomeres.

Chromosome segregation. Once decatenated, how do the duplicated chromosomes assume their new positions in the prospective daughter cells and how does FtsZ ring formation correlate with bacterial chromosome partitioning? It has been found that, within experimental error, FtsZ-ring formation in E. coli coincides with termination of DNA replication (16). Because the nucleoid borders (115) as well as the duplicated oriC origins (78, 97) maintain a constant distance from the cell poles in an elongating cell, the question can be asked whether the replicated termini become pushed apart by an active mechanism. Conceptually, one could invoke cytoplasmic components and/or the invaginating cell envelope. For FtsZ, the indications for a role in chromosome partitioning are twofold. First, a delay in nucleoid segregation has been inferred in FtsZ-depleted cells (113), and second, nucleoid segregation appeared affected in an ftsZ temperature-sensitive mutant at the nonpermissive temperature (45; C. L. Woldringh, personal communication). Evidence for an envelope role also arose from the interesting observation that separation of the replicated chromosomes is delayed in an ftsI temperature-sensitive mutant at the nonpermissive temperature (45, 110) or after impairment of FtsI (PBP3)-specific peptidoglycan synthesis by specific penicillins (45). This suggests that division-specific envelope synthesis contributes to the partitioning of daughter chromosomes. Perhaps the above-mentioned role of FtsZ in chromosome partitioning is indirect, because it is plausible that impairment of membrane-associated FtsZ affects division-specific peptidoglycan synthesis.

Does DNA condensation play a role during chromosome segregation? Like eukaryotes, prokaryotes possess structural maintenance of chromosomes (SMC) proteins. In eukaryotes, these proteins are involved in mitotic chromosome condensation and in chromatid cohesion. These activities are carried out with the help of other proteins, and the complexes have been termed condensin and cohesin, respectively. In these multiprotein complexes, different SMCs are present and occur as heterodimers (for reviews, see references 5, 42, and 70).

How does E. coli know where to divide? The simple question of how E. coli knows where to divide has generated much speculation. One view is based on the observation that cytokinesis takes place between duplicated nucleoids when they are sufficiently separated (nucleoid occlusion). Thus, the location of the division plane is directly related to the cellular position of the replicating bacterial chromosome (74, 124). Whereas the original nucleoid occlusion model was based on observing cellular constrictions as such, recent studies have confirmed that the positioning of the FtsZ ring is affected by the nucleoid in a similar way (36, 112). The observed coincidence of termination of DNA replication and FtsZ ring formation (16) also fits the model.

View larger version (145K): [in this window] [in a new window]   FIG. 10.   Time-lapse images in seconds (numbers) of GFP-MinD oscillations in living E. coli cells. Fluorescent GFP-MinD appears as bright polar regions. The cells to the right were imaged with differential interference contrast. For details, see reference 94 (reprinted with permission of the publisher).

DIFFERENT SOLUTIONS

Top Previous Next References

What Is the E. coli Equivalent of the Bipolar Spindle Apparatus?

Mitosis in eukaryotic cells requires DNA compaction into discrete chromosomes after the cells have traversed through the S and G₂ cell cycle phases. In prokaryotes, no distinction can be made between interphase and mitosis. DNA segregation goes hand in hand with replication (122, 125, 127). Therefore, prokaryotes have no mitosis in the classical (eukaryotic) sense (Fig. 2). Even so, this does not detract from the concept that cytokinesis in prokaryotes requires a bipolar cellular arrangement of the replicating or replicated chromosome.

How can the macromolecular differences be assessed? I will start with the well-known eukaryotic situation and enumerate some perhaps obvious aspects (2, 61). (i) DNA replication occurs through numerous origins (Fig. 9). (ii) After completion of DNA replication, the sister chromosomes (chromatids) stay together until mitotic anaphase. (iii) The replication origins play no role in chromatid separation. (iv) Chromatid separation is preceded by integral chromosome compaction. (v) A bipolar spindle apparatus is formed in between duplicated MTOCs. (vi) Sister chromosome separation requires the dissolution of a so-called cohesin complex (reference 117 and references therein) (vii) Chromatid segregation is effected by shortening of microtubules between the kinetochore and MTOC. (viii) The kinetochore binds to specific nucleotide sequences (centromere).

I will now list, with some generalization, the prokaryotic solution. (i) DNA replication occurs through one origin only (Fig. 9). (ii) The bacterial "chromatids" do not stay together; they move apart during the DNA replication process. (iii) Nucleotide sequences near the origin of replication provide binding sites for proteins that may assist in DNA separation, i.e., in maintaining a constant distance of the origin to the cell pole (note that B. subtilis is an exception [105]). (iv) No microtubules or kinetochores are present in bacteria. (v) Apart from SMC-like proteins, classical motor proteins resembling myosin, dynein, and kinesin have not yet been detected in prokaryotes.

As suggested above, one can consider the origin of replication and/or associated proteins to be the eukaryotic centromere analogue (reference 33, and references therein). The prokaryotic terminus of DNA replication would then be the equivalent of the telomere. In a general mechanistic sense, this appears plausible. However, in macromolecular terms, this comparison does not hold. Eukaryotic origins of DNA replication play no role in mitosis, and telomeres have no functional relationship to the prokaryotic terminus of replication.

The bacterial analogue of the MTOC is more difficult to define. During mitosis, centromeres (while bound to kinetochores) approach MTOCs. Does the E. coli origin of replication, as a centromere analogue, approach something? Because the average distance of duplicated oriC origins from the cell poles remains constant during most of the DNA replication cycle while the cell elongates (78, 97), the cellular poles might be considered the MTOC analogues. Consequently, according to this reasoning, the E. coli centromere analogue (oriC) and its MTOC analogue (the cellular pole) coincide. However, it has also been shown that individual GFP-labeled gene regions can move independently of cell elongation (reference 33 and see above), which suggests some autonomy with respect to the mobility of subcompartments of the segregating nucleoid. On the other hand, the E. coli nucleoid as a whole seems to separate along with cell elongation (115). Thus, globally, it would seem that the replicating and simultaneaously segregating bacterial chromosome, with its origins directed toward the cell poles, represents the E. coli equivalent of the BSA.

Comparison of E. coli and S. pombe with respect to polarity and envelope extension will also serve to illuminate basic structural differences between prokaryotes and eukaryotes (Fig. 11). Cell wall extension in S. pombe occurs at the poles, either at one or at two poles (for a review, see reference 48). Presumably, building blocks of the growing poles are carried to their destination with the aid of a microtubule-based transport system. As mentioned above, Tea proteins might direct microtubules to their proper polar location. By contrast, E. coli poles are largely inert and envelope extension takes place through intercalation of peptidoglycan precursors in the lateral wall (for reviews, see references 43, and 76). Here, gene products needed for envelope assembly (membrane-anchored PBPs) find their way through coupled transcription-translation-membrane insertion. It might be speculated, that in the latter case the topological arrangement of genes being transcribed is directly related to the pattern of envelope assembly (10, 76) (Fig. 12). Clearly, the spatial separation of transcription and translation by a nuclear envelope precludes this possibility in a eukaryote. In fission yeast, directionality of cell wall assembly is probably aided by a polarized cytoskeleton.

View larger version (31K): [in this window] [in a new window]   FIG. 11.   Comparison of cell wall extension in S. pombe and in E. coli. S. pombe extends at the poles, presumably aided by the polarized microtubular framework. In E. coli, the poles are inert with respect to cell wall extension and cell elongation takes place between the poles. Here the poles are not connected through a cytoskeletal structure in the cytoplasm. A diffusible component oscillates between the poles instead. MT, microtuble. For further details, see the text.

View larger version (49K): [in this window] [in a new window]   FIG. 12.   Cotranslational insertion of membrane proteins in the plasma membrane of E. coli. OM, outer membrane; PM, plasma membrane. (Copyright C. L. Woldringh.)

In S. pombe, microtubules are arranged in such a way that their positive ends point to the respective poles. The presumed structural communication between the poles might thus be mediated through a defined cellular structure (Fig. 4). How does E. coli fit in this scheme?

In the absence of a classical cytoskeleton in E. coli, i.e., one located in the cytoplasm, it is not immediately clear how polar communication is achieved or whether it is at all relevant. However, as mentioned in a previous section, recent evidence has shown that MinC and MinD can oscillate between poles (Fig. 10). It would thus seem that the absence of a cytoskeleton in a small prokaryote is compensated for by an oscillation mechanism presumably based on protein diffusion.

View larger version (54K): [in this window] [in a new window]   FIG. 13.   Schematic model of E. coli nucleoid segregation. The nucleoid is transiently connected to the plasma membrane through transcription loops. The model refers to slowly growing cells, where segregation takes place in the long axis of the cell. Reprinted from reference 127 with permission of the publisher.

Role of small GTPases. In a previous section, membrane-anchored Rho-like GTPases were mentioned as mediators between the eukaryotic cytoskeletal organization and assembly of the cell envelope, be it the extracellular matrix or the cell wall. An example was Rho1p of S. cerevisiae, which is involved in activation of 1,3--glucan synthase (21, 89) and also in the reorganization of the actin cytoskeleton (46). So far, the only known protein of the E. coli divisome exhibiting GTPase activity is FtsZ. However, tubulin-like FtsZ does not belong to the group of Rho-like GTPases.

To allow redistribution of duplicated chromosomes over the new daughter cells, chromosomes become aligned in the center of the BSA during eukaryotic metaphase. In E. coli, as discussed above, this refers to the distribution of genes rather than to that of chromosomes. In the classic model of DNA replication, the polymerase holoenzyme travels around the circular chromosome in two directions. Alternatively, as originally proposed by Sueoka and Quinn (111) and by Dingman (18), the DNA to be replicated passes through a stationary structure or replisome. The term "replisome" has been defined as a supramolecular assembly "with helicases, primosome, and polymerase holoenzyme operating coordinately at the replication fork" (52). Indeed, recent experiments indicate that DNA replication takes place in the cell center (in E. coli [51] and B. subtilis [53, 54]). It should be mentioned that the idea of a local aggregation of replisomes had already gained some popularity in the eukaryotic field some time ago (85). Of course, in this case the fixed replisome has no relevance for the central position of the metaphase plate. However, a central cellular position of the prokaryotic replisome would collect the genes to be replicated in the cell center before being distributed over the new daughter cells. In this sense, the prokaryotic replisome is the analogue of the eukaryotic metaphase plate (Fig. 14).

View larger version (66K): [in this window] [in a new window]   FIG. 14.   Bipolarization in a eukaryote and in a prokaryote. In a eukaryotic cell, the compacted chromosomes are arranged in the metaphase plate, which is symmetrically located between the MTOCs. In a prokaryote like E. coli, the replisome is symmetrically located between the duplicated origins (oriC). In this representation, the oriC region of the chromosome and associated proteins are analogues of the MTOC. However, in E. coli, no distinction can be made between the centromere and MTOC functions. Note also that the chromosomes have completed replication in the metaphase plate whereas it is the replisome, active in DNA replication, that is centrally located.

By contrast, duplication of a eukaryotic genome does not suffice to establish a bipolar cell. Duplication of the MTOC is required as a prelude to mitosis, and it is here that the dynamic cytoskeleton comes to the fore. Eukaryotes have thus created a higher level of organization, where the MTOC has taken over the putative segregating function of the prokaryotic origin of replication and its associated proteins (Fig. 14). As mentioned above, in E. coli DNA strands become distributed over new daughter cells, whereas in eukaryotes this applies to a set of sister chromatids.

Evolution of cytokinesis can be considered on two levels of cellular organization. First, it can be asked whether the eukaryotic cytokinetic ring evolved from its prokaryotic counterpart. Second, it could be asked whether chloroplasts and mitochondria fission in a prokaryotic way. This question arises because according to the endosymbiosis hypothesis, these organelles are of prokaryotic origin. I will start with the latter topic (for reviews, see references 26 and 84).

FtsZ in chloroplasts and mitochondria. Initial attempts to find FtsZ in mitochondria have met with little success. By contrast, in chloroplasts of Arabidopsis thaliana two types of FtsZ proteins have been found. They are nucleus encoded; one of them, FtsZ1, appears to be located in the stroma of the chloroplast, whereas the other one, FtsZ2, occurs the outside of the chloroplast (83). It has been suggested that FtsZ1 would be equivalent to the prokaryotic FtsZ in aiding constriction whereas the external FtsZ2 would carry out a squeezing role (26). In this sense, FtsZ2 would assume a role which in gram-positive and gram-negative bacteria is performed by the ingrowing peptidoglycan (75, 76).

Evolution of FtsZ, ZipA, and FtsA into microtubules, MAPs, and actin filaments, respectively? Until now, little, if anything, was known about the evolution of the eukaryotic cytokeleton. Because of the resemblance of FtsZ to tubulin, it has often been suggested that microtubuli arose from FtsZ (reference 25 and references therein). Interestingly, ZipA (Fig. 7) appears to contain sequence elements in its N-terminal cytoplasmic domain resembling those of eukaryotic microtubule-associated proteins (MAPs) (95). In vitro, ZipA was found to promote bundling of FtsZ protofilaments (95), thus strengthening its MAP resemblance. Whereas FtsZ is present in all prokaryotes studied so far (for a review, see reference 66), ZipA does not seem to occur in gram-positive bacteria and archaebacteria, suggesting that in these cases other proteins might have acquired a similar role (95). Of course, these findings have not yet elucidated how the prokaryotic FtsZ ring evolved into a eukaryotic spindle apparatus for the separation of replicated chromosomes. One clue linking FtsZ to chromosome segregation might be, as mentioned above, that impairment of FtsZ affects the segregation of completed chromosomes in E. coli (45, 113).