Topological cues appear to override temporal events in the assembly of the Escherichia coli cell division ring. When a procedure that allows the recruitment of ring components based on their topological properties is used, a concerted mode of assembly of several components of the divisome, rather than a strict linear mode, is revealed. Three multimolecular complexes, the proto-ring, the periplasmic connector and the peptidoglycan factory, show some degree of concertation for their assembly. In addition, back-recruitment of all late proteins except FtsN into the division ring occurs even in the absence of proteins incorporated at earlier stages, i.e. FtsA or FtsQ.
Ten essential proteins must assemble at the midcell to form a division ring for an Escherichia coli cell to divide (revised in Vicente et al., 2006). All of them interact with at least one other member of the group. How these proteins find their place into the divisome has been investigated by making extensive use of mutants and fusions to fluorescent proteins. The results of these studies have led to the proposal that the E. coli divisome components are recruited in a hierarchical linear order (Fig. 1A; Buddelmeijer and Beckwith, 2002).
Given the ability of many of the ring components to interact readily with each other (Di Lallo et al., 2003; Karimova et al., 2005), the operation of a strictly linear hierarchical assembly mode would require mechanisms to prevent simultaneous assembly. For example, the ring components could be synthesized as precursors that are unable to interact with other elements and that must be processed or modified in order to become activated for assembly, or their transcription could be temporally repressed. In both cases, the immediate upstream division protein would be needed to control directly or indirectly the release of the blocking mechanism. Processing has been described solely for FtsI, and none of the more upstream division proteins has been shown to participate in the production of the mature protein (Nagasawa et al., 1989). Although a temporal control of the transcription of ftsZ, coding for the first protein of the ring, has been described (Garrido et al., 1993), temporal regulators of transcription, if present, do not play an essential role in cell division (Vinella et al., 2000). It is then unlikely that the assembly of divisome components involves extensive protein maturation or essential transcriptional regulatory mechanisms that would account for the linear hierarchical assembly. On the contrary, assembly of FtsZ, FtsA and ZipA into rings cannot be visualized as separate events (Rueda et al., 2003), suggesting that some elements of the division ring do not follow a strictly linear assembly hierarchy.
Goehring et al. (2005) have recently developed an ingenious method to circumvent the need for pre-assembly of upstream proteins at the ring for the recruitment of downstream ring components. Further application of this technique, reported in this issue of Molecular Microbiology (Goehring et al., 2006), provides a different view of the mode of division ring assembly that seems to be closer to the concerted mechanism proposed for other bacteria such as Bacillus subtilis (Errington et al., 2003).
Topological recruitment of ring components
The assay developed by Goehring et al. (2005) is based on the ability of ZapA, a dispensable protein that promotes in vitro bundling of FtsZ filaments, to bind to FtsZ (Gueiros-Filho and Losick, 2002). The procedure allows direct assessment of the ability of a ring component to recruit other cell division components independently of the presence of the upstream components normally required for its assembly. Essentially, this procedure allows the recruitment of ring components based on their topological properties, rather than on their temporal appearance within a structure. The first results obtainedusing this procedure indicated that FtsQ can coordinate the recruitment of the downstream proteins, except FtsN, in the absence of the upstream FtsA and FtsK proteins (Goehring et al., 2005). Even more surprising was the finding that prematurely localized FtsQ can back-recruit FtsK even in the absence of the upstream protein FtsA. This observation suggested a direct protein–protein interaction between FtsQ and FtsK in the division complex and was the first in vivo assay to show back-recruitment of division proteins.
The existence of interactions between the E. coli cell division proteins in vivo has been revealed by two different bacterial two-hybrid assays (Di Lallo et al., 2003; Karimova et al., 2005). They confirm that FtsZ binds to FtsA, ZipA and FtsK, the other components of the cytoplasmic ring (Fig. 1D). FtsA and FtsK interact with the proteins of the periplasmic ring, including FtsQ, the component showing the highest number of potential interactions with other members of the division ring (see fig. 5 in Vicente et al., 2006). Paradoxically, in the absence of intermediate proteins, the observed interactions between some pairs (e.g. FtsZ–FtsK, FtsA–FtsI, FtsQ–FtsN) are not proficient for recruiting the second member of the pair into the midcell.
Interactions between FtsA, an early assembly protein, and FtsI and FtsN, two proteins that assemble later, has been reported based on a polar recruitment assay in which a fusion between FtsA and DivIVA is driven to the cell pole together with interacting partners by the ability of DivIVA to migrate and integrate into the poles (Corbin et al., 2004). However, neither this polar recruitment nor the two-hybrid assays can show whether such interactions are direct or indirect.
Recovering the lost order
The application of a topological recruitment assay to probe additional members of the division ring (Goehring et al., 2006) substantially modifies our views on the hierarchy of the assembly process (Fig. 1B). The previously accepted linear assembly sequence described the recruitment into the division ring as a temporal sequence of events in which the incorporation of an individual protein was strictly dependent on the prior incorporation of all the upstream components (Fig. 1A).
The first hint showing that a strictly linear hierarchy does not apply to the entire assembly process was provided by data on the assembly of FtsZ, FtsA and ZipA (Pichoff and Lutkenhaus, 2002). We can define them as forming the proto-ring by the concerted assembly of all three components (Fig. 1C) and in which at least one membrane anchor, provided either by the FtsA carboxy end or the ZipA membrane domain, is required to stabilize the FtsZ molecules accumulated at the midcell. These three proteins remain together for some time until the rest of the division proteins localize into the ring (Aarsman et al., 2005; Fig. 1E), which suggests that the proto-ring might signal the cell to initiate constriction via peptidoglycan synthesis (Vicente et al., 2006). The data obtained with the topological recruitment assays show that FtsQ fused to ZapA can back-recruit FtsK into the final division ring even in the absence of FtsA (Goehring et al., 2005). FtsK is a large bifunctional protein that is bound to the cytoplasmic membrane and that was previously considered to require FtsA for assembly. Three proteins, FtsQ, FtsB and FtsL, each containing cytoplasmic, membrane and periplasmic domains, could connect the cytoplasmic ring (formed by the proto-ring and FtsK) and the proteins attached to the peptidoglycan sacculus, namely FtsI and FtsN (Fig. 1C). The topological recruitment assay illuminates the peculiar assembly mode of this trio. The three proteins seem to assemble together before their incorporation into the ring (Buddelmeijer and Beckwith, 2004) but FtsL and FtsB can contact with each other independently of the presence of FtsQ and probably do not need any other divisome component for their interaction prior to their recruitment into the ring (Goehring et al., 2006).
Two elements of the divisome likely to form part of the septal peptidoglycan factory (Fig. 1C) are FtsW, a putative transporter of the precursors, and FtsI, which is a septation-specific transpeptidase. In the absence of FtsA or FtsQ, both elements of the peptidoglycan factory can recruit each other into the divisome (Goehring et al., 2006). Finally, FtsN is a late assembling protein that needs the presence of all the components that assemble earlier in order to be recruited. Moreover, even if the entire FtsA-independent complex is assembled, FtsN fails to be recruited in the absence of FtsA (Fig. 1F; Goehring et al., 2006), suggesting that an additional signal activated by the presence of every other division protein is required for its incorporation into the ring.
Ordering the future
Based on our current knowledge, the assembly of the divisome components to form a division ring appears to be a multistage process in which some concerted events bring together components of the proto-ring, the periplasmic connector and the peptidoglycan factory (Fig. 1B). How these three complexes assemble together, and how they interact with the individual FtsK and FtsN proteins remains to be discovered. In particular the role of FtsN, considered in the linear model as being the last protein to assemble, but able to complement defects in a proto-ring component, FtsA (Dai et al., 1993), remains to be elucidated.
One caveat of the experimental procedures used to probe divisome assembly is their reliance on protein fusions to fluorescent proteins and overexpression. Although genes encoding fluorescent chimeras are able to complement conditional lethal mutations in the corresponding chromosomal gene, there is no guarantee that their interaction properties duplicate exactly those of their wild-type counterparts. A similar concern can be expressed regarding the overexpression issue. In addition, we cannot fully exclude that temporal regulation of gene expression contributes to the fine-tuning of the assembly process. Six of the proteins that assemble into the division ring (FtsZ, A, Q, L, W and I) are encoded by genes that map together in the dcw cluster, a conserved region in which an abundance of promoters might exert subtle influences on the timing of cell division (Palacios et al., 1996). In addition the dcw cluster harbours genes involved in the synthesis of peptidoglycan precursors (Tamames et al., 2001). Genomic channelling has been invoked as a possible reason for the phylogenetic conservation of genes and gene order in the dcw cluster (Mingorance et al., 2004). It would be no big surprise to find that it also has some role in regulating the assembly of the components of the divisome to effect septation in a fully efficient way.
Many open questions on the assembly and functioning of the cell division machinery are likely be answered by in vitro approaches; the reconstitution of the divisome from its parts will be hoped to reveal the exact assembly order and the precise roles of the ring components.
Work in our laboratory is funded by projects GEN2003-20234-C06-02 and BIO2005-02194 from Ministerio de Educación y Ciencia.