Canonical cell division in Bacteria is initiated by invagination of the cytoplasmic membrane (septation) brought about by the formation of a contractile ring (the Z-ring) composed of the tubulin - like protein FtsZ, the product of the ftsZ gene (Lutkenhaus and Addinall, 1997; Bramhill, 1997; Erickson, 1997; Rothfield et al., 1999). Formation of the Z-ring is accompanied by synthesis of septal peptidoglycan, and interference with peptidoglycan synthesis by inhibitors such as penicillin blocks cell division. FtsZ is an indispensible participant in the cell division of most Bacteria, and in Archaea, only the kingdom Crenarchaeota is without it. Chloroplasts also use FtsZ in cell division, but most mitochondria use a dynamin - related protein instead (Erickson, 2000; Gilson and Beech, 2001). The genome of Ureaplasma urealyticum is without ftsZ. The three sequenced Mycoplasma genomes all have ftsZ but none of the genes of the peptidoglycan biosynthetis pathway and the organisms are not penicillin-sensitive (see Table 2 for references).

The situation in Chlamydiaceae is so far unique [see also: ftsZ and peptidoglycan]. The genomes of all three sequenced species, C. trachomatis, C. muridarum, and C. pneumoniae, are without ftsZ but they have a complete (or nearly complete) set of genes for the biosynthesis of peptidoglycan [see: Genome comparisons for references]. Chlamydiaceae have no demonstrable peptidoglycan sacculus and their division is inhibited by penicillin. Most Bacteria have peptidoglycan and are penicillin-sensitive; the few without peptidoglycan are penicillin - insensitive (Moulder, 1993). I have called this apparently contradictory state of affairs the chlamydial anomaly. How this anomaly might be resolved has been the subject of much speculation (Chopra et al., 1998; Ghuysen and Goffin, 1999; Bavoil et al., 2000) to which I will not add. I will ask instead how the anomaly might have evolved and how it might contribute to chlamydial fitness.

It is clear that loss of neither FtsZ nor peptidoglycan is prerequisite to obligate intracellular existence. The obligate intracellular parasite R. prowazekii (Pang and Winkler, 1994; Andersson et al., 2001) and the obligate endosymbiont Buchneri sp. APS (Shiginoba et al., 2000) have both FtsZ and peptidoglycan.

Binary fission in Chlamydia and Chlamydophila, as observed in thin-section electron micrographs, is unremarkable except that septae are almost never seen in constricted and presumably dividing RBs. However, both C. trachornatis and C. psittaci have a non-protein antigen that localizes in the plane of division where the septum should be (Brown and Rockey, 2000). Could it be septal peptidoglycan? Of the nine or more genes essential to the division of E. coli, the bacterium which has been most thoroughly studied (Rothfield et al., 1999; Bramhill, 1999), chlamydiae have only ftsI and ftsW, both of which in E. coli are probably involved in the synthesis of septal peptidoglycan. Chlamydiae also code for other E. coli - like genes associated with, but not essential for, cell division. Therefore, Chlamydiaceae is more likely to have evolved some modification of the typical bacterial cell division mechanism rather than coming up with an entirely new one. A functional substitute for the Z-ring must have appeared, or else RBs would never constrict and separate into daughter cells. As just noted, mitochondria have substituted dynamin for FtsZ, but the chlamydial genomes provide no hint of a dynamin-like molecule. The observation that, in HeLa cells transfected with a dominant negative dynamin mutant, growth of C. trachomatis serovar L2 and C. cavaiae is inhibited (Boleti et al., 1999) is hard to interpret in terms of effect on cell division.

In light of the almost universal occurrence of the gene in Bacteria, the remote ancestors of Chlamydia and Chlamydophila probably divided by an ftsZ-based mechanism. What evolutionary advantage could possible have accrued from dropping a highly successful mode of division and substituting a different one that accomplished the same thing? Obligate intracellular bacteria have small genomes, probably because gene products not needed in the intracellular milieu have been eliminated [see: Genome degradation] but ftsZ is an essential gene in bacteria that have it. In E. coli, ftsZ mutants grow as long, non-dividing filaments at permissive temperatures and not at all at non-permissive ones. How could loss of an essential gene make Chlamydiaceae better adapted to intracellular life? Perhaps some peculiarity of the inclusion favored mutants with a hypothetical pseudo-Z-Ring, but then how is the absence of FtsZ in Ureaplasma to be explained? It has been suggested that chlamydiae exploit host machinery for their own cell division (Erickson, 2000), but I think that unlikely. The presence or absence of ftsZ in the other families of Chlamydiaceae is not known. If they are all without ftsZ, then it must have dropped out of the Chlamydiales lineage at an early date . If one or more of the families still divides by an ftsZ-based mechanism, then Chlamydiaceae must have lost the gene after it diverged from the familial LCA.

A number of Bacteria are without peptidoglycan (mycoplasmas, ureaplasms, planctomycetes, and the rickettsial agent of scrub typhus, for example) and none of the Archaea have it (Moulder, 1993). What is peculiar about Chlamydiaceae is that they have the genes for making peptidoglycan and yet no peptidoglycan sacculus. Their retention of inhibition of cell division by penicillin is an almost certain sign that peptidoglycan synthesis of some sort is nevertheless still involved. As with loss of the Z-ring, something had to take the place of the peptidoglycan sacculus. The integrity of EBs is maintained by disulfide bond cross-linked membrane proteins (Hatch, 1999; Hackstadt, 1999) but RBs, lacking these linkages, must depend on the protection offered by the inclusion. Again as with loss of FtsZ, what is the survival advantage of this loss? Proteins that recognize peptidoglycan are conserved from insects to humans, and peptidoglycan - binding is thought to be an important recognition signal in the activation of innate immunity (Kang et al., 1998). Did the loss of peptidoglycan make Chlamydiaceae less likely to trigger an innate immune response?

When did Chlamydiaceae lose its peptidoglycan sacculus? Of the four Chlamydiales families, only the multiplication of Chlamydiaceae is blocked by penicillin [see: Divergence in Chlamydiales]. Suppose that the LCA of the Chlamydiales families had a functional set of genes for biosynthesis of peptidoglycan and made a peptidoglycan saccculus. Then Simkaniaceae, Waddliaceae, and Neochlamydiaceae have lost all or part of the peptidoglycan pathway so that they neither make it nor need it in their multiplication. On this basis, Chlamydiaceae then occupies an intermediate position, having dispensed with its peptidoglycan sacculus but retained an unexplained need for peptidoglycan in cell division. Other explanations will require even more assumptions. It would be instructive, and relatively easy, to look for penicillin-binding proteins in representatives of the penicillin-resistant families (see Barbour et al., 1982). In Part I, I suggested that the basic properties of the Chlamydiales developmental cycle had already been established when the four families diverged. May I suggest further that the penicillin - sensitive step in multiplication of Chlamydiaceae is one that the other families have eliminated without loss of the definitive Chlamydiales characteristics?
[JWM]