The genome sequence of Rickettsia prowazekii, the agent that causes typhus, has been determined. What emerges is a snapshot of genome re-tailoring in a parasitic bacterium, and a new look at the evolutionary connection between Rickettsia and mitochondria.
On page 133 of this issue, Andersson et al.1 report the complete genome sequence (1,111,523 base pairs) of Rickettsia prowazekii, the causative agent of louse-borne typhus. This bacterium and its relatives represent one of biology's great ironies. On the one hand, the historical ancestors of R. prowazekii precipitated some of the greatest plagues to afflict the human race (see box overleaf). On the other hand, an evolutionary antecedent of R. prowazekii participated in one of the seminal events in the evolution of eukaryotic (nucleus-containing) cells — the formation of mitochondria2, cellular organelles that contain their own DNA and, during oxidative breakdown of glucose, produce the ATP that powers these cells. With the complete genome sequence of R. prowazekii, we can now examine this important genetic blueprint for clues both as to what makes R. prowazekii such a great killer, and what allowed one of its ancestors to contribute so fundamentally to the emergence of eukaryotic cells in the first place.
R. prowazekii is an obligate intracellular parasite — that is, it can only live within other cells. Its gene content, like that of other parasitic eubacteria, has been reduced and tailored to suit its dependent lifestyle. Andersson et al.1 have found that the R. prowazekii genome encodes 834 complete open reading frames, DNA sequences that specify protein sequences. This number is far less than the 4,288 protein-coding genes found in the fourfold larger genome of Escherichia coli, its free-living cousin3. However, R. prowazekii contains ten times as many genes as the most bacteria-like mitochondrial genome described to date, the 69,034-bp mitochondrial (mt)DNA of the freshwater protozoon Reclinomonas americana 4. Surprisingly, the R. prowazekii genome also contains the highest fraction of non-coding DNA (24%) found in any microbial genome so far, much of which may represent inactive genes that have been degraded by mutation, but have not yet been eliminated from the genome.
By comparing the sequences of bacterial and mitochondrial genes, we get the best evidence that Rickettsia and mitochondria are specific evolutionary relatives. Evolutionary trees based on small-subunit ribosomal RNA (SSU rRNA) originally pinpointed members of the 
-division of the so-called 'purple bacteria' (proteobacteria) as the closest contemporary bacterial relatives of mitochondria5. More recent SSU rRNA trees divide the -proteobacteria into two groups, with the rickettsial sub-division (to which R. prowazekii belongs) the one that is specifically affiliated with mitochondria2 (Fig. 1).
Figure 1: Relationship between the R.prowazekii genome and mitochondrial DNA.
The tree shown is the -proteobacterial/mitochondrial (MT) portion of a eubacterial/organellar small-subunit (SSU) ribosomal RNA tree. Extreme differences in the rate of mitochondrial sequence divergence are responsible for the separation of mitochondria into 'short-branch' (plants, Reclinomonas americana) and 'long-branch' groups. (The analysis used an aligned data set of 275 published eubacterial and organellar SSU rRNA sequences, and the tree was generated using DNADIST from Phylip version 3.5 (ref. 9) with the ML option and default parameters2. Courtesy of D. F. Spencer, Dalhousie University.)
The same result is seen with evolutionary trees based on mitochondrial proteins encoded by the nuclear genome6. Such nuclear genes are assumed to have been transferred from mitochondria during the drastic down-sizing that characterized evolution of the mitochondrial genome after it was acquired by the eukaryotic cell2. In their analysis, Andersson et al.1 constructed evolutionary trees comparing the amino-acid sequences encoded by mitochondrial and bacterial genes involved in energy metabolism (subunits of NADH dehydrogenase) and genetic processes (ribosomal proteins). True to expectation, these results show that R. prowazekii is more closely related to mitochondria than is any other bacterium whose genome has been investigated at this level of detail.
Andersson and colleagues rightly point out that the DNAs of Rickettsia and mitochondria are "stunning examples of highly derived genomes, the products of several modes of reductive evolution". Both lack genes for metabolizing sugars in the absence of oxygen (anaerobic glycolysis), as well as all or most of the genes involved in synthesizing amino acids and nucleotides. However, the R. prowazekii genome contains a complete set of genes encoding components of the tricarboxyclic acid cycle, a metabolic pathway involved in respiration, and respiratory-chain complexes. A subset of the same genes is found in mtDNA, with the remainder in the nuclear genome. The functional profiles of Rickettsia and mitochondria are strikingly similar, with production of ATP occurring in basically the same way in the two systems.
Do these similarities mean that mitochondria evolved directly from a Rickettsia -like organism that was already highly reduced? The answer is almost certainly no. If one compares the organization of the same genes in the R. prowazekii 1, E. coli 3 and Reclinomonas mitochondrial4 genomes, vestiges of bacterial operon organization (gene clustering) are still clearly seen in the Reclinomonas mtDNA. However, the Reclinomonas mitochondrial and R. prowazekii genomes do not specifically share any derived characteristics of gene organization; indeed, certain gene clusters are uniquely rearranged in the Rickettsia genome, relative to what was probably the ancestral bacterial order. These comparisons emphasize that the Rickettsia and mitochondrial genomes independently descended from an -proteobacteria-like ancestor, each undergoing a separate process of reductive evolution. The observed congruence in the functional profiles of Rickettsia and mitochondria is certainly intriguing, but it remains to be seen whether this seeming example of convergent evolution is more than a coincidence.
By examining other rickettsial genomes we should learn more about genome reduction in bacterial parasites, a challenging question in its own right. However, such studies are unlikely to provide more information about the nature of the genome in the most recent common ancestor of mitochondria and the Rickettsiae. While the search continues for organisms with mtDNAs that are even more ancestral than in Reclinomonas, it will be important to identify and explore the genomes of those minimally diverged, free-living -proteobacteria that are specific but more distant relatives of both the Rickettsiae and mitochondria. Such genomes should yield additional clues relevant to the origin and evolution of mitochondria, a process that is central to the emergence of eukaryotic life7,8.
