J Mol Evol (2001) 53:377–386 DOI: 10.1007/s002390010227

© Springer-Verlag New York Inc. 2001

Accumulation of Species-Specific Amino Acid Replacements That Cause Loss of Particular Protein Functions in Buchnera, an Endocellular Bacterial Symbiont Shuji Shigenobu,1,2 Hidemi Watanabe,2 Yoshiyuki Sakaki,2,3 Hajime Ishikawa1 1

Department of Biological Sciences, Graduate School of Science, University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan RIKEN Genomic Sciences Center, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohamashi, Kanagawa, 230-0045, Japan 3 Human Genome Center, Institute of Medical Science, University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108-8639, Japan 2

Received: 14 December 2000 / Accepted: 12 March 2001

Abstract. Endosymbiotic bacteria live in animal cells and are transmitted vertically at the time of the host’s reproduction. In view of their small and asexual populations with infrequent chances of recombination, these endocellular bacteria are expected to accumulate mildly deleterious mutations. Previous studies showed that the DNA sequences of these bacteria evolved faster than those of free-living bacteria. In this study, we compared all the ORFs of Buchnera, an endocellular bacterial symbiont of aphids, with those of 34 other prokaryotic organisms and estimated the effect of the accelerated evolution of Buchnera on the functions of its proteins. It was revealed that Buchnera proteins contain many mutations at the sites where sequences are conserved in their orthologues in many other organisms. In addition, amino acid replacements at the conserved sites are mostly changes to physicochemically different amino acids. These results suggest that functions and conformations of Buchnera proteins have been seriously impaired or strongly modified. Indeed, extensive loss of functional motifs was observed in some Buchnera proteins. In many Buchnera proteins mutations were not detected evenly throughout each molecule but tended to accumulate in some functional units, possibly leading to loss of specific functions. As Buchnera has an unusual and limited gene repertory, it is conceivable that the manner of interactions among its proteins has been changed, and thus,

Correspondence to: H. Ishikawa; email: [email protected]

functional constraints over their amino acid residues have also been changed during evolution. This may account for the loss of some functional units only in the Buchnera proteins. We obtained evidence that amino acid replacements in Buchnera were not always deleterious, but neutral or, in some cases, even positively selected. Key words: Buchnera — Symbiont — Mutation — Protein evolution — Reductive evolution — PROSITE — Positive selection — Neutral mutation Introduction Many eukaryotic cells provide the sole habitat for a vast and varied array of prokaryotic lineages (Buchner 1965). These endosymbiotic bacteria are locked inside their hosts for an evolutionarily long time, with a small population size and substantially no recombination. As a result, their genomes are expected to have accumulated mildly deleterious mutations in a process referred to as Muller’s ratchet (Andersson and Kurland 1998; Chao 1990; Moran 1996). For example, Buchnera sp., an endocellular bacterial symbiont of aphids, grows inside specialized host cells, called bacteriocytes, that are maternally transmitted through each host generation (Baumann et al. 2000; Buchner 1965; Douglas 1998; Houk and Griffiths 1980). It was reported that the genes of Buchnera exhibit unusually low ratios of synonymousto-nonsynonymous substitutions compared with those for the corresponding genes of enterics (Brynnel et al.

378

Fig. 1. Schematic representation of Buchnera- and E. coli-specific mutation sites. Vertical bars represent the percentage of the most major amino acid at each site. Blue and red bars indicate that the amino acids at the sites are species-specific mutations of likely and unlikely amino acid exchanges, respectively (details are given under Materials and Methods). Conserved sites are represented in dark gray. The percentage threshold for detecting conserved sites is 80%. a PriA, primosomal protein N⬘. b TrpB, tryptophan synthase ␤ chain. c SuhB, extragenic suppressor protein suhB. d HscB, chaperone protein hscB. e RecC, exodeoxyribonuclease V 125-kD polypeptide. The plots of other proteins are available at our website, http://buchnera.gsc.riken.go.jp/.

1998; Clark et al. 1999; Wernegreen and Moran 1999; Moran 1996). This accumulation of nonsynonymous substitutions is compatible with the hypothesis of increased fixations of deleterious mutations, whose consequence will be the inactivation and eventual deletion of nonessential genes (Moran 1996; Wernegreen, 1999). It is apparently consistent with this hypothesis of reductive evolution that the Buchnera genome consists of 640,681 base pairs of DNA and contains only 583 genes. This is the smallest bacterium in both genome size and number of genes among the bacteria sequenced to date, except for Mycoplasma genitalium (Charles and Ishikawa 1999; Shigenobu et al. 2000). An increase in number of complete genomic sequences available permitted systematic comparisons of homologous genes among different organisms (Aravind et al. 2000; Koonin et al. 2000; Tekaia and Dujon 1999;

Watanabe et al. 1997). Important sequence domains and amino acid residues responsible for the biological activities of proteins are under very tight constraints throughout evolution, and these domains and sites have been conserved universally among a wide variety of organisms. We compared protein sequences of Buchnera with those of the homologues of other bacteria and examined amino acid replacements in Buchnera proteins at the sites where common amino acids are conserved in many other bacteria. Such Buchnera-specific mutations will have effects, positive or negative, on the proteins’ specific functions. Materials and Methods Database A comparative genomic analysis was performed, using the complete genomic sequence data on 35 prokaryotes: Haemophilus influenza

379 (NC_000907), Mycoplasma genitalium (NC_000908), Methanococcus jannaschii (NC_000909), Synechocystis sp. (NC_000911), Mycoplasma pneumoniae (NC_000912), Helicobacter pylori 26695 (NC_000915), Helicobacter pylori J99 (NC_000921), Escherichia coli (NC_000913), Methanobacterium thermoautotrophicum (NC_000916), Bacillus subtilis (NC_000964), Archaeoglobus fulgidus (NC_000917), Borrelia burgdorferi (NC_001318), Aquifex aeolicus (NC_000918), Pyrococcus horikoshii (NC_000961), Pyrococcus abyssi (NC_000868), Mycobacterium tuberculosis (NC_000962), Treponema pallidum (NC_000919), Chlamydia trachomatis (NC_000117), Rickettsia prowazekii (NC_000963), Chlamydia pneumoniae (NC_000922), Aeropyrum pernix (NC_000854), Thermotoga maritina (NC_000853), Campylobacter jejuni (NC_002163), Chlamydia pneumoniae AR39 (NC_002179), Chlamydia pneumoniae JI38 (NC_002491) Chlamydia trachomatis MoPn (NC_002178), Deinococcus radiodurans (NC_001263, NC_001264, NC_000958, NC_000959), Neisseria meningitidis Z2491 (NC_002203), Neisseria meningitidis MC58 (NC_002183), Ureaplasma urealyticum (NC_002162), Vibrio cholerae (NC_002505), Rhizobium sp. NGR234 complete plasmid (NC_002491), Xylella fastidiosa (NC_002488), Pseudomonas aeruginosa (NC_002516), and Buchnera sp. APS (NC_002528). For the present analysis, we constructed a database of the protein sequences of 35 genomes, which were obtained from the genome division of the Entrez system at the NCBI website (http:// www.ncbi.nlm.nih.gov/entrez/query.fcigi?db⳱Genome) and from our website, iBGD (integrated Buchnera genome database, http:// buchnera.gsc.riken.go.jp/).

Detection of Species-Specific Mutation Sites Each Buchnera protein was compared against our protein sequence database to search for its homologues, using the BLASTP program (Altschul et al. 1997). Proteins of ⱖ59.8 bits were regarded as homologues. We then constructed a multiple alignment for each Buchnera protein and its homologues with CLUSTALW (Thompson et al. 1994). In the multiple alignments, we searched for Buchnera-specific mutation sites using our custom program SPEC2 (SPECies-SPECific mutation detection program). A Buchnera-specific mutation site was defined by two criteria: (1) more than the given percentage threshold of the homologues of a Buchnera protein have the same amino acid as that of the E. coli orthologue at the site—the sites satisfying this criterion were defined as a conserved site, and the amino acid residue at the site was defined as a consensus; (2) Buchnera has an amino acid different from the consensus. According to the above definitions, it is probable that the last common ancestor’s (LCA’s) orthologue corresponding to an orthologue pair of Buchnera and E. coli had the consensus amino acid at each conserved sites. The information on the sequences of the LCA proteins allows us to know the actual amino acid changes during the evolution of a lineage. The E. coli orthologue of a Buchnera protein was defined as the protein of E. coli that was the most similar to the Buchnera protein. This simple procedure is feasible in this study because Buchnera has no species-specific duplicated proteins besides grpE (Shigenobu et al. 2000).

Motif Search We searched all proteins of Buchnera for motifs in the PROSITE database (Release 16.0, July 1999) (Hofmann et al. 1999) using the online service at GenomeNet (http://motif.genome.ad.jp/kegg/).

Results Detection of Buchnera-Specific Mutation Sites in Proteins Multiple alignments of homologous proteins indicated that Buchnera proteins had species-specific amino acid replacements even at the sites where their homologues from other bacteria have been conserved (examples shown in Fig 1). To see if this trend appeared after the divergence between Buchnera and E. coli, which is the closest relative of Buchnera among the 35 species examined, we compared the number of species-specific mutation sites between Buchnera and E. coli, using 538 multiple alignments each of which contained the E. coli orthologue of a Buchnera protein (Table 1). Four thousand sixty-seven Buchnera-specific mutation sites were identified, whereas only 1066 were identified for E. coli orthologues. A Buchnera-specific mutation site was defined as the site where ⱖ70% of the homologues have the same amino acid residue as does the E. coli orthologue but different from the residue of the Buchnera protein. This difference became augmented with a higher threshold: 2672 versus 603 at 80%, 1593 versus 325 at 90%, and 1285 versus 268 at 100%. Such skewed frequencies were observed for each Buchnera protein. Pairwise comparisons of orthologues revealed that Buchnera genes always accumulated more lineage-specific mutations than the E. coli orthologues at all threshold values, with a few exceptions (Fig. 1). Thus, it is conceivable that the frequent amino acid replacement at the conservative sites is characteristic of the Buchnera proteins. Unusual Patterns of Amino Acid Replacements in Buchnera Proteins If the Buchnera-specific mutations cause exchanges between similar amino acids, the effect on the protein conformation and function may be small. To estimate how significantly the Buchnera-specific mutations affect the functions of the Buchnera proteins, we evaluated the pattern of amino acid replacements at the Buchneraspecific mutation sites using the BLOSUM62 matrix, in which a more likely amino acid exchange is represented by a larger positive value, whereas a less likely exchange is represented by a lower negative value (Henikoff and Henikoff 1992). We counted all types of amino acid replacement from the consensus to the residue of the Buchnera protein (Fig. 2). As a result, we detected many mutated sites whose scores were of negative value in BLOSUM62 (correspondent cells are shaded gray in Fig. 2). Buchnera-specific unlikely mutation sites, i.e., the sites of negative BLOSUM scores, are summarized and compared with those of E. coli in Table 1, showing that Buchnera has about six times as many species-specific

380 Table 1.

Comparison of frequencies of Buchnera- and E. coli-specific mutation sites Buchnera

E. coli

Threshold (%)a

Species-specific mutation sites

Species-specific mutation sites (BLOSUM < 0)

Species-specific mutation sites

Species-specific mutation sites (BLOSUM < 0)

100 90 80 70 60 50

1,285 1,593 2,672 4,067 6,719 10,275

397 478 769 1,144 1,945 3,006

268 325 603 1,066 2,210 4,157

59 69 127 208 454 955

a

The percentage threshold used in detecting conserved sites.

Fig. 2. Frequency of each type of amino acid replacement from the consensus to those of Buchnera at Buchnera-specific mutation sites. Unlikely amino acid replacements (corresponding BLOSUM62 scores are negative) are shaded. The threshold for detecting conserved sites is 90%.

unlikely mutations as E. coli. In most cases, unlikely amino acid exchanges are the ones between amino acids of physicochemically different characteristics and, thus, must have negative effects on protein functions. Buchnera-Specific Mutations in Motifs and Their Effects on Functions We searched 571 proteins shared between Buchnera and E. coli for sequence motifs in the PROSITE database (Hofmann et al. 1999) and compared the inventories of the motifs between these species. One hundred twentyeight motifs in E. coli proteins were not found in the corresponding regions of the Buchnera orthologues, whereas only 38 motifs in the Buchnera proteins are missing in the E. coli orthologues. Such a large difference was not detected between E. coli and Haemophilus influenzae, a close relative of Buchnera and E. coli, indicating that the large loss of functional motifs charac-

terizes the evolution of Buchnera proteins. To see whether the loss of motifs in the Buchnera lineage is a consequence of the Buchnera-specific mutations at conserved sites, we examined the correlation of the positions of the lost motifs in Buchnera proteins with those of the Buchnera-specific mutation sites. We found 118 Buchnera-specific mutation sites in 65 regions corresponding to some PROSITE motif in 55 E. coli orthologues. Since these mutations in Buchnera proteins are found at the conserved sites of proteins of other organisms, i.e., the sites of strong functional constraints, the functions of the Buchnera proteins bearing species-specific mutations may be impaired or modified partly or even completely. We found several intriguing examples of possible loss of functional motifs in Buchnera proteins. PriA, the primosomal protein N⬘, is necessary for the formation of the ⌽X-type primosome and is an indispensable protein responsible for the initiation of recombination-dependent replication (Marians 1996). The priA gene is detected in

381

all completely sequenced genomes except for those of mycoplasmas and Deinococcus radiodurans. Interestingly enough, as many as 43 Buchnera-specific mutations have accumulated in PriA of Buchnera, and these mutations have been introduced in the ATP-binding motif, helicase motif (Al-Deib et al. 1996), and DEEH box, but only one Buchnera-specific mutation was observed in the two zinc finger motifs (Fig. 3a), indicating that the helicase activity of PriA has been impaired but the DNA binding activity remains unaffected. Similarly, RecC, the ␥ subunit of exodeoxyribonuclease V (RecBCD complex), which is involved in recombinational repair (Kuzminov 1999; Lloyd and Low 1996), was mutated in highly conserved regions (Fig. 1e). In the case of the SuhB protein, two motifs responsible for inositol monophosphatase activity are less conserved, whereas other regions are apparently conserved (Figs. 1c and 3b). In the case of HscB, a heatshock protein, the DnaJ-like Nterminal region seems to be conserved among all organisms including Buchnera, but the other part of Buchnera HscB is not well conserved (Fig. 1d). The Buchneraspecific mutations are concentrated in the N-terminus regions in TrpB, the tryptophan synthase ␤ subunit (Fig. 1b). Taking these observations into consideration, it was concluded that in Buchnera, mutations were not fixed evenly throughout genes but selectively in certain regions. Relation of Amino Acid Replacement and AT Bias in the Buchnera Genome We examined whether the increased AT content of codons accounted for the amino acid exchange patterns observed in Buchnera. We focused on the first and second positions of the codons, because nucleotide substitutions at these positions contribute to amino acid replacements. Figure 4 indicates that the tendency in amino acid replacement was strongly influenced by the increase in AT in codons: mutations with increasing AT amounted to 72%; those with decreasing AT, only 9%. This means that the genome of the LCA was, at least in ORFs, richer in GC than that of Buchnera. We looked at the preference of the amino acid replacements at the Buchnera-specific mutation sites. Amino acids coded by GC-rich codons, such as Val, Ala, Gln, Leu, and Met, frequently changed to other amino acids in the Buchnera lineage compared to the LCA (Fig. 5a), whereas amino acids encoded by AT-rich codons, Ile, Ser, and Asn, increased their occurrence strikingly in the Buchnera lineage (Fig 5b).

Discussion In this work, we demonstrated a unique evolution of Buchnera proteins. In Buchnera proteins, amino acid re-

placements have frequently occurred even at the sites that are universally conserved in other organisms. In addition, the tendency of exchanges between consensus amino acids and those of Buchnera showed an unusual pattern. These results suggest that the functions and the conformations of Buchnera proteins have been significantly affected by these mutations. Indeed, many functional motifs of proteins have been lost specifically in the Buchnera lineage. This finding apparently supports the hypothesis that endocellular bacteria should accumulate deleterious mutations (Andersson and Kurland 1998, Wernegreen and Moran 1999; Moran 1996). However, our results suggest that Buchnera-specific mutations are not always deleterious. Once a protein becomes useless for an organism, it will be freed from functional constraints and accumulate mutations randomly throughout its sequence. Interestingly, however, it was revealed that the Buchnera-specific mutations had not occurred at random, but were concentrated in some conserved domains or sites (Figs. 1 and 3). This feature may be explained by the characteristic gene repertory of the Buchnera genome. For example, recG, the gene for structure-specific DNA helicase, is missing from the Buchnera genome, though all other bacteria except chlamydias possess it. RecG is an essential protein and its null mutant of E. coli is particularly morbid (Lloyd and Low 1996). According to a genetic study using a recGnull mutant of E. coli, suppressors of this phenotype, called srgA, have been shown to be alleles of priA and specify mutant PriA proteins. These proteins have singleamino acid replacements in or close to one of the conserved helicase motifs that brought about dysfunction of helicase activity (Al-Deib et al. 1996; Zavitz and Marians 1992). Interestingly, most of the point mutations of E. coli srgA correspond to the Buchnera-specific mutations of PriA (Fig. 3a). This finding suggests that the Buchnera PriA protein has also lost its helicase activity, canceling out the serious effect of the loss of recG from the Buchnera genome, implying that the mutations in PriA of Buchnera have been fixed by positive selection. However, it is unlikely that these mutations of PriA brought about its complete dysfunction, since two intact zinc finger motifs and some conserved regions in this protein still remain. The RecBCD enzyme, a heterotrimer comprised of RecB, RecC, and RecD, is essential for homologous recombination. The prominent biochemical activities of this enzyme include DNA helicase, dsDNA exonuclease, and ssDNA exonuclease (Kuzminov 1999; Lloyd and Low 1996). The DNA-dependent ATPase activity of RecB and RecD is essential for these activities. We found that ATP-binding domains of RecB and RecD were conserved in Buchnera (data not shown), but RecC accumulated many Buchnera-specific mutations (Fig. 1e). It has been reported that E. coli RecBCD mutated at the recC gene lacks Chi recombination hotspot activity,

Fig. 3. Examples of multiple alignments showing the Buchnera-specific mutations at functional motifs. Abbreviations of organism names are buch (Buchnera), ecoli (E. coli), vcho 1 (Viblio cholelae chromosome 1), hinf (Haemophilius influenzae), paer (Pseudomonas aeruginosa), xfas (Xylella fastidiosa), nmen (Neisseria meningtidis), ctra (Chlamydia trachomatis), cpneu (Chlamydia pneumoniae), hpyl (Helicobayter pylori), bsub (Bucillus subtilis), synecho (Synechocystis sp.), tpal (Treponema pallidum), rpxx (Rickettsia prowazekii), aquae (Aquefex aeolicus), mtub (Mycobacterium tuberculosis), dra 1 (Deinococcus radiodurans R1 complete chromosome 1), aful (Archaeoglobus fulgidus), bbur (Borrelia burgdorferi), tmar (Thermotoga maritina), aquae (Aquifex aeolicus), and mjan (Methanococcus jannaschii). In the case where one species has more than one different strain (nmen, hpyl, ctra, cpne), only one sequence is represented. Conserved columns are shaded. a PriA protein. Known mutations of recG suppressor mutant, srgA, of E. coli are shown italicized. b SuhB.

382

383

Fig. 3.

Continued.

though its nuclease activity is nearly the same as that of the wild type (Arnold et al. 1998, 2000). This is reminiscent of the characteristic feature of the Buchnera genome, the AT richness, and the low frequency of Chi sites (5⬘-GCTGGTGG-3⬘) and other repetitive elements (Shigenobu et al. 2000). Another possible reason for the accumulation of mutations in RecC is the absence of RecA in Buchnera. In the homologous recombination system in E. coli, RecBCD functions in concert with RecA, which is a key molecule for homologous recom-

bination (Lloyd and Low 1996). It is likely that in Buchnera the regions of RecC responsible for interaction with RecA, which have not yet been identified, have accumulated mutations after the loss of RecA. These observations imply that RecC of Buchnera is modulated by mutations according to the genome structure and gene repertory. It is expected that the accumulation of mutations in specific regions will lead to the eventual deletion of the functional units. We found a few examples of such cases

384

Fig. 4. Increase in AT content at the first and the second codon base positions by amino acid replacements. The average AT content is used for each amino acid.

in Buchnera proteins. Buchnera PolA retains only the N-terminal region responsible for the 5⬘→3⬘ exonuclease activity but has lost the Klenow region for the polymerase and 3⬘→5⬘ exonuclease activities. It is intriguing that M. genitalium, whose genome is the smallest in bacteria, also has a PolA protein of a similar structure. We found that DnaX, AceF, FtsL, and RnfC of Buchnera had also lost some regions. To understand the mechanisms of the Buchneraspecific mutations, we examined the correlation between amino acid changes and AT bias imposed on the Buchnera genome. Previous analyses using several genes of Buchnera have shown through comparisons with those of E. coli that the bias in nucleotide composition affects the amino acid composition of gene products, resulting in a proportional increase in abundance of amino acids for which corresponding codons contain more A and T in the DNA sequences (Clark et al. 1999; Moran 1996; Ohtaka and Ishikawa 1993; Wernegreen and Moran 1999). However, since E. coli is not an ancestor of Buchnera, it is not guaranteed that the apparent base bias and the difference in the amino acid composition are due to some evolutionary events in the Buchnera lineage. In the present study, we inferred the ancestral protein sequences at the conserved sites and found that the LCA of Buchnera and E. coli had proteins richer in amino acids encoded by high-GC codons than those of Buchnera. This confirmed that a large part of the amino acid replacements in Buchnera proteins resulted in an increase in amino acids encoded by high-AT codons. Taking into account the higher AT content in intergenic regions than that in ORFs in Buchnera, it is concluded that the major driving force of the biased mutations in Buchnera proteins includes AT pressure to the Buchnera genome. Reduction of the genome size of parasites and endosymbionts is due mainly to the loss of genes (Andersson et al. 1998; Fraser et al. 1995; Himmelreich et al. 1996; Shigenobu et al. 2000; Zomorodipour and Andersson 1999). In the present study, we found that the genome of

Buchnera is undergoing reductive evolution through not only loss of genes but also loss of functional units. Another significant finding in this study is that all Buchnera-specific mutations are not deleterious, despite the loss of functional units. Once a protein has been lost from an organism, functional units of proteins that used to interact with the lost protein may be freed from their functional constraints and accumulate neutral mutations

Fig. 5. Preference in amino acid replacements in the proteins of Buchnera and E. coli. a Ratio of the number of amino acid residues changed by Buchnera- or E. coli-specific mutations against the total number of each type of consensus residues. b Number of amino acid residues derived by Buchnera- or E. coli-specific mutation. Statistics of a and b were both calculated using the matrix in Fig. 1.

385

Fig. 6. A hypothetical model for reductive evolution of the genome by sequential loss of functional units. Four proteins are represented as four filled circles and the interactions between the proteins are represented by arrows. Mutated proteins/units are colored in gray, and lost ones are represented as open circles with dotted lines. (1) Accumulation of deleterious mutations in protein D, leading to loss of protein D. (2) Loss of protein D and the interaction with proteins A and C. (3) Accumulation of deleterious or neutral mutations in protein C and the domain of protein A for the interaction with protein D. (4) The consequent deletion of all functional units related to the function of protein D. The functional units that were not related to the function of protein D are intact.

(Fig. 6). Such events could trigger subsequent loss of units that had been involved in a functional network together with the lost units. It should be noted that the mutations causing such a series of loss of functional units triggered by the first loss event can be neutral rather than deleterious if the first loss caused malfunction of the functional network, e.g., a sequential metabolic pathway. Interesting enough, the mutations found in Buchnera PriA seem to be the result of positive selection from the viewpoint that the mutations in PriA overcome the preceded deleterious effect of the recG loss. These losses of specific functions may be a consequence of the changes in the functional constraints on sequences of proteins during Buchnera evolution. Because Buchnera has lost a large number of genes during evolution, interactions among proteins in the Buchnera cell must be different from those in other species. Acknowledgment. We thank T.D. Taylor for suggestions and review of the manuscript.

References Al-Deib AA, Mahdi AA, Lloyd RG (1996) Modulation of recombination and DNA repair by the RecG and PriA helicases of Escherichia coli K-12. J Bacteriol 178:6782–6789 Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ (1997) Gapped BLAST and PSI-BLAST: A new generation of protein database search programs. Nucleic Acids Res 25:3389–3402 Andersson SG, Kurland CG (1998) Reductive evolution of resident genomes. Trends Microbiol 6:263–268

Andersson SG, Zomorodipour A, Andersson JO, Sicheritz-Ponten T, Alsmark UC, Podowski RM, Naslund AK, Eriksson AS, Winkler HH, Kurland CG (1998) The genome sequence of Rickettsia prowazekii and the origin of mitochondria. Nature 396:133–140 Aravind L, Watanabe H, Lipman DJ, Koonin EV (2000) Lineagespecific loss and divergence of functionally linked genes in eukaryotes. Proc Natl Acad Sci USA 97:11319–11324 Arnold DA, Bianco PR, Kowalczykowski SC (1998) The reduced levels of chi recognition exhibited by the RecBC1004D enzyme reflect its recombination defect in vivo. J Biol Chem 273:16476–16486 Arnold DA, Handa N, Kobayashi I, Kowalczykowski SC (2000) A novel, 11 nucleotide variant of chi, chi*: One of a class of sequences defining the Escherichia coli recombination hotspot chi. J Mol Biol 300:469–479 Baumann P, Moran NA, Baumann L (2000) Bacteriocyte-associated endosymbionts of insects. In: Dworkin M (ed) The prokaryotes. Springer, New York Brynnel EU, Kurland CG, Moran NA, Andersson SG (1998) Evolutionary rates for tuf genes in endosymbionts of aphids. Mol Biol Evol 15:574–582 Buchner P (1965) Endosymbiosis of animals with plant microorganisms. Wiley and Sons, New York Chao L (1990) Fitness of RNA virus decreased by Muller’s ratchet. Nature 348:454–455 Charles H, Ishikawa H (1999) Physical and genetic map of the genome of Buchnera, the primary endosymbiont of the pea aphid Acyrthosiphon pisum. J Mol Evol 48:142–150 Clark MA, Moran NA, Baumann P (1999) Sequence evolution in bacterial endosymbionts having extreme base compositions. Mol Biol Evol 16:1586–1598 Douglas AE (1998) Nutritional interactions in insect-microbial symbioses: Aphids and their symbiotic bacteria Buchnera. Annu Rev Entomol 43:17–37 Fraser CM, Gocayne JD, White O, et al. (1995) The minimal gene complement of Mycoplasma genitalium. Science 270:397–403 Henikoff S, Henikoff JG (1992) Amino acid substitution matrices from protein blocks. Proc Natl Acad Sci USA 89:10915–10919 Himmelreich R, Hilbert H, Plagens H, Pirkl E, Li BC, Herrmann R

386 (1996) Complete sequence analysis of the genome of the bacterium Mycoplasma pneumoniae. Nucleic Acids Res 24:4420–4249 Hofmann K, Bucher P, Falquet L, Bairoch A (1999) The PROSITE database, its status in 1999. Nucleic Acids Res 27:215–219 Houk EJ, Griffiths GW (1980) Intracellular symbiotes of the Homopltera. Annu Rev Entomol 25:161–187 Koonin EV, Aravind L, Kondrashov AS (2000) The impact of comparative genomics on our understanding of evolution. Cell 101: 573–576 Kuzminov A (1999) Recombinational repair of DNA damage in Escherichia coli and bacteriophage lambda. Microbiol Mol Biol Rev 63:751–813, Table of Contents Lloyd RG, Low B (1996) Homologous Recombination. In: Neidhardt FC (ed) Escherichia coli and Salmonella. ASM Press, Washington, DC, pp 2236–2255 Marians KJ (1996) Replication fork propagation. In: Neidhardt FC (ed) Escherichia coli and Salmonella. ASM Press, Washington, DC, pp 749–763 Moran NA (1996) Accelerated evolution and Muller’s rachet in endosymbiotic bacteria. Proc Natl Acad Sci USA 93:2873–2878 Ohtaka C, Ishikawa H (1993) Accumulation of adenine and thymine in a groE-homologous operon of an intracellular symbiont. J Mol Evol 36:121–126

Shigenobu S, Watanabe H, Hattori M, Sakaki Y, Ishikawa H (2000) Genome sequence of the endocellular bacterial symbiont of aphids Buchnera sp. APS. Nature 407:81–86 Tekaia F, Dujon B (1999) Pervasiveness of gene conservation and persistence of duplicates in cellular genomes. J Mol Evol 49:591– 600 Thompson JD, Higgins DG, Gibson TJ (1994) CLUSTAL W: Improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22:4673–4680 Watanabe H, Mori H, Itoh T, Gojobori T (1997) Genome plasticity as a paradigm of eubacteria evolution. J Mol Evol 44(Suppl 1):S57– S64 Wernegreen JJ, Moran NA (1999) Evidence for genetic drift in endosymbionts (Buchnera): Analyses of protein-coding genes. Mol Biol Evol 16:83–97 Zavitz KH, Marians KJ (1992) ATPase-deficient mutants of the Escherichia coli DNA replication protein PriA are capable of catalyzing the assembly of active primosomes. J Biol Chem 267:6933–6940 Zomorodipour A, Anderson SG (1999) Obligate intracellular parasites: Rickettsia prowazekii and Chlamydia trachomatis. FEBS Lett 452: 11–15