Early investigations on obligate intracellular bacteria suggested that restriction to intracellular habitats had been accompanied by exploitation of host biosynthetic pathways and disappearance of corresponding pathways in the bacteria, presumably by gene loss and inactivation. This view, first expressed by Lwoff (1944) and then by Zamenhof and Eichhorn (1967), was validated years later by determination of the complete genome sequences for a number of obligate endoparasitic and endosymbiotic bacteria. The remarkable reduction in genome size among obligate intracellular bacteria is now widely appreciated and is the subject of both experimental and theoretical interest. Bacteria that thrive in a variety of niches tend to have "large" genomes (4 to 6 X 10⁶ base pairs) whereas those confined to narrowly defined habitats tend to have "small" ones (0.5 to 2.0 X 10⁶ base pairs). The following table lists the genome sizes of bacteria representative of both categories to illustrate the point:. 

Table 2. Influence of life style on genome size.

The Chlamydiales occupy a specialised intracellular niche and would thus be expected to have a small genome. With their genomes of only about 1 X 10⁶ base pairs, Chlamydiaceae fit this expected pattern. In the other Chlamydiales families, only the size of the S. negevensis genome is known, about 1.7 X 10⁶ base pairs (Kahane et al., 1999).

At least three kinds of restricted habitat have favored evolutionary reduction in genome size  - extreme environments; epicellular association with hosts; and obligate intracellular parasitism / symbiosis. It remains to be seen whether similar evolutionary forces were at work in these environments. There seems little difference in genome size between obligate intracellular parasites and obligate intracellular symbionts. Does this mean that the 2 intracellular life styles are equally complex? Surprisingly, epicellular parasites, bacteria that live in close association with hosts but not inside them, have genome sizes in the same range as the obligate intracellular organisms. Are the genomes of Chlamydiaceae still shrinking? Although other explanations are possible, differential gene loss may be inferred from the distribution of the genes for tryptophan biosynthesis and cytotoxins. Tryptophan biosynthesis genes are found in C. muridarum but not in C. trachomatis or C. pneumoniae. Genes for the cytotoxin described by Belland et al., (2001) are present in C. muridarum and are expressed when host cells are infected at high multiplicity. C. trachomatis serovar D also codes for and expresses a cytotoxin, but the gene is smaller and its gene product is less active., whereas C . trachomatis serovar L2 has only a small part of the cytotoxin gene sequence and no toxin is expressed. Finally, the genomes of all 3 sequenced strains of C. pneumoniae have no sequences related to the cytotoxin. Rickettsiae seem to be in process of eliminating the gene for synthesis of S-adenosylmethionine, and none of the sequenced Chlamydiaceae have the gene at all. A variety of approaches have been used to estimate the size of the "minimal genome", i.e the smallest number of genes that will support the existence of a free-living organism ( Hutchinson et al., 1999; Mushegian, 1999; Koonin, 2000). The magic number turns out to be about 300. How big a minimal genome would be needed by an obligate endosymbiont or intracellular parasite?

It has been suggested that Rickettsia and the aphid endosymbiont Buchnera have descended from free-living ancestors with genomes of about 4 X 10⁶ base pairs (Andersson and Kurland, 1998). Since they are both proteobacteria, most of whom have genomes of about this size, the suggestion is plausible. Unfortunately, with chlamydiae there is no similar point of comparison. All known Chlamydiales are obligately intracellular (Part I). It would be my guess that the first intracellular Chlamydiales had a "large" genome and that the LCA of the four families already had a "small" one.

Chlamydiae reproduce asexually in small populations isolated in hosts and host cells and constantly go through bottlenecks in the passage of only a few individuals from host to host. This makes them susceptible to the operation of Muller's ratchet, which predicts that such populations will accumulate deleterious mutants more rapidly than large populations of free-living bacteria (Bell, 1997; Moran, 1996). In this manner, pseudogenes, non-coding residues of genes destroyed by mutation, should accumulate in the genome before being eliminated by genome reduction. The protein-coding capacities of the C. trachomatis and C. pneumoniae genomes have been calculated at 89.5 to 90.5%, virtually identical to the 91% average for all sequenced genomes of Bacteria (Lawrence et al., 2001; Mira et al., 2001), whereas the corresponding value for C. muridarum has been reported as either 82% (Lawrence et al., 2001) or 90.5% (Mira et al., 2001). In contrast, the protein-coding capacity of Rickettsia prowazekii (the agent of louse-borne typhus) is only 76% (Zomorodipour and Andersson, 1998; Andersson et al., 1998; Lawrence et al., 2001; Mira et al., 2001). The coding capacity of the recently sequenced agent of boutoneuse fever, Rickettsia conorii, is somewhat higher, 82% (Ogata et al., 2001) It appears that rickettsiae do indeed exhibit the loss of protein-coding capacity predicted by operation of Muller's ratchet, whereas chlamydiae do not. With so few genome sequences available, this conclusion must be tentative. The only other obligate intracellular bacterium with a sequenced genome, Mycobacterium leprae, shows massive genomic decay (Cole et al., 2001). With a genome of 3.27 X 10⁶ bp and a coding capacity of only 49%, its estimated functional genome contains only 1.60 X 10⁶ bp. Has Chlamydiales evolved a more efficient way of getting rid of its genomic garbage or has its lineage lived as intracellular parasites for a much longer time?

The genetic mechanisms responsible for genome degradation in obligate intracellular bacteria have been considered at length elsewhere (Andersson and Andersson, 1997; 1999; 2001; Andersson et al., 1998; 1999; Muller and Martin, 1999; Andersson and Kurland, 1998; Zomorodipour and Anderson, 1999; Lawrence et al., 2001; Mira et al., 2001). I will not even attempt to summarize these considerations and will pass on to asking what evolutionary advantage was to be gained by genome shrinkage. After all, facultative intracellular bacteria have learned to live inside host cells without obvious reduction in genome size (Table 2), showing that gross degradation of the genome is not a necessary accompaniment of adaptation to intracellular life. However, I know of no obligately intracellular bacterium, parasite or symbiont that does not have a drastically reduced functional genome. The selective advantage of small genomes may lie in a biological version of the "more bang for the buck" principle. Small genomes may allow production of more infectious units per unit of metabolic effort (whatever that is!) with less disruption of host activities and the presentation of a smaller antigenic cross-section to the immune system of the host. However, it is not at all apparent that facultatively intracellular bacteria such as the mycobacteria or the invasive enteric bacteria are any less successful in exploiting the intracellular niche.

Loss of genes has not been random. Genes involved in handling informational macromolecules tend to be conserved, while genes for the biosynthesis of small molecules such as cofactors, amino acids, and purine and pyrimidine nucleotides tend to be lost. However, the obligate endosymbionts of insects furnish an exception to this generalization. These organisms have conserved the ability to make many amino acids that are essential for their hosts (Moran and Baumann, 2000). If a biosynthetic pathway essential for an obligate intracellular bacterium is to be non-lethally eliminated, the product(s) of that pathway must be supplied by the host and made accessible in adequate concentration. With chlamydiae, the transport problem may be doubly critical because the host-supplied surrogate must pass through the membranes of both the inclusion and the chlamydial cell. It should also be remembered that the presence of a metabolite in the host cytoplasm does not necessarily mean that it is freely available to the intracellular bacterium. Host cells can act as both habitat and competitor (Moulder, 1969; Hatch, 1975).
[JWM]

[MEW Comment April 2002: Palmer (2002) points out that bacterial pathogens with small genomes, such as rickettsiae, chlamydiae, spirochaetes, mycoplasma and Ehrlichia devote a high percentage of their reduced genome to paralogous families of polymorphic surface proteins. It is suggested these are important in evasion of the immune response].