Because of the prokaryotic heritage shared by all plastids, bacterial cell division systems (11,12) have been used to guide investigation of the plastid division apparatus. The first bona fide component of the plastid division apparatus was identified through a reverse-genetic approach when an FtsZ-like sequence with similarity to Escherichia coli FtsZ was found in an Arabidopsis expressed sequence tag (EST) collection (13). Escherichia coli FtsZ is a tubulin-like, polymer-forming GTPase (14) that is essential for cell division and forms a ring-shaped structure (the ‘Z-ring’) at the division site (15). FtsZ is thought to be one of the first protein components to arrive at the division site, where it acts as a scaffold for other division proteins (16). It may also provide a contractile force that facilitates cytokinesis (12). FtsZ is conserved among almost all prokaryotes, including the cyanobacteria, and all sequenced cyanobacterial genomes have only one FtsZ gene (17). In photosynthetic eukaryotes, plastidic forms of FtsZ seem to be encoded by at least two gene families (18), suggesting that FtsZ gene duplication may have been critical for evolution from endosymbiont to organelle. In a small number of unicellular eukaryotes, FtsZ proteins derived from the α-proteobacterial ancestor of mitochondria also function in mitochondrial division, but mitochondrial FtsZ has been lost from most eukaryotes, including plants (19).
In plants, plastid division is mediated by two phylogenetically distinct FtsZ families called FtsZ1 and FtsZ2 (18,20). Both proteins reside in the stromal compartment of the chloroplast and localize to rings at the plastid division site (21–23). The two most noted differences between the FtsZ1 and FtsZ2 proteins are (a) the variability of a single amino acid residue in the otherwise conserved ‘tubulin signature motif’ (24) and (b) the conservation in FtsZ2 proteins, but not FtsZ1, of a short C-terminal motif found in most bacterial FtsZ proteins. In E. coli, this motif mediates an interaction between FtsZ and two membrane-associated Z-ring stabilizing factors, FtsA and ZipA (25–27). The significance of the difference between FtsZ1 and FtsZ2 in the tubulin signature motif is not yet known, but the C-terminal motif in FtsZ2 has been shown to mediate an FtsZ2-specific interaction with the chloroplast division protein ARC6 (28) (described below). Neither ZipA nor FtsA are found in plants, but the observation that ARC6 stabilizes plastidic FtsZ filaments in vivo (29) suggests it could be functionally related to one of these proteins. One possibility is that ARC6 could promote bundling of FtsZ protofilaments similar to EcZipA (27), thereby promoting Z-ring assembly.
In addition to the differences in amino acid sequence, there is some evidence that FtsZ1 and FtsZ2 differ in their biochemical properties and in vivo behavior. El-Kafafi et al. (30) have shown that purified recombinant FtsZ1 is capable of forming high-molecular weight complexes in the presence of GTP, and electron microscopy shows that these complexes resemble filaments formed by recombinant E. coli FtsZ (31,32). FtsZ2 did not form any filament-like structures in vitro in these experiments, but seemed instead to form aggregates in the presence of GTP. While these are intriguing differences, all the above experiments were performed with precursor proteins bearing the transit peptides; thus their relevance to the in vivo behavior of mature FtsZ1 and FtsZ2 are unclear. El-Kafafi et al (30) also found that FtsZ1 and FtsZ2 have differing membrane association characteristics upon fractionation of isolated spinach chloroplasts (30); FtsZ1 was mostly present as a soluble stromal protein, while most of the FtsZ2 was tightly associated with the envelope membranes. We have obtained similar results in some experiments (J. M. Glynn, B. Olson and R. S. McAndrew, unpublished observations) but not others (22). The reasons for the variation in FtsZ2 membrane association are unclear. One possibility is that FtsZ2 localization varies at different stages of the plastid division cycle, although this has not been investigated. Typical chloroplast preparations from Arabidopsis or other higher plants are likely heterogenous; they do not contain a uniform population of dividing plastids, frozen at the same stage of division. It is likely that the Z-ring and the other components of the division apparatus are undergoing dynamic assembly and disassembly throughout the organelle division cycle. Therefore, synchronizable algal experimental systems may yield important insights into FtsZ properties and dynamics, possibly linking FtsZ behavior with a discrete stage of plastid division.
Experiments showing that either overexpression or depletion of FtsZ1 or FtsZ2 causes dose-dependent defects in chloroplast division, as indicated by reduced numbers of oversized chloroplasts (20,33), suggest that their stoichiometry may be important for Z-ring function (33). Consistent with this possibility, quantitative analysis of FtsZ1 and FtsZ2 protein levels in isolated Arabidopsis chloroplasts reveals an approximate 1:2 molar ratio (34). Whether this reflects the stoichiometry in the Z-ring of dividing chloroplasts is not yet known and further work is needed to clarify this matter as well.
Control of Z-Ring assembly and placement: The Min system
In bacteria, Z-ring positioning is regulated by the Min system (named for the mutant minicelling phenotype). The Min system allows Z-ring assembly at the mid-cell and prevents Z-ring assembly near the cell poles, ensuring symmetric division (12). An evolutionarily divergent Min system appears to actively position the placement of the Z-ring in the chloroplast.
The Min system has been studied extensively in the Gram-negative bacterium E. coli, in which it is composed of three protein components: MinC, MinD and MinE (12). MinC binds directly to and inhibits polymerization of FtsZ (35–37), but MinC activity is restricted to the cell poles by the concerted action of MinD and MinE (38,39). MinD, a polymer-forming ATPase, assembles on the inner leaflet of the cell membrane near the cell poles, forming a polar zone (12,40). By binding to MinD, MinC becomes tethered to the membrane in the polar zones, inhibiting FtsZ polymerization at the cell poles (41,42). MinD, and hence MinC, is prevented from localizing to the division site by MinE, which forms a dynamic ring-like structure near the mid-cell at the edge of the polar zone (43,44). By stimulating MinD ATPase activity, MinE causes MinD, and hence MinC, to dissociate from the membrane, keeping the concentration of membrane-associated MinD/MinC low near the cell center. The absence of MinC at the cell center allows FtsZ polymerization and Z-ring assembly to occur at this position, initiating division. In E. coli, the positioning of the Min system itself is directed in part by an oscillatory mechanism wherein all three Min proteins move from pole-to-pole (45). Oscillation of the Min system is critical both for inhibiting Z-ring assembly at the cell poles and permitting assembly at the cell center in E. coli (12). However, MinE is not found in the Gram-positive bacterium Bacillus subtilis. Instead, MinCD activity is restricted to the cell poles, and hence Z-ring assembly to the cell center, by the action of DivIVA in a distinct mechanism that does not involve pole-to-pole oscillation of MinC and MinD (11,12).
Homologs of MinD and MinE have been identified in plants and green algae and the Arabidopsis proteins AtMinD and AtMinE have been functionally characterized by Møller and colleagues (8,46–48). Mature AtMinD and AtMinE are stromal proteins, and green fluorescent protein (GFP) fusion proteins localize near the poles of chloroplasts (46,47,49). AtMinD and AtMinE are capable of interacting in vivo (28) and their biochemical characteristics are mostly consistent with those of their bacterial cousins, although AtMinD ATPase activity is calcium rather than magnesium dependent (50,51). Unlike EcMinD, AtMinD ATPase activity is not stimulated by phospholipid (50)—perhaps unsurprising because the plastid inner envelope membrane contains very little phospholipid (52). The same group found that AtMinE can form homodimers in yeast and in planta and propose that AtMinE homodimers near the mid-plastid might quench the activity of an AtMinD-containing FtsZ-inhibitory complex, much like the Min system of E. coli (12). How the reported polar localization of AtMinE (49) relates to this hypothesis is unclear, however. Whether AtMinD and AtMinE form oscillating complexes in chloroplasts is unknown.
Mutant analysis in Arabidopsis has shown that both AtMinD and AtMinE play roles in positioning of the chloroplast division site (47,48). Consistently, assembly and placement of the plastidic Z-ring is altered in the corresponding mutants (D. W. Yoder, S. Vitha, K. W. Osteryoung, unpublished observations) further suggesting that the Arabidopsis proteins have roles analogous to MinD and MinE in E. coli. Depletion of AtMinD by expression of an antisense transgene results in multiple asymmetric constrictions within a single chloroplast, reminiscent of E. coli minD mutants (48); the position of each Z-ring coincides with a constriction (29). A mutant allele of AtMinD, arc11, confers similar Z-ring phenotypes (Figure 2). In contrast, when AtMinD is overexpressed, FtsZ is not observed in rings, but is instead detected in short filaments inside a greatly enlarged plastid (29). These results are consistent with a role for plastidic MinD in preventing Z-ring assembly away from the mid-plastid. AtMinE has the opposite effect on plastidic Z-ring formation. Recently, arc12 was identified as an allele of AtMinE (S. Miyagishima, D. W. Yoder, K. W. Osteryoung, unpublished observations). A frameshift mutation in arc12 introduces a premature stop codon into the AtMinE open reading frame (Figure 2, legend), and the wild-type AtMinE gene complements the arc12 chloroplast division defect. arc12 mutants have short FtsZ filaments within a single oversized plastid, like AtMinD overexpressors (29), whereas AtMinE overexpressors have multiple sites of constriction, and presumably multiple Z-rings, within the chloroplasts (49,53), similar to AtMinD depletion and arc11 mutants (48). Taken together, the morphological and FtsZ localization phenotypes in AtMinD and AtMinE overexpressors and mutants suggest that plastidic MinD and MinE antagonistically regulate Z-ring assembly and placement in plant cell chloroplasts.
Figure 2. Phenotypes associated with arc3 and AtMinD/E alleles. Top panel. A–D) Immunofluorescent micrographs of AtFtsZ2-1 filament morphology in chloroplasts of mature leaf mesophyll cells from wild-type Col-0 (A), arc11 (an allele of AtMinD) (B), arc12 (an allele of AtMinE) (C) and arc3 (D). Bottom panel. E–H) Light micrographs of chloroplast morphology in mature leaf mesophyll cells from wild-type Col-0 (E), arc11 (an allele of AtMinD) (F), arc12 (an allele of AtMinE) (G) and arc3 (H). Arrowheads in (H) show points of constriction in arc3 that are probable sites of Z-ring formation. The mutant arc12 harbors a lesion (1196GC to A–) in AtMinE creating a frameshift and premature stop, changing 108AWKI111 to 108IGRStop111. Size bars = 20 μm.
Although MinD and MinE are conserved in plants, no MinC-like sequences have been found in plant genomes or EST collections. This is somewhat surprising because all three Min components are conserved among the cyanobacteria (17). In E. coli, Z-ring assembly is inhibited at the cell poles by direct interaction of FtsZ with MinC, which is thought to destabilize FtsZ polymers (54). Neither AtMinD nor AtMinE binds FtsZ1 or FtsZ2 (28), suggesting they do not have a MinC-like function—therefore, it seems plausible that a functional analog of MinC may exist in plants that are divergent from cyanobacterial MinC. The chloroplast division protein ARC3, a plant-specific protein with some similarity to FtsZ (55), has been suggested as a possible chloroplastic MinC replacement based on its mutant phenotype and ability to interact with AtMinD, AtMinE and FtsZ1 (56). Further, we have observed that arc3 mutants exhibit multiple Z-rings within a single enlarged chloroplast (Figure 2D), similar to arc11 (Figure 2B) and AtMinD antisense mutants (29). Taken together, these findings suggest that ARC3 may indeed regulate Z-ring positioning and assembly in a MinC-like manner. Although ARC3 has a weakly predicted N-terminal transit peptide and is reported to be cytosolic (55), the above findings also suggest that ARC3 is a stromal protein. In agreement with this likelihood, we find that a rice ortholog of ARC3 is strongly predicted by TargetP (57) to be chloroplast-targeted. While the reported localization of ARC3 at the division site (55) may seem inconsistent with its proposed MinC-like function and colocalization with AtMinD near the poles of the chloroplast (49,56), mid-cell and polar localization of GFP-MinC has been reported for B. subtilis (58). Moreover, cyanobacteria encode relatives of both MinE (found in Gram-negative bacteria) and DivIVA (found in Gram-positive bacteria) (17), suggesting that cyanobacteria and plastids employ a mechanism for Z-ring placement that is distinct from those used by E. coli or B. subtilis. Further work on ARC3 is needed to verify its localization, topology and function in chloroplast division.
Beyond the Min system, a second mechanism of Z-ring placement, called nucleoid occlusion, is also active in many bacteria (12). The nucleoid occlusion system inhibits assembly of the Z-ring over the bacterial chromosome, averting scission of the genome during cytokinesis (59,60). However, recent evidence suggests that a nucleoid occlusion mechanism does not operate in the cyanobacterium Synechococcus elongatus, as Z-rings were noted to form around the chromosome in this organism (17). There is no strong evidence to support the existence of an analogous system in other cyanobacterial species or plastids based on blast searches using E. coli SlmA (59) as a query sequence (J. M. Glynn, Michigan State University, East Lansing, Michigan, unpublished observations).
Promoting Z-ring assembly: ARC6
Previous work from our laboratory has shown that ARC6 is a bitopic inner envelope membrane protein that acts as a positive regulator of Z-ring formation (29). ARC6-GFP localizes to a ring-like structure at the mid-plastid (29), similar to AtFtsZ1 and AtFtsZ2 proteins (23). ARC6 is a descendant of the cyanobacterial cell division gene Ftn2 (61), and ARC6 and its orthologs are only found in cyanobacteria and chlorophytes. ARC6 and Ftn2 proteins possess a conserved region at their N-termini with sequence similarity to J-domains, implicating them as possible Hsp70-associated co-chaperones. arc6 mutants have short FtsZ filaments within a single large chloroplast and ARC6 overexpressors have abnormally long, branched FtsZ filaments held within an oversized plastid. These phenotypes suggest that ARC6 could play a role in bundling of short FtsZ filaments into a ring at the chloroplast division site.
The N-terminus of ARC6 resides in the stroma (29) and a conserved N-terminal segment of ARC6 interacts with FtsZ2 but not FtsZ1 (28). This interaction represents an important functional difference between the FtsZ1 and FtsZ2 families. The interaction requires the short C-terminal motif in FtsZ2 described above, reminiscent of the FtsZ–ZipA and FtsZ–FtsA interactions observed in E. coli (62). The J-domain of ARC6 is not required for interaction with FtsZ2 (28). In contrast, Ftn2 is reported to require the J-domain for interaction with cyanobacterial FtsZ (63) but the significance of this difference is not yet understood.
Despite its relevance to our understanding of plastid division, only a few studies have identified components of the cyanobacterial cell division apparatus (17,61). While some of these are orthologs of division components from Firmicutes and Proteobacteria, others like ARC6 are unique to the cyanobacterial lineage and provide an opportunity for greater understanding of the chloroplast division apparatus. Discrete analysis of division components may be more convenient and efficient in cyanobacteria rather than Arabidopsis or other model systems because of the short generation time, ease of transformation and gene replacement, and the ability to obtain a near-synchronous culture.