If Chlamydiaceae has indeed descended from a large-genome forebear, present species must have lost many of the ancestral genes, but they have also gained a myriad of capabilities not possessed by that distant hypothetical organism. As the many eubacterial lineages that gave rise to intracellular parasites and symbionts radiated into different hosts and different host cells, there evolved about as many successful ways of intracellular living as there are different kinds of intracellular bacteria (Moulder, 1985; Finlay and Falkow, 1989; 1997; Corsaro et al., 1999). The central question in chlamydial evolution is how and why, of the many pathways they could have taken, Chlamydiaceae evolved their own uniquely peculiar way of intracellular life. Some appreciation of the magnitude of this question may be gained by comparing the hypothetical situation at the very beginning of intracellular existence for Chlamydiales with that obtaining in extant Chlamydiaceae spp..

Phagocytosis probably appeared in ancient eukaryotes as a way of feeding on food particles too large to be otherwise ingested. It is hard to visualize phagocytosis as occurring other than by invagination of the cell membrane about the object to be ingested, followed by pinching off of the invagination with the production of a vacuole. Some of the ingested particles must have been early bacteria which, on rare occasions, were not consumed but survived and multiplied. They were the ancestors of mitochondria and chloroplasts and perhaps, as I have suggested, chlamydiae (Part I). Let us assume that the chlamydial ancestor found itself inside a primitive phagosome. With no time to evolve evasive devices, how did it escape destruction (also discussed in Part I)? It may be that the ancient eukaryotic host also had no time to evolve efficient phagosomal killing mechanisms. Survival might have been due more to forbearance on the part of the host than to enterprize on the part of the parasite. There is no agreement as to what the early eukaryotic hosts were like or what extant organisms most closely resemble them (Kurland and Andersson, 2000), so we have no idea what phagocytosis and killing in the first Chlamydiales host was like. However, recent demonstration of the ancient origin of some of the mechanisms of innate immunity (Hoffman et al., 1999; Medzhitov and Janeway, 2000) suggests that past versus present differences may be less than once thought.

Now to compare the adventitious interaction between the chlamydial ancestor and its first host with the complex chain of events set in motion when a present-day chlamydia encounters a receptive host cell. The interaction of modern chlamydiae with host cells is well-described in several recent reviews which should be consulted for details (Hackstadt, 1999; Hatch, 1999; McClarty, 1999). Here, I will outline this interaction in the briefest and most general of terms, which will probably satisfy no one but which will serve the purposes of the discussion to follow. Chlamydiae attach to and enter host cells by unusually efficient mechanisms that are a function of the infecting chlamydial EB. Once inside, the EB almost immediately begins to modify its phagocytic vacuole to make it a more suitable place for chlamydial survival and multiplication. This it does by incorporating chlamydial proteins and host lipids into the membrane of the vacuole (the inclusion membrane) so that the chlamydia-containing vacuole (inclusion) follows an exocytic pathway to a safe intracellular haven. At the same time, the EB becomes an RB which undergoes several cell divisions before some of the RBs reorganize into a new generation of EBs which exit the host cell to infect new cells. This is, of course, the developmental cycle which must be regulated by the sequential expression of appropriate genes. At the same time, chlamydial proteins are being inserted into the inclusion membrane and other chlamydial proteins are secreted into the host cytoplasm where they modify host activities to the benefit of the chlamydiae. The Chlamydiales lineage, which started out as an accidental intracellular visitor, has gradually assumed control of the host cell without immediate major disruption of its function.

What genomic changes have supported adaptation of the Chlamydiales lineage to intracellular life? How do the genomes of today's chlamydiae differ from those of the first intracellular Chlamydiales? These questions can be answered only in a vague and tentative way. Extant chlamydiae are almost certainly much smaller, but even with drastic genome degradation, over half of the genes in Chlamydia and Chlamydophila can be functionally related to genes in other bacteria (Stephens et al., 1998; Read et al., 2000). They have probably descended from genes of the extracellular ancestor. But it is unlikely that the intracellular activities of chlamydiae I have just summarized are the product of genes already present in the ancestral extracellular Chlamydiales. They must have been added to the Chlamydiales lineage as it adapted to intracellular life. There are certainly enough genes of unknown function in the chlamydial genomes to accommodate new acquisitions. New genes had to come either from the outside by horizontal transfer or from the inside by re-tooling of genes of the ancestral genome. Although Chlamydiales may have picked up genes from both plants and animals in early evolutionary days, evidence of recent horizontal transfer is limited (Stephens et al., 1998; Kalman et al., 1999; Stephens, 1999; Wolf et al., 1999; Bush and Everett, 2001). Two examples of possible outside origin of genes of importance to the intracellular life of chlamydiae will be discussed separately [see: ATP/ADP transferases; Type Ill protein secretion systems). Jacob (1977) introduced the concept of ‘molecular tinkering’; the adaptation of existing resources to new situations. I suspect that there was considerable ‘tinkering’ going on as the Chlamydiales lineage shifted from extracellular to intracellular habitats. The "new" genes, the ones that were evolved in response to pressures of the intracellular habitat, were more likely to be derived from "old" genes of at least remotely parallel function in the extracellular ancestor rather than being created de novo. For example, I have already suggested (Part 1) that the developmental cycle may have had its beginning in the tendency of all bacteria to exhibit different phenotypes in their multiplying and quiescent stages. On a more specific level, a gene coding for a protein homologous to the integration host factor (IHF) and the heat-stable nucleoid protein HU may be involved in regulation of the late stages of the developmental cycle (Zhong et al., 2001).

However all these genetic changes may have occurred, the genotypes of extant Chlamydiaceae must represent optimum sets of mutations assembled from mutations occurring separately in different isolated populations. Genetic exchange among chlamydiae has not been conclusively demonstrated, but over a long time even low rates of exchange would be enough. Whether or not extant species of Chlamydiaceae are optimally adapted to their several niches is a question yet to be answered.

When I contemplate the intricate web of interactions between modern chlamydiae and their host cells, I can appreciate the feelings of the nineteenth-century savants who doubted that the vertebrate eye could have been formed by natural selection. Darwin refuted their doubts by showing how the eye could have evolved by discrete steps identifiable in lower life forms. Some day the steps in evolution of the chlamydia - host cell interaction will no doubt become apparent. Perhaps in time, a comparative genetic anatomy of Chlamydiales will emerge, but all the truly interesting events seem to have happened a long time ago. Extant Chlamydiaceae are too much alike to be of any help, and even the different Chlamydiales families do not seem to depart very far from the same plan.
[JWM]