Bacteriocyte-Associated Endosymbionts of Insects

Baumann, Paul; Moran, Nancy A.; Baumann, Linda

doi:10.1007/0-387-30741-9_16

Paul Baumann,
Nancy A. Moran &
Linda Baumann

4558 Accesses
36 Citations
3 Altmetric

Access provided by Autonomous University of Puebla. Download reference work entry PDF

Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

Intracellular associations between bacteria and insects are widespread in nature (Baumann and Moran, 1997; Buchner, 1965; Dasch et al., 1984; Douglas, 1989; Houk and Griffiths, 1980). Extensive studies of the natural history of such associations have led to the conclusion that they are commonly found in insects that utilize diets containing an excess of one class of compounds but a deficiency of some essential nutrients (Buchner, 1965; Dadd, 1985). It was thought that the function of the endosymbionts was to rectify this imbalance by the synthesis of these essential nutrients for the host. Extensive compilations of the occurrence of endosymbionts in different groups of insects are found in Buchner (1965) and Dasch et al. (1984). Because most of the prokaryotes involved in such associations are not cultivable on common laboratory media, their characterization had to await the development of recombinant DNA methodology. The past 10 years have witnessed the initiation of studies on the intracellular association of prokaryotes with a variety of insect hosts. In this chapter we will provide an overview of the evolution, and where possible, genetics and physiology of such recently studied associations. A summary of some of their features is presented in Table 1, and the phylogeny of the endosymbionts based on 16S rDNA is presented in Fig. 1.

Table 1. General properties of the considered endosymbiotic associations.^a

Full size table

The diversity of symbiotic associations and problems of definitions have been previously discussed and will not be considered here (Smith and Douglas, 1987; Werren and O’Neill, 1997). Some of the phylogenetic studies have included few host taxa and are thus not entirely conclusive; nevertheless, current results suggest that most of the associations considered in this chapter have common features and represent a relatively well-defined type. To aid presentation we will describe these common features, which are established from recent, largely molecular, studies as well as from older investigations based on morphological analyses. References to the earlier studies are found in Buchner (1965), who arrived at similar conclusions. References to recent studies are given as each association is considered.

The associations listed in Table 1 and Fig. 1 are the results of infections of various insect lineages with different prokaryotes. These associations became stable, resulting in the emergence of a new composite (of host and endosymbiont) organism. The endosymbiont became heritable through the acquisition of mechanisms ensuring vertical, maternal transmission to progeny. The association also became obligate, or beneficial, for host growth. Because the host depended on the association, and because horizontal or infectious transmission between hosts did not occur, the phylogeny of the endosymbionts is congruent with the phylogeny of the hosts. With some exceptions, heritable associations tend to become mutualistic (Lipsitch et al., 1995; Werren and O’Neill, 1997). In most cases, the host cannot survive without the endosymbiont, or the elimination of the endosymbiont has a deleterious effect. Although the advantage for the host is in most cases apparent, the advantage for the endosymbiont is not always clear. Perhaps it is more correct to think that the host domesticates the endosymbiont for its own welfare, utilizing functions that are present in the prokaryote but lacking in the host (Douglas and Smith, 1989; Maynard Smith and Szathmáry, 1995). In this chapter, the organism which is present in all the species of an insect group and which appears to be of essential value to the host is designated by either its scientific name or, if one is lacking, by the term primary (P-) endosymbiont.

Superimposed on this fundamental association may be associations with additional endosymbionts. Although these are heritable, they appear to be the result of multiple independent infections, horizontal transmission, or both. Since these endosymbionts may be absent in some hosts, their contribution to the welfare of the organism may not be major or essential. These organisms are designated as secondary (S-) endosymbionts. In this connection it is relevant that some bacterial strains may exist within the body cavity of insects for long periods without obvious deleterious effects, thus serving as potential endosymbiont precursors of endosymbiotic associations (Boman and Hultmark, 1987; Faye, 1978).

These conclusions are tentative and, with the possible exception of the Sitophilus-endosymbiont association, are probably applicable to most of the endosymbiotic associations considered in this chapter. One well-studied, contrasting association is between many arthropods and the intracellular prokaryote, Wolbachia (O’Neill et al., 1997). Although this organism is typically heritable, being transmitted maternally, results of phylogenetic studies imply some incidence of horizontal exchange between very different lineages. Wolbachia causes a number of different reproductive alterations favoring the spread of infected host lineages and has properties characteristic of a parasite (O’Neill et al., 1997). Recent work may necessitate a modification of this conclusion because newly discovered Wolbachia in filarial worms appear to be essential for host survival and may show phylogenetic congruence with their hosts, indicating vertical evolution (Bandi et al., 1998, 1999).

Symbionts of Insects Which Utilize Plant Sap as Food

Aphids, psyllids, whiteflies and mealybugs share a number of common structural and nutritional properties (Borror et al., 1989) and constitute four separate lineages within the suborder Sternorrhyncha (order Homoptera; Campbell et al., 1994; von Dohlen and Moran, 1995). All of these groups feed predominantly or exclusively on plant phloem sap. This mode of life necessitates the penetration of plant tissue by flexible tubular mouthparts (stylets) and the ingestion of plant phloem sap. This diet is unbalanced, as it is rich in carbohydrates but deficient in amino acids and other nitrogenous compounds (Dadd, 1985; Sandström and Pettersson, 1994). Because of the low concentrations of nitrogenous compounds, phloem-feeding insects ingest a large amount of plant sap and then excrete the excess sugar as honeydew. This mode of feeding is conducive to the transmission of plant viruses, and members of the Homoptera are important vectors of viral plant pathogens (Blackman and Eastop, 1984; Gray and Banerjee, 1999; Sylvester, 1985). In addition, these insect populations may reach enormous numbers on plants, causing nutrient deprivation, leaf curling, and gall formation (Borror et al., 1989).

In spite of these common properties, aphids, whiteflies, psyllids, and mealybugs have different prokaryotic P-endosymbionts (Table 1, Fig. 1). These insects, like other animals, require ten essential amino acids, and endosymbionts are thought to upgrade the diet by synthesizing these essential amino acids and providing them to the host (Baumann et al., 1995, 1997a, b; Douglas, 1989; Moran and Telang, 1998). Of these four symbiotic associations, the most extensively studied is that between Buchnera (the P-endosymbiont) and aphids. This association will be considered in some detail and followed by a brief discussion of three other associations.

Aphid Endosymbionts

Buchnera—the Primary Endosymbiont of Aphids

Phylogeny

The initial characterization of the 16S-rDNA gene of Buchnera involved the use of an Escherichia coli 16S rDNA hybridization probe to perform a restriction enzyme and Southern blot analysis on total Acyrthosiphon pisum DNA, which established that this gene was present as a single copy (Unterman et al., 1989). Subsequently three overlapping DNA fragments were cloned and the 16S-rDNA gene was sequenced. In addition, bacteriomes were dissected from the aphid and the DNA purified. Restriction enzyme and Southern blot analysis gave the same results with whole aphid DNA and DNA obtained from dissected bacteriomes, indicating that the bacteriomes were the source of endosymbiont DNA. In all subsequent studies, the 16S rDNA was obtained by PCR amplification using whole aphid DNA preparations cloned into plasmid vectors and then sequenced (Munson et al., 1991b).

Based on 16S rDNA analysis, Buchnera is a distinct lineage within the γ-3 subgroup of the Proteobacteria (Fig. 1; Moran et al., 1993; Munson et al., 1991b; Unterman et al., 1989; van Ham et al., 1997). The closest known organisms are the endosymbionts of carpenter ants, endosymbionts of tsetse flies (Wigglesworthia), and members of the Enterobacteriaceae as defined by Brenner (Brenner, 1984; Aksoy, 1995a, b; Schröder et al., 1996). Phylogenetic analyses based on 16S rDNA indicate that these organisms are four separate lineages but do not permit firm conclusions regarding their relationships to one another. Buchnera contains a single copy of rRNA genes which are arranged as two transcription units, 16S rRNA and tRNAGlu-23S rRNA-5S rRNA (Munson et al., 1993; Rouhbakhsh and Baumann, 1995). This organization of the rRNA genes into two transcription units is somewhat rare but also has been found in Wolbachia (Bensaadi-Merchermerk et al., 1995) and Rickettsia (Andersson et al., 1998), organisms which are in the α-subdivision of the Proteobacteria and also associated with insects. In the endosymbionts of carpenter ants, the rRNA genes are also split into two transcription units (C. Sauer and R. Gross, personal communication) whereas in Wigglesworthia (Aksoy, 1995b) and the Enterobacteriaceae the order is 16S–23S (Berlyn, 1998). The organization of the rRNA genes into two transcription units suggests a possible relationship between Buchnera and the endosymbionts of carpenter ants.

The results of phylogenetic analyses involving all of the currently available Buchnera sequence information are presented in Fig. 2. Most of the characterized endosymbionts are from the family Aphididae. Based on 16S rDNA, Buchnera forms one clade within which two well-supported subclades are apparent. These are the aphids of the Aphididae and the Sc and Mr from the tribe Fordini in the family Pemphigidae. Additional studies using a portion of trpB (Fig. 2b) confirmed some of these relationships and provided further resolution within the genus Uroleucon. These relationships are in broad agreement with the results of evolutionary studies of plasmid-associated trpE, leuBCD, and repA1 (Fig. 2c, d, e; Baumann et al., 1997b, 1999b; Bracho et al., 1995; Rouhbakhsh et al., 1996, 1997; Silva et al., 1998; van Ham et al., 1997, 1999). Within the genus Uroleucon, the phylogeny based on trpB is in good agreement with the more extensive analysis of host phylogeny based on mitochondrial and nuclear genes (Clark et al., 2000; Moran et al., 1999).

Table 2. Abbreviations of aphid species used in this chapter.

Full size table

The congruence of phylogenies derived from Buchnera chromosomal and plasmid genes, as well as host mitochondrial and nuclear genes, is strong evidence for a vertical mode of evolution with no exchange of either bacteria or plasmids among host lineages (Moran and Baumann, 1994). An implication of the congruence between the phylogenies of Buchnera and corresponding aphid hosts is that dates for branch points inferred from fossil aphids can be extended to ancestral Buchnera (Moran et al., 1993). A further implication is that modern Buchnera descend from an infection of a common ancestor of all modern aphids. From the aphid fossil record, we can infer that this infection by a free-living bacterium must have occurred at least 150–250 million years ago. The divergence in 16S rDNA of modern Buchnera is consistent with this hypothesis of an ancient infection followed by cospeciation of Buchnera and hosts.

Buchnera shows a rate of base substitution in its 16S rDNA that is about twice as great as that in related free-living bacteria based both on relative rate comparisons with free-living taxa and on comparisons of rates calibrated with respect to absolute time (Clark et al., 1999b; Moran, 1996). The elevated substitution rate of Buchnera relative to that of related free-living bacteria extends to genes encoding proteins (Brynnel et al., 1998; Clark et al., 1999b; Moran, 1996; Wernegreen and Moran, 1999). Based on calibrated rates for protein-coding genes, synonymous sites evolve about twice as fast and nonsynonymous sites about six times as fast in Buchnera as in E. coli/Salmonella typhimurium, based on an absolute time scale (Table 3). The rate differences are considerably greater on a scale based on generations, since Buchnera appears to have fewer generations per year than do natural populations of enteric bacteria (Clark et al., 1999b).

Table 3. Substitution rates in Buchnera and enteric bacteria.^a

Full size table

The most plausible explanation for the faster rate of substitution in Buchnera is that the population structure of Buchnera, involving strictly vertical transmission of a small inoculum between hosts, results in greater levels of genetic drift, which can increase the fixation rate of mildly deleterious mutations. Several observations support this explanation. First, the rate increase is found at all genes and is concentrated at sites, such as nonsynonymous sites, that are expected to be under selective constraint (Moran, 1996; Wernegreen and Moran, 1999). Second, polypeptide compositions are consistently biased towards amino acids that allow more adenine and thymine in the DNA sequence, indicating that mutational bias has affected protein evolution. Third, faster substitution rates in 16S rDNA are observed in other insect endosymbionts that share a similar transmission mode, suggesting that the endosymbiotic lifestyle has repeatedly produced the same changes in patterns of molecular evolution. Finally, the 16S rRNA secondary structure of Buchnera and other endosymbionts has lower thermal stability than that of related free-living bacteria, as expected if the DNA base substitutions are mildly deleterious (Lambert and Moran, 1998).

The 16S rDNA substitution rate of Buchnera is about 35 times greater than that of homologous regions of 18S rDNA of hosts, based on comparisons of pairwise divergences of corresponding aphid and Buchnera taxa (Moran et al., 1995). Thus, the hypothesis of a universal rate of substitution in rDNA is not even approximately true.

Taxonomy

The genus Buchnera contains one species, Buchnera aphidicola, and the type strain is the endosymbiont of the aphid Schizaphis graminum (Munson et al., 1991a). Currently this species name designates the lineage consisting of the P-endosymbionts of aphids. There are over 4,000 species of aphids (Blackman and Eastop, 1984; Remaudière and Remaudière, 1997) of which only 35 have been characterized by molecular methods. Consequently our conclusions are based on a very small sample of aphid species. Although 16S rDNA has been useful for showing the monophyletic origin of aphid endosymbiosis and the establishment of major aphid subgroups, it is far too conserved to be useful for defining relationships among endosymbionts of closely related aphids. Some success has been obtained by the use of other, less conserved, molecules (Fig. 2). The 16S rDNA sequence difference of Buchnera in Sg and Sc (the most distantly related aphids) is about the same as that between E. coli and Proteus vulgaris. Thus, subsequent studies using less conserved molecules will probably indicate that Buchnera should be subdivided into new species. So far, no studies have addressed the range of variation within endosymbionts of a single aphid species.

Habitat

Location and Ultrastructure

During their reproductive phase, aphids contain within their body cavity a bilobed structure called a bacteriome consisting of 60–90 uninucleate, polyploid cells called bacteriocytes (Douglas and Dixon, 1987). These cells are filled with host-derived vesicles containing Buchnera (Fig. 3a). This organism is spherical or oval in shape, 2–4 m in diameter, with a cell wall consisting of two-unit membranes, as is characteristic of Gram-negative bacteria (Fig. 3b; Akhtar and van Emden, 1994; Griffiths and Beck, 1973; Hinde, 1971b; McLean and Houk, 1973). A thin layer corresponding to peptidoglycan has been detected (Houk et al., 1977). The presence of peptidoglycan also is indicated by penicillin-induced alterations of the cell wall as well as by chemical analysis (Griffiths and Beck, 1974; Houk et al., 1977).

Number of Endosymbionts

Buchnera contains one copy of the 16S rRNA gene (Munson et al., 1991b, 1993; Unterman et al., 1989) and one copy of groEL (Ohtaka et al., 1992; Hassan et al., 1996). The number of copies of Buchnera 16S rRNA genes in the aphid Sg was studied by quantitative PCR (Baumann and Baumann, 1994) and was found to be 0.5–1.2 × 10⁷ mg⁻¹ aphid wet weight. Using quantitative hybridization of a Buchnera groEL probe, the number of genome copies in Ap was estimated at 1–2 × 10⁷ mg⁻¹ aphid wet weight (Humphreys and Douglas, 1997). In both of these studies the number of Buchnera cells was assumed to be identical to the number of genome copies. However a recent study has demonstrated that Buchnera (Ap) is polyploid, containing an average of about 120 genomes per cell (Komaki and Ishikawa, 1999). If the average number of genomes per endosymbiont is relatively constant, then the number of endosymbionts is about 100-fold less than estimated previously, or about 105 mg⁻¹ aphid wet weight. This value is considerably lower than the estimates for Ap of 1.6–1.8 × 10⁶ endosymbionts mg⁻¹ aphid wet weight, based on microscopic enumeration of the endosymbionts (Harrison et al., 1989).

Growth and Reproduction

In their most active stage, aphids are wingless females, which reproduce by parthenogenesis, giving birth to live young. There is telescoping of generations in that the mother aphid contains embryos that, in turn, may contain other embryos (Dixon, 1973, 1992). Studies on the growth of Sg (Baumann and Baumann, 1994) have indicated that newly born aphids weigh 24 µg and contain 2 × 10⁵ copies of the Buchnera genome. The increase in the number of endosymbiont genomes parallels the increase in the weight of the aphid. The maximum weight is reached in 9–10 days at which time the aphid weighs 540 µg and contains 5.6 × 10⁶ Buchnera genomes. The endosymbionts are partitioned between maternal and embryonic bacteriocytes. In a mature aphid, most of the Buchnera genomes are found in the embryos (Humphreys and Douglas, 1997). The first young are born in 8–9 days; each aphid can produce 60 to 80 live young during its lifetime. Douglas and Dixon (1987) showed that, during the period of growth, there is a concomitant increase in the maternal bacteriocyte volume as well as a small drop in bacteriocyte number. In the adult aphid the number of maternal bacteriocytes in the bacteriome undergoes a sharp decrease probably due, in part, to their dispersion within the abdomen, their degradation, as well as the degradation of Buchnera (Brough and Dixon, 1990; Douglas and Dixon, 1987; Griffiths and Beck, 1973; Hinde, 1971a).

Aphids may also produce sexual forms with the females depositing eggs that overwinter and hatch in the spring. Buchnera is maternally transmitted (transovarial transmission) to both developing embryos and eggs. Maternal bacteriocytes adjacent to an embryo near the blastoderm stage form a small opening through which the endosymbionts pass. Buchnera then moves through the intervening hemolymph and enters a nearby opening on the oocyte surface. During early embryonic development, the presumptive bacteriocytes form, and the endosymbionts migrate into these cells (Buchner, 1965; Blackman, 1987; Hinde, 1971a). Symbiont invasion of eggs also occurs from the dispersed bacteriocytes and they can be observed as an aggregate at the posterior pole of the mature egg (Buchner, 1965; Brough and Dixon, 1990). Buchnera and bacteriomes appear to be nearly universal in aphids (Buchner, 1965). However, some species of the tribe Cerataphidini lack both and instead contain yeast-like extracellular symbionts within their body cavity (Buchner, 1965; Fukatsu and Ishikawa 1992a, 1996). Some species of aphids may produce dwarf males and/or sterile female soldiers that may also lack endosymbionts (Buchner, 1965; Fukatsu and Ishikawa, 1992b; Fukatsu et al., 1994).

Physiology

Nutrition and Metabolism

Plant sap, the diet of aphids, has an excess of carbohydrate relative to amino acids and other nitrogenous compounds (Dadd, 1985; Douglas, 1998; Sandström and Pettersson, 1994; Sandström and Moran, 1999). Aphids, like other insects, are thought to require 10 preformed amino acids and these essential amino acids are present in low amounts in plant sap. Some species of aphids can grow on synthetic diets even in the absence of essential amino acids. Adding antibiotics to such diets results in the elimination of endosymbionts and the failure of the aphids to reproduce. There is some sparing effect when the essential amino acids are included in the antibiotic-containing diet. These experiments have generally been interpreted as indicating that one of the functions of Buchnera is the synthesis of essential amino acids for the aphid host (reviewed by Baumann et al., 1995; Douglas, 1998). A major problem is that compared with growth on plants, growth on artificial diets is poor and generally limited to a few generations. In addition, aphid growth on complete synthetic diets in the presence of antibiotics is even worse, indicating that Buchnera provides nutrients or functions that cannot be provided by the artificial diets. The effects of antibiotics on a number of aphid properties have been recently reviewed (Wilkinson, 1998). There is a further complication with some of the nutritional studies, in which physiological effects have been attributed to the removal of Buchnera. The aphid strain used may also contain S-endosymbionts, as is true of the strain of the aphid Ap used in the studies of A. E. Douglas and her collaborators (Douglas and Prosser, 1992; Wilkinson 1998). The S-endosymbiont probably does not perform any essential functions for the host (see section on S-endosymbionts of aphids in this Chapter); nevertheless, the use of antibiotics eliminates both endosymbionts, and consequently the observed effects of this loss may not be attributable solely to the loss of Buchnera.

Currently one of the more complete studies involves the essential amino acid tryptophan. Using a strain of Ap containing an S-endosymbiont, Douglas and Prosser (1992) have shown a sparing effect of tryptophan in chlortetracycline-containing synthetic diets on aphid growth. In addition they detected low levels of tryptophan synthase in Buchnera and found that activity was absent in chlortetracycline-treated and thus endosymbiont-free aphids. The assays used (Smith and Yanofsky, 1962) crude extracts of whole aphids as well as preparations enriched in endosymbionts and measured the disappearance of the substrate indole and not the appearance of the product tryptophan. Indole or indole derivatives may be substrates for a variety of reactions catalyzed by enzymes found in crude extracts of insects. No information was provided about the linearity of increasing enzyme activity with increasing crude extract concentration. In spite of possible difficulties with this assay, the dependence of the reaction on “the substrate [sic] pyridoxal phosphate” (Douglas and Prosser, 1992) is consistent with it being a measure of tryptophan synthase activity.

Using synthetic diets containing radiolabelled sulfate, it was shown that Buchnera can reduce this compound to the level of hydrogen sulfide and incorporate it into methionine and cysteine and that these amino acids are found in aphid tissue (Douglas, 1988). Using ¹⁴C-radiolabelled amino acids, it was found that the synthesis of the essential amino acids arginine, threonine, isoleucine and lysine was reduced or eliminated by the inclusion of rifampicin in the diet (Liadouze et al., 1996). Sasaki and Ishikawa (1995) also showed that treatment of aphids with rifampicin eliminated the incorporation of dietary ¹⁵N-glutamine into the essential amino acids arginine, histidine, isoleucine and/or leucine, phenylalanine, threonine, and valine.

Glutamine is the predominant amino acid in phloem and also in aphid hemolymph (Sandström and Pettersson, 1994; Sasaki et al., 1990). Isolated bacteriocytes were found to take up glutamine and convert it to glutamate, which subsequently was taken up by Buchnera (Sasaki and Ishikawa, 1995). Isolated endosymbionts incorporated the nitrogen of glutamine into the essential amino acids isoleucine, leucine, valine, and phenylalanine as well as a number of other amino acids, and these amino acids were excreted into the suspending medium. Whitehead and Douglas, however, could not find any evidence for excretion of amino acids by Buchnera (cited in Douglas, 1997).

Using synthetic diets, Nakabachi and Ishikawa (1999) demonstrated a requirement for riboflavin by rifampicin-treated aphids. These results indicate that Buchnera is required for the synthesis of at least one vitamin for the aphid host.

Whitehead and Douglas (1993) isolated vesicles containing Buchnera and showed that they readily took up acetic, glutamic and aspartic acid as well as tricarboxylic acid cycle intermediates and oxidized them to CO₂. Oxygen consumption was also detected and was greatly reduced by KCN. These results suggest the presence of a tricarboxylic acid cycle in the endosymbionts and indicate a respiratory metabolism. The latter conclusion is consistent with the presence of a gene for a subunit of NADH dehydrogenase I, an enzyme involved in the generation of a proton motive force during respiration, and of all the genes for ATP synthase, a membrane-associated enzyme which utilizes the proton motive force for the synthesis of ATP (Table 4).

Table 4. Genes of Buchnera from the aphid S. graminum.^a

Full size table

Table 5. Order of genes on DNA fragments of Buchnera from the aphid S. graminum.^a

Full size table

Gene Expression

Buchnera messenger RNA (mRNA) has been detected for a variety of genes encoding proteins involved in amino acid biosynthesis (Table 4). This includes genes for amino acids of the glutamate (argA) and aspartate (thrB) families (Nakabachi and Ishikawa, 1997), the shikimate pathway (aroH), as well as the biosynthetic pathway for tryptophan (trpE, trpD, trpA), branched-chain amino acids (ilvI, ilvD, leuA), and histidine (hisG; Baumann et al., 1999a). Buchnera mRNA has been detected also for a gene involved in the biosynthesis of riboflavin (ribE; Nakabachi and Ishikawa, 1999) as well as for genes involved in the heat shock response (groEL, groES, dnaK, dnaJ; Table 4; Sato and Ishikawa, 1997a, b). Numerous Buchnera proteins have also been detected by immunological methods. These include GroEL, GroES, and DnaK (Kakeda and Ishikawa, 1991; Sato and Ishikawa, 1997b) as well as ribosomal protein S1 (the product of rpsA; Clark et al., 1996) and the protein involved in septum formation during cell division (the product of ftsZ; Table 4; Baumann and Baumann, 1998).

In bacteria, rRNA genes are transcribed from strong promoters. Comparisons of the regions upstream of rRNA genes from six species of Buchnera indicated conservation of sequences resembling the −35 and −10 regions of σ⁷⁰ promoters as well as boxA and boxC (Munson et al., 1993; Rouhbakhsh, 1995). Similar putative −35 and −10 regions were found in Buchnera plasmids containing genes for tryptophan and leucine biosynthesis (Baumann et al., 1999b; Rouhbakhsh et al., 1996; Silva et al., 1998).

GroEL Overproduction and Its Significance

In Buchnera the chaperonin, GroEL, constitutes a major fraction of the total protein (Sato and Ishikawa, 1997a). In addition, GroEL is present in aphid hemolymph (van den Heuvel et al., 1994). Overproduction of GroEL is a characteristic of some endosymbionts and pathogens in the intracellular environment (Hogenhout et al., 1998). This protein mediates the folding of peptides into their functional forms as well as the repair of damaged proteins (Gross, 1996). Buchnera GroEL is able to complement E. coli mutants (Ohtaka et al., 1992). GroEL has been localized in maternal and embryonic Buchnera by immunohistochemistry (Fukatsu and Ishikawa, 1992c). Electron micrographs indicate that the purified Buchnera GroEL has the characteristic double-ring appearance observed with the E. coli protein (Filichkin et al., 1997; Hara and Ishikawa, 1990). The endosymbiont protein has ATPase activity and in the presence of E. coli, GroES could reconstitute denatured Rhodospirillum rubrum ribulose-1,5-biphosphate carboxylase (Kakeda and Ishikawa, 1991).

The Buchnera groESL operon organization resembles that of E. coli (Hassan et al., 1996; Hogenhout et al., 1998; Ohtaka et al., 1992). Upstream of groES are nucleotide sequences characteristic of the −35 and −10 regions of σ³² promoters. A message of 2.1 kb (containing both groES and groEL) is made by the endosymbiont using only this promoter (Sato and Ishikawa, 1997a). It is not understood why GroES is low in the endosymbiont, in contrast to the high quantities of GroEL (Kakeda and Ishikawa, 1991). The genes for σ³² (rpoH) as well as dnaKJ have been cloned and sequenced (Sato and Ishikawa, 1997a, b). The latter also are transcribed solely from a σ³² promoter. In E. coli as well as other organisms, transcription of the groESL operon and the dnaKJ operon is part of the σ³² regulon, and their synthesis is increased by heat shock (Gross, 1996). It would appear that this mode of regulation is modified in Buchnera (Sato and Ishikawa, 1997a, b). Synthesis of groESL and dnaKJ mRNA is constitutive and is not increased by heat shock. This conclusion is supported by the observation that there is no increase in the level of total GroEL in aphids shifted from 23 to 33°C for one day (Baumann et al., 1996).

Baumann et al. (1996) arrived at an estimate of the amount of GroEL per Buchnera cell, based on the premise that the endosymbiont contained only one genome copy. Recently it has been shown that Buchnera is polyploid, containing an average of 120 genome copies per cell (Komaki and Ishikawa, 1999). If this result is incorporated into the calculation, an impossible excess of GroEL would be present in each endosymbiont cell. Since protein extracts of whole aphids were used in the estimation of GroEL content, this result could be explained by the finding that GroEL is also present in the hemolymph (van den Heuvel et al., 1994).

Perhaps the major economic effect of aphids on agriculture is their ability to transmit plant viruses (Blackman and Eastop, 1984; Gray and Banerjee, 1999; Sylvester, 1985). Buchnera-derived GroEL has been implicated in the survival of luteoviruses in the hemolymph (Filichkin et al., 1997; van den Heuvel et al., 1994; Hogenhout et al., 1998). These viruses replicate in the plant and are ingested by aphids when they feed on phloem sap. Subsequently they are transported from the digestive tract into the hemolymph and from there, into the salivary gland for transmission to plants via salivary secretions. The viruses are retained in an infective form (without replication) in the hemolymph throughout the life span of the aphid. There is evidence that the GroEL that is found in the hemolymph coats the virus particles and protects them from host defenses. A region in Buchnera GroEL has been identified which is essential for binding to the virus (Hogenhout et al., 1998), and similarly a portion of a viral capsid protein has been identified as the region to which the endosymbiont GroEL binds (van den Heuvel et al., 1997). Transmission of plant viruses may be advantageous to the aphid because infected plants have higher levels of nutrients in their sap (Blua et al., 1994).

An additional reason for the constitutive synthesis of high amounts of GroEL by Buchnera may be to compensate for the accumulated amino acid substitutions which have occurred at a high rate in this endosymbiont (Moran, 1996; Table 3). These slightly deleterious changes may result in proteins of decreased stability, and the high levels of GroEL may compensate for these changes, allowing proper folding and retention of function. A similar role for chaperones in masking deleterious mutations has been recently suggested on the basis of work on Drosophila heat shock protein (Pennisi, 1998; Rutherford and Lindquist, 1998).

Genetics

Genome Analysis

The guanine + cytosine (G+C) content of Buchnera is about 28 mol% (Clark et al., 1998a; Ishikawa, 1987). The genome size of the endosymbiont from the aphid Ap has been found to be 0.657 Mb (Charles and Ishikawa, 1999). This is considerably below such free-living organisms as E. coli (4.6 Mb; Blattner et al., 1997) and Haemophilus influenzae (1.8 Mb; Fleischmann et al., 1995) and the intracellular pathogens, Chlamydia trachomatis (1.0 Mb; Stephens et al., 1998) and Rickettsia prowazekii (1.1 Mb; Andersson et al., 1998). The Buchnera genome is somewhat larger than that of the pathogen Mycoplasma genitalium (0.58 Mb; Fraser et al., 1995). An unusual property of the Buchnera genome is that it appears to be present as about 120 copies per cell (Komaki and Ishikawa, 1999).

Approximately 130 kb of DNA have been sequenced from Buchnera (from Sg) (Baumann and Baumann, 1998; Baumann et al., 1995; Clark et al., 1998a, b; Thao and Baumann, 1998). The choice of this aphid was predicated on the fact that it contains only one endosymbiont as indicated by morphological examinations as well as extensive restriction enzyme and Southern blot analysis of whole aphid DNA, using probes for many different genes. The latter results indicated that, with the exception of plasmid amplified DNA, only one copy of the targeted genes was present in the endosymbiont genome. In many cases, the DNA that was used for cloning was also endosymbiont-enriched. There were many independently cloned DNA fragments with identical overlapping sequences, indicating that the aphids did not contain several closely related endosymbionts. At least 20 kb of DNA also have been sequenced from Buchnera of each of the aphids Dn, Sc and Mr (Baumann et al., 1998a; Clark et al., 1999b; Lai et al., 1995, 1996). Table 4 lists, under different functional categories, the genes found in Buchnera, primarily in (Sg). The order of these genes in the DNA fragments is presented in Table 5. A total of 126 open reading frames were detected, of which 101 corresponded to E. coli genes with known function. The remaining 25 open reading frames all had homologs of no known function in the E. coli chromosome (Blattner et al., 1997). Table 6 presents the codon usage of the Buchnera structural genes. As expected from the G+C content, there is a strong bias for A and T, especially in the third codon position. This bias also affects the composition of proteins, favoring amino acids for which codons contain more A and T (Moran, 1996; Clark et al., 1999b).

Table 6. Codon usage of Buchnera from the aphid S. graminum.^a

Full size table

Buchnera was found to contain dnaA, encoding a protein which initiates bidirectional chromosome replication, and ftsZ, encoding a protein involved in septum formation during cell division (Baumann and Baumann, 1998; Lai et al., 1992a). Among other genes that were found are those encoding proteins for: peptidoglycan synthesis, cell division, DNA replication, DNA transcription, ribosomal proteins, amino acid tRNA synthases, ATP synthase, electron transport, protein secretion and glycolysis. In addition, genes for three tRNAs were detected. Genes encoding homologs of proteins involved in the E. coli heat shock response (groEL, groES, htrA, dnaK, dnaJ) and the cold shock response (hscA, hscB) were also detected (Clark et al., 1998a; Hassan et al., 1996; Ohtaka et al., 1992; Sato and Ishikawa, 1997b). Nakabachi and Ishikawa (1999) detected a gene (ribE) encoding a protein involved in riboflavin biosynthesis. In addition some of the genes encoding enzymes for the biosynthesis of aromatic amino acids (shikimate pathway, tryptophan branch), branched-chain amino acids (isoleucine, valine, leucine), lysine, cysteine and serine as well as genes for the complete pathway of histidine biosynthesis were found (Clark et al., 1998a, b; Thao and Baumann, 1998). The presence of genes for enzymes of amino acid biosynthesis is in marked contrast to the obligate intracellular pathogens Rickettsia prowazekii and Chlamydia trachomatis, as well as such fastidious organisms as Mycoplasma genitalium and Borrelia burgdorferi, all of which lack genes encoding enzymes of amino acid biosynthesis (Andersson et al., 1998; Fraser et al., 1995, 1997; Stephens et al., 1998). Retention of amino acid biosynthetic genes by Buchnera probably reflects the role of these pathways in the mutualistic association with the host aphids. Overall, these results on gene content indicate that Buchnera has many of the properties of free-living bacteria and would appear to be, in many respects, a self-contained, physiologically autonomous unit enclosed within bacteriocyte-derived vesicles.

Currently the most interesting comparison of the Buchnera genome is with the recently sequenced obligate intracellular pathogen R. prowazekii, an organism which is a member of the α-subdivision of the Proteobacteria (Andersson et al., 1998). Compared to the sequenced bacterial genomes, this organism is unusual in that 24% of its DNA is noncoding. In addition it has a number of pseudogenes. These findings are interpreted as a stage in the adaptation of R. prowazekii to an intracellular lifestyle, involving the loss of genes encoding metabolic pathways for products that are provided by the host. Currently about 20% of the Buchnera genome has been sequenced, and the organization of the genes in these genome fragments is highly compact and similar to that of other bacteria (Clark et al., 1998a). These findings indicate that if, as seems probable, Buchnera originated from an organism with a larger genome (Charles and Ishikawa, 1999), then the reduction in the genome size has already been accomplished and what has been retained is the essential gene complement required for the endosymbiotic association.

Plasmid-Associated Amplification of Biosynthetic Genes

Some species of aphids have Buchnera in which genes for enzymes of amino acid biosynthesis are amplified on plasmids (Bracho et al., 1995; Lai et al., 1994; van Ham et al., 1999). This plasmid-associated gene amplification has been interpreted as an adaptation of Buchnera to an endosymbiotic association in which one of its functions is the overproduction of amino acids. This interpretation is based on analogies with other prokaryotic systems in which gene amplification is viewed as an attribute of prokaryotic genome plasticity allowing the organism to adapt to new environments (for recent reviews see Romero and Palacios, 1997; Roth et al., 1996). Currently Buchnera has been found to contain three different types of plasmids. The properties of these plasmids are summarized in Table 7. The plasmid-amplified genes encode the first enzyme of the tryptophan biosynthetic pathway (TrpEG) and four enzymes (LeuA, LeuB, LeuC, LeuD) of the leucine portion of the branched-chain amino acid biosynthetic pathway. For purposes of grouping of plasmid types and ease of presentation, we have used the plasmid designations given in Table 7 followed by the abbreviation corresponding to the aphid species (Table 2). When first used, the designation is followed by the original plasmid name given in parentheses (if applicable). There are a number of other examples of plasmid amplification of biosynthetic genes. Amplification has been interpreted as a means of increasing the rate of synthesis of endproducts. These include genes of purine biosynthesis in Borrelia (Margolis et al., 1994), cysteine biosynthesis in Synechococcus (Nicholson et al., 1995), and histamine biosynthesis in Vibrio (Barancin et al., 1998).

Table 7. Designations and characteristics of Buchnera plasmids.^a

Full size table

The trpEG-Containing Plasmids and Gene Silencing

Figure 4 is an outline of the aromatic amino acid biosynthetic pathway. It consists of a common portion leading to chorismate (shikimate pathway) and branches to 1) phenylalanine and tyrosine as well as 2) tryptophan. In the shikimate pathway, arrows that have designations correspond to genes detected in Buchnera (Sg). The activity of the tryptophan biosynthetic pathway is regulated by anthranilate synthase (TrpEG) which is feedback inhibited by tryptophan (Crawford, 1989). In Buchnera from 11 species of aphids, trpEG has been found to be plasmid-associated (Fig. 5;Baumann et al., 1997b; Lai et al., 1994, 1996; Rouhbakhsh et al., 1996, 1997; van Ham et al., 1999). The remaining genes of the pathway [trpDC(F)BA] have been found to have a chromosomal location in all cases examined [Buchnera (Sg, Dn, Sc, Mr)] (Baumann et al., 1998a; Clark et al., 1999b; Lai et al., 1995). In contrast to the situation in Aphididae, in Buchnera (Sc, Mr), trpEG is not plasmid-associated but is present as one copy on the endosymbiont chromosome (Fig. 4;Clark et al., 1999b; Lai et al., 1995).

The structure of plasmids of the pTrpEG-type usually consists of tandem repeats of a nearly identical unit (Fig. 5a–h). In Buchnera (Sg, Rp) the plasmids contain four tandem repeats of a 3.6 kb unit, in Buchnera (Rm) the plasmid consists of one 3.6 kb unit, while in Buchnera (Ap) plasmids containing 5, 6, or 10 tandem repeats are found. Buchnera (Sg) contains about 4 plasmids per endosymbiont chromosome resulting in a 16-fold trpEG amplification. In Buchnera (Sc, Mr), in which trpEG is chromosomal, these two genes are found between fpr and hslU (Fig. 6). In Buchnera (Sg), in which trpEG is plasmid-associated, trpEG is absent from this chromosomal location, consistent with its transfer to the plasmid (Fig. 6). Instead ibp is present at this position, suggesting a concomitant or a subsequent acquisition of this gene (Clark et al., 1999a).

The plasmids in Fig. 5a–h all share in common the presence of 2–5 DnaA boxes which are 9 nt-long sequences to which the DnaA protein binds, thereby initiating chromosomal replication (Messer and Weigl, 1996). There is also considerable variation in the length of the repeated units (2.6 to 3.6 kb). Within this plasmid group, a readily recognized subset contains a unique arrangement of three DnaA boxes and a conserved region upstream of trpEG which has been designated as ori-3.6 (Fig. 5a–c; Lai et al., 1996; Rouhbakhsh et al., 1996). Buchnera in aphids of the genus Uroleucon (which is derived from Buchnera within the cluster that has ori-3.6-containing plasmids, Fig. 2b, c) have trpEG units which show a considerable size range and substantial differences in the arrangement of the DnaA boxes (Fig. 5d–g; Baumann et al., 1997b; Rouhbakhsh et al., 1997). All of these Buchnera are from aphid species of the family Aphididae (Fig. 2a). Plasmid pTrpEG-Tc (pBTc2; Fig. 5h) is from an aphid within the family Pemphigidae (van Ham et al., 1999). The predominant form of this plasmid consists of one 3.0 kb unit; a minor form consists of two units. A distinctive feature is the presence of DnaA boxes between trpE and trpG (Fig. 5h) and not upstream of trpEG as is the case of the other plasmids (Fig. 5a–g). However, since there is considerable rearrangement of the DnaA boxes within plasmids in Buchnera of Uroleucon (Fig. 5d–g), it is plausible that the arrangement of DnaA boxes within pTrpEG-Tc is not a fundamental difference but a variation on the arrangement observed in the other trpEG-containing plasmids (Fig. 5a–g).

A totally different trpEG-containing plasmid is pTrpEG-R-Ps (pBPs2; Fig. 5i), which does not contain DnaA boxes but instead has putative replication initiation proteins (RepAC) which are related to those of plasmids of the broad host-range group IncA/C (van Ham et al., 1999). Within the DNA encoding the C-terminal portion of repAC and/or downstream of it are 4–12 repeats of a 19 nt-long sequence corresponding to a putative interon. In addition, there is a single copy of a 19 nt-long sequence similar to the interon sequence of IncAC plasmid RA1 (Llanes et al., 1996). Curiously pTrpEG-Rm (Fig. 5b) contains in its DNA a gene for a putative RepAC protein and the 19 nt-long sequence similar to that found in plasmid RA1, while pTrpEG-Rp from a closely related aphid has a remnant of repAC (van Ham et al., 1999).

In several pTrpEG plasmids, the expression of most of the trpEG copies appears to be silenced (Baumann et al., 1997b; Lai et al., 1996; van Ham et al., 1999). In Buchnera (Dn) there are about two copies of pTrpEG-Dn for each endosymbiont genome (Lai et al., 1996). Plasmid pTrpEG-Dn consists of a single 3.2 kb unit containing an open reading frame corresponding to the putative protein TrpEG (Fig. 7a). This is followed by a 2.6 kb unit containing trpEG pseudogenes and 5–7 repeats of a 3.2 kb unit also containing trpEG pseudogenes. (By pseudogenes, we mean segments of DNA which are clearly recognizable as trpEG but which contain numerous frameshifts and stop codons preventing the synthesis of an intact protein.) A comparison of the sequences between the 3.2 kb fragments with and without pseudogenes indicated 244 differences of which 93% were localized in an approximately 900-bp DNA segment which included the putative promoter and the N-terminus of trpE (Fig. 7a). These changes should result in the reduction or elimination of mRNA synthesis; if messenger is made, it would be translated into short peptides because of the numerous frame shifts and stop codons in the region of the N-terminus of trpE.

Another instance of trpEG silencing is illustrated by pTrpEG-Usn (Fig. 7b). This plasmid consists of a 3.2 kb unit of trpEG followed by 10–14 2.1 kb units consisting of DNA with a deletion of about 56% of the N-terminal region of trpE and an intact trpG. A more remarkable example of gene silencing is found in pTrpEG-R-Ps (Fig. 7c), which consists of a 3.6–3.8 kb unit containing repAC2-trpEG followed by 1.8 kb units consisting of repAC1, a deletion of trpE, and an intact trpG (van Ham et al., 1999). Preceding the latter is a short DNA segment which appears to be a remnant of DNA encoding the C-terminus of trpE. The structure of this plasmid suggests that, in Buchnera (Ps) of the family Pemphigidae, trpEG amplification had an origin independent of that of the remaining plasmids shown in Fig. 5. The initial plasmid probably contained tandem repeats of repAC-trpEG. Subsequently there was selection pressure for gene silencing resulting in a plasmid with one intact copy of repAC-trpEG and tandem repeats of repAC-trpG which contain a deletion of the putative promoter region and most of trpE.

Speculation Concerning trpEG Amplification

The following is a summary of the results obtained from studies of trpEG in Buchnera. 1) trpEG amplification is widespread in Buchnera within the family Aphididae and is also present in at least two members of the family Pemphigidae. In Buchnera from two aphid species of the latter family, trpEG is chromosomal and is found in the same location. In most cases trpEG amplification is affected by plasmids consisting of tandem repeats of the same or similar unit. 2) Evolution of plasmid-associated trpEG is vertical, that is, Buchnera from different aphid species do not exchange plasmids. 3) The trpEG-containing plasmids constitute at least two replicon types, one of which is based on the presence of DnaA boxes, while the other is based on the presence of interons and replication-initiation proteins related to plasmids of the IncA/C group. Buchnera from the related aphids Rm and Rp which belong to the first replicon type also have a gene and/or the remnant of a gene for a replication-initiation protein related to plasmids of the IncA/C group. 4) In trpEG plasmids of both replicon types, gene silencing of some of the trpEG tandem repeats is observed. 5) In Buchnera (Sg) the sequence of two of the 3.6-kb units is virtually identical, as is the sequence of the 2.6 and 3.2-kb trpEG pseudogene-containing units of Buchnera (Dn). While results above are derived from genetic analyses, the interpretations that follow are speculative and based in large part on analogies with other prokaryotic systems.

Gene amplification is currently viewed as a reversible aspect of genome plasticity which occurs at a frequency considerably higher than that of mutation in structural genes (Romero and Palacios, 1997; Roth et al., 1996). Gene amplification is frequently used by an organism as a means of increasing the amount of a growth-limiting enzyme to levels beyond that achieved by gene regulation of expression. One of the functions of Buchnera is the synthesis of essential amino acids (including tryptophan) for the aphid host. In almost all prokaryotes, the limiting enzyme TrpEG (which is feedback inhibited by tryptophan) regulates the activity of the tryptophan biosynthetic pathway. The Buchnera enzyme is probably also feedback inhibited by tryptophan since trpE has the conserved amino acid residues that are involved in feedback inhibition (Lai et al., 1994). To overproduce tryptophan, the activity of TrpEG must be increased, the potential effect of tryptophan accumulation on activity must be overcome, or both. Since even in high concentrations of tryptophan the activity of TrpEG is not fully inhibited, an increase in enzyme protein will result in increased tryptophan production. This has been the case in experiments on overproduction of tryptophan for industrial purposes; an increase in allosterically inhibitable TrpEG is the primary means of achieving excretion of high amounts of tryptophan into the medium (Katsumata and Ikeda, 1993).

In free-living bacteria, gene amplification is frequently transient. Its persistence depends on a constant selective pressure, the absence of which leads to a rapid decrease in number of repeats primarily by means of RecA-mediated homologous recombination (Roth et al., 1996). There may be differences or fluctuations in the levels of tryptophan in aphid diets; furthermore, the high level of TrpEG protein may impose an energy burden on the endosymbiont. Both of these conditions could provide short-term selective pressure for the elimination of plasmids from Buchnera (Baumann et al., 1997a). Consequently some mechanism of stabilization may be necessary for the maintenance of trpEG-containing plasmids. The gene recA has been cloned and sequenced from many bacterial taxa, and adequate oligonucleotide primers are available for its amplification by PCR (Eisen, 1995). We have made extensive unsuccessful attempts to detect this gene by PCR, suggesting that it may be absent from Buchnera or is greatly modified. Once plasmid stabilization occurs, the aphid may encounter conditions in which trpEG amplification is no longer necessary due to availability of tryptophan in the diet. If the usual mechanisms (homologous recombination?) which effect a decrease in trpEG amplification are absent, then one way of reducing the potentially wasteful synthesis of TrpEG is gene silencing.

In the past we have speculated that trpEG amplification is a property of rapidly growing aphids and that gene silencing may occur when following plasmid stabilization, the diet of aphids is nutritionally enriched (Baumann et al., 1997a; Lai et al., 1994, 1995, 1996). Recent studies have indicated that these speculations are overly simplistic. It has been suggested that the lack of trpEG amplification in Buchnera (Sc, Mr) which are in the Pemphigidae is due to the slow growth rate of these aphids compared to that of aphids within the Aphididae (Lai et al., 1995). There are, however, few studies on the growth rate of aphids, and the finding of amplification in Buchnera (Ps, Tc) which are in the family Pemphigidae and presumably also have a slow growth rate makes this explanation questionable. Dn causes major tissue histolysis of plants and it has been suggested that the presence of higher amounts of tryptophan in the diet may be the explanation for gene silencing (Lai et al., 1996). Recent studies on amino acid composition of ingested phloem sap are not strongly supportive of this hypothesis as a sole explanation for the presence of pseudogenes (Sandström and Moran, 1999; Sandström et al., 2000; Telang et al., 1999). Buchnera (Rp) and Buchnera (Sg) are similar in that both have pTrpEG consisting of four tandem repeats of a 3.6-kb unit (Fig. 5a). In spite of this similarity, Rp causes essentially no modification of the amino acid composition of plant phloem while Sg causes substantial increases (Sandström et al., 2000). Dn contains a plasmid with trpEG pseudogenes (Fig. 7a). The ingested diet of Dn has approximately doubled concentrations of tryptophan, yet the changes it causes in the amino acid composition of phloem are less than those caused by Sg. This suggests that pseudogene formation is not solely the result of increased dietary tryptophan (Sandström et al., 2000; Telang et al., 1999). However, in the case of Usn, the presence of trpEG pseudogenes is consistent with the finding of unusually high levels of essential amino acids in the phloem diet (Sandström and Moran, 2000). Thus the availability of nutrients in plant sap may be a partial explanation for trpEG amplification and pseudogene formation, but other factors must also be involved.

In this connection it should be mentioned that the past speculations attempt to correlate adaptations of Buchnera with properties of the aphid host (growth rate, modification of nutrient content of plant sap). The environment of the endosymbiont is the bacteriocyte vesicle, which harbors the endosymbiont. This environment is a reflection of the activities of the aphid host and is a function of its ability to obtain nutrients from the plant as well as its demands on the biosynthetic attributes of the endosymbiont. Therefore, host properties, such as the efficiency of nutrient uptake from the plant and their transformation and delivery, may determine the nutritional parameters within the bacteriocyte vesicles and impose the selective pressure resulting in Buchnera adaptation to the endosymbiotic association.

Some of the phenomena encountered in pTrpEG from Buchnera also have been found in other systems. Promoter inactivation by multiple sequence changes is the mechanism used for silencing the expression of the Bordetella pertussis toxin gene (Gross and Rappuoli, 1988) and the expression of the Bordetella urease gene cluster (McMillan et al., 1998). The changes resemble those observed in gene silencing of trpEG in Buchnera (Dn; Fig. 7a). Multiple copies of nearly identical chromosomal enzyme-encoding genes have been found in Thiobacillus ferroxidans (Kusano et al., 1991) and Nitrosospira sp. (Norton et al., 1996). This situation is similar to that found with the repeats of trpEG and trpEG pseudogenes and has led to the postulation of mechanisms for the preservation of sequence identity of the repeated units (Klotz and Norton, 1998).

The phylogenetic trees constructed on the basis of plasmid-associated genes are congruent with the phylogenetic trees based on Buchnera chromosomal genes (Fig. 2a–c). One exception, the basal position of Buchnera (Tc; Fig. 2c), is probably an artifact arising from the more rapid change of trpE in this lineage (van Ham et al., 1999). These results strongly suggest that there is no exchange of trpEG-containing plasmids between endosymbionts of different aphids. Plasmids of the pTrpEG type (Fig. 5a–h) could have an endogenous origin. DnaA boxes are found in other locations of the Buchnera genome (Clark et al., 1998c), and their assembly with trpEG could generate a separate replicon. In contrast to these plasmids, the repAC genes of pTrpEG-R are related to replication initiation proteins of IncA/C plasmids. Thus, this plasmid may be the result of an invasion of Buchnera by an exogenous plasmid that recombined with endosymbiont genes, resulting in their amplification. It has been established that some bacteria may persist for a long time in insects; conceivably, such organisms transferred their plasmids to Buchnera during the infection of embryos or eggs at a stage at which the endosymbionts are not sequestered within bacteriocytes. It is, however, difficult to explain the presence of repAC and its remnant in Buchnera (Rm, Rp). One possibility is that it is the result of another invasion by a similar plasmid. However, organization of ori-3.6 in Buchnera (Rm, Rp) closely resembles that of Buchnera of related aphids, suggesting a common plasmid origin for this group (Fig. 5a–c). Alternatively, van Ham et al. (1999) suggested that a plasmid of the pTrpEG-R type is the ancestor of all trpEG amplification plasmids. This hypothesis requires the subsequent occurrence of multiple losses of repAC and the interons in pTrpEG-R type plasmids and the acquisition of DnaA boxes in Buchnera of Aphididae.

pLeu Plasmids

Figure 8 is an outline of the pathway of branched-chain amino acid biosynthesis. The gene for aspartokinase (thrA) has been found in Buchnera (Sg). The genes ilvIH, ilvC, and ilvD have been found in Buchnera (Sg, Dn, Sc, Mr) and encode three enzymes which function in both the isoleucine and valine biosynthetic pathways (Clark et al., 1998c, 1999b; Thao and Baumann, 1998). The pathway of leucine biosynthesis is a branch off the valine pathway (Fig. 8). Bracho et al. (1995) found that in Buchnera (Rp), the genes for leucine biosynthesis (leuABCD) were present on a plasmid (Fig. 9a). This plasmid represents a third type (Table 7), designated pLeu, which is characterized by the presence of genes (repA1, repA2) encoding putative replication initiation proteins related to those of plasmids of the IncFII incompatibility group. Besides these genes, pLeu also contains ORF1 encoding a putative membrane-associated protein. The closely related Buchnera (Sg, Rp, Dn) all contain very similar plasmids of 7.8 to 8.0 kb in which the genes are arranged in the same order (Fig. 9a;Baumann et al., 1999a; Bracho et al., 1995). In Buchnera from the more distantly related pLeu-Ppo (Fig. 2e), there are rearrangements of the repA genes and ORF1 (Fig. 9b; Silva et al., 1998). All of these aphids are within the family Aphididae. In pLeu-Ts (pBTs1; Fig. 9c), which is from an aphid of the family Thelaxidae, there is also a rearrangement of the leu genes as well as an acquisition of ibp which encodes a heat shock protein (van Ham et al., 1997). All of these plasmids have a conserved region, downstream of repA1, which is probably an origin of replication (ori) (Baumann et al., 1999; Bracho et al., 1995; van Ham et al., 1997). Remarkably pBTc1 (Fig. 9d), a 1.7 kb plasmid from Buchnera (Tc), which is found in an aphid belonging to the family Pemphigidae, contains only ori, repA1, and ORF1 and probably constitutes a minimal replicon.

In Buchnera (Sg), there are about 24 copies of pLeu per endosymbiont genome while in Buchnera (Dn) there are only 2 copies (Thao et al., 1998). This difference in functional gene copy number parallels that observed with pTrpEG in the endosymbionts of these two aphid species. In the case of pLeu, in which only one copy of the genes is present, the reduction in amplification in Buchnera (Dn) is achieved by means of a reduction of copy number. In pTrpEG, which contains tandem repeats of the same unit, the reduction in amplification is accomplished by means of pseudogene formation (Fig. 7a).

The similarities of pLeu plasmids suggest a single origin with pBTc1 (Fig. 9d) being the ancestral state (Baumann et al., 1999). In the lineage common to the Aphididae and the Thelaxidae, there was probably a duplication of repA1 and the acquisition of leu genes. This was followed by a rearrangement of the genes and in one lineage the acquisition of ibp [for another interpretation, see van Ham et al. (1997)]. Silva et al. (1998) sequenced repA2 from Buchnera of six additional aphids. A phylogeny based on this gene (Fig. 2e) as well as a more limited analysis based on leu genes (Fig. 2d) is congruent with trees established on the basis of other Buchnera chromosomal and plasmid genes. These results indicate that the pLeu plasmids are not exchanged among endosymbionts from different aphid species and that their evolution is vertical, as is the case with the trpEG-containing plasmids.

Unanswered Questions; Other Possible Adaptations

One unanswered question is, why are only the trp and leu genes amplified? The endosymbiont produces other essential amino acids for the aphid host and their overproduction would in principle also be enhanced by plasmid amplification. It has been speculated that aphids make indole acetic acid which is involved in gall formation (Forrest, 1987). In many plant pathogens that cause gall formation, tryptophan is the precursor of indole acetic acid (Patten and Glick, 1996). Amplification of trpEG is, however, found in aphids that do not produce galls (Sg, Dn, Rp, Usn) and is absent in some that do (Sc, Mr). Consequently this does not appear to be a probable explanation for trpEG amplification. Leucine, lysine, valine, arginine and threonine are the most common amino acid in aphids (Sandström and Moran, 1999), yet only genes for leucine biosynthesis have been detected on plasmids. It is possible that, in other cases where amino acids are overproduced for the host, an increase in enzyme activity is obtained by increasing the expression of the gene(s) by promoter modification. Alternatively changes of the allosteric properties of regulated enzymes may allow retention of activity in the presence of endproducts.

A possible example of the latter is cysE of Buchnera (Sg; Lai and Baumann, 1992b). This gene encodes an enzyme of the biosynthetic pathway of cysteine, and its activity is regulated by cysteine feedback inhibition (Kreditch, 1996). It has been established that the amino acids at the C-terminus of the E. coli enzyme are involved in cysteine feedback inhibition (Denk and Böck, 1987). The Buchnera (Sg) enzyme lacks these C-terminal amino acids and consequently is probably not subject to feedback inhibition by the endproduct; this change would result in cysteine overproduction (Lai et al., 1992b).

Secondary Endosymbionts of Aphids

Besides Buchnera, many aphids have additional endosymbionts usually called secondary (S-) endosymbionts (Buchner, 1965; Houk and Griffiths, 1980; Moran and Baumann, 1994). In many cases these endosymbionts are spheres or rod-shaped with different width and length; they were initially recognized by differences in size and shape from the round or oval Buchnera. The S-endosymbionts are also maternally inherited. They have not been extensively studied, and most of the available information is for the S-endosymbionts of the aphid Ap. Electron microcopic studies have shown that the rod-shaped S-endosymbionts are located within vesicles found in the flattened, syncytial, sheath cells which surround the bacteriome (Griffiths and Beck, 1973). Using a probe derived from E. coli 16S rDNA and restriction enzyme and Southern blot analysis of total aphid DNA, it was found that the S-endosymbiont from Ap contained a single copy of the 16S rDNA-gene (Unterman et al., 1989). DNA obtained from dissected bacteriocytes gave the same restriction pattern, indicating that the S-endosymbiont was located in the bacteriome. Two DNA fragments of 2.3 kb each were cloned and the 16S rDNA sequence determined. A phylogenetic analysis indicated that the S-endosymbiont was a member of the Enterobacteriaceae (Fig. 1). As in other members of this family, the 16S-rDNA gene of the S-endosymbiont was directly upstream of 23S rDNA (Unterman and Baumann, 1990).

Chen and Purcell (1997) found that 88% of the strains of Ap had the S-endosymbiont. In addition, the S-endosymbionts of Ap and Macrosiphum rosae were identical, suggesting recent infection or horizontal transmission. Interestingly, it was also found that the hemolymph of 48% of Ap strains contained a rod-shaped organism which had a 16S rDNA sequence nearly identical to that of Rickettsia bellii, an organism found in ticks (Chen et al., 1996). As a result of these studies strains of Ap became available which had 1) only the S-endosymbiont, 2) only the Rickettsia sp., as well as 3) neither of these two organisms. Chen (1996) injected one or both of these organisms into Ap, which originally lacked both, and observed their effect on fecundity, longevity and the length of the reproductive period. The results were complex in that they were affected by the plant on which the aphids grew and by the temperature of growth. At 20°C both the S-endosymbiont and the Rickettsia sp. reduced the fecundity, longevity and reproductive period of Ap on clover but had no significant effects on Ap grown on alfalfa or sweet pea (Chen, 1996). In some cases, both of these organisms appeared to cause an increase in the fitness of Ap when grown at 25°C. These results suggest that both the S-endosymbiont and the Rickettsia sp. can have either a deleterious or a beneficial effect on the host, depending on the environmental conditions (Chen, 1996). The S-endosymbiont, upon injection into the closely related species, Acyrthosiphon kondoi, was found to be pathogenic. Although the rate of maternal transmission of both the S-endosymbiont and the Rickettsia sp. was high, one instance of S-endosymbiont loss was observed (Chen and Purcell, 1997).

Studies based on light and electron microscopy have suggested that some S-endosymbionts in some aphid species may inhabit syncytial cells or, possibly, bacteriocytes that appear similar to those containing Buchnera (Buchner, 1965; Hinde, 1971b; Iaccarino and Tremblay, 1973). Fukatsu and Ishikawa (1993) surveyed 61 aphids for the presence of S-endosymbionts. Previously it was found that Buchnera (Ap) overproduced GroEL (Kakeda and Ishikawa, 1991). Using anti-E. coli-GroES for immunoprecipitation of Buchnera (Ap) GroES, it was concluded that Buchnera produced low levels of this protein (Kakeda and Ishikawa, 1991). This antiserum as well as anti-E. coli-GroEL was used to detect GroEL and GroES in immunoblots of whole aphid extracts as well as for histochemical detection in thin sections of aphids. From these experiments it was concluded that the synthesis of substantial amounts of GroES by S-endosymbionts distinguishes them from Buchnera, and that this property can be used for the identification of S-endosymbionts (Fukatsu and Ishikawa, 1993). These studies have a number of problems that limit their general applicability to the survey of S-endosymbionts. The principal one is the use of antisera against E. coli GroES for the detection of cross reactivity of GroES from organisms that have an unknown relationship to E. coli. Since it is probable that many of the S-endosymbionts are members of the Enterobacteriaceae, a stronger cross-reaction would be expected with their proteins than with the proteins from Buchnera; consequently, an increased reactivity need not indicate a major difference in the amount of the protein. Conversely, in those cases where the S-endosymbiont is not a member of the Enterobacteriaceae, the distant relationship may preclude a strong cross-reaction (Eremeeva et al., 1998). The cross-reactivity of the Buchnera and S-endosymbiont proteins with the anti-E. coli protein antisera has not been compared. Finally, the relative production of GroES may vary among S-endosymbionts of different types.

The studies of Fukatsu and Ishikawa (1993, 1998), in which thin sections of aphids were stained by immunohistochemistry and examined by light microscopy, do suggest that in many aphids the S-endosymbionts occupy bacteriocytes distinct from those containing Buchnera. The authors also state that the S-endosymbionts have a variety of different shapes. The methods used and the photographs presented do not, however, allow adequate visualization of cell shape and the resolution of bacteriocyte structure. In addition the designation of some of the endosymbionts as Buchnera or S-endosymbionts appears to be arbitrary.

Fukatsu et al. (1998) used group-specific oligonucleotide probes for in situ detection of aphid P- and S-endosymbionts. A universal eubacterial 16S rRNA probe was used as well as probes specific for the 23S rRNA γ- and β-subdivision Proteobacteria. The sequences of the latter two probes differ by only one nucleotide. Buchnera and S-endosymbionts all hybridized with the 16S eubacterial probe, although the intensities of the signal differed considerably. Curiously the putative P-endosymbionts of two out of seven aphids did not hybridize with the 23S γ-subdivision probe. Using the total DNA preparation from these aphids, the 16S rDNA was amplified, cloned, and sequenced. Two sequences were detected in each aphid DNA preparation, and one of these was related to Buchnera 16S rDNA. Based on this result, it was concluded that these aphids contained Buchnera but that their 23S-rDNA gene was changed to such an extent that hybridization with the γ-subdivision probe no longer occurred. This conclusion is questionable since the 23S rDNA of Buchnera (Sg, Dn, Sc, Mr), which span the diversity of aphid hosts (Fig. 2), contains the exact sequence complementary to the probe used (Clark et al., 1999b; Rouhbakhsh and Baumann, 1995). The S-endosymbiont of Tetraneura radicicola hybridized with the probe to the β-subdivision but not to the γ-subdivision, suggesting that this endosymbiont is a member of the former group. Since there is only a single nucleotide difference between these two probes, a confirmation of this conclusion by sequencing the rDNA from this organism seems desirable. The S-endosymbiont from two other aphid species did not hybridize with either the γ- or β-subdivision probe. In view of the technical difficulties encountered with some of the specimens, these conclusions cannot be interpreted as indicating that the S-endosymbionts of these aphids belong to different bacterial groups.

Absence of a Stable Bacterial Flora in Aphid Guts

Aphids maintained under clean conditions do not appear to have a bacterial gut flora (Douglas, 1990; Grenier et al., 1994; Harada and Ishikawa, 1993). Older aphids as well as aphids reared under crowded conditions may acquire a gut flora which appears to consist of members of the Enterobacteriaceae (Serratia, Erwinia), Pseudomonas, Staphylococcus, and Bacillus (Grenier et al., 1994; Harada and Ishikawa, 1993). These organisms are frequently associated with plant surfaces. None of these organisms has a close relationship to Buchnera, precluding a recent common ancestor from which they and Buchnera are descended. The presence of this bacterial flora has a deleterious effect on aphid performance (Grenier et al., 1994). The actual bacterial numbers have not been established. Harada et al. (1996) have isolated 38 bacterial strains from the guts of 20 aphids. This hardly suggests the presence of an indigenous bacterial flora and the numbers are insignificant compared to the numbers of Buchnera or the S-endosymbionts. These studies indicate that the guts of aphids are generally sterile but under certain conditions a transient bacterial gut flora may be present.

Endosymbionts of Other Plant Sap-Utilizing Insects

Psyllids, whiteflies and mealybugs are three separate lineages of the suborder Sternorrhyncha and contain P-endosymbionts from three distinct bacterial groups (Table 1, Fig. 1). These insects have obligatory sexual reproduction with the young hatching from eggs (Borror et al., 1989). The endosymbionts are housed within bacteriocytes, and at least during some stage of the insect’s life cycle, the bacteriocytes are associated with the ovarioles resulting in the transmission of endosymbionts to the eggs (Buchner, 1965).

Psyllid Endosymbionts

Fukatsu and Nikoh (1998) sequenced the 16S rDNA of endosymbionts from Anomoneura mori, while Spaulding and von Dohlen (1998) performed a similar study of the endosymbionts of Blastopsylla occidentalis, Pachypsylla venusta, and Trioza magnoliae. These authors came to the same conclusion, namely that the P-endosymbionts of psyllids constitute a distinct lineage within the γ-subdivision of the Proteobacteria (Fig. 1). These endosymbionts have an unusual property, namely, the lowest known G+C content of any 16S rDNA (36.4 mol%). In addition, this lineage appeared to have a substantial acceleration of the rate of evolutionary change within the 16S sequence. A. mori, B. occidentalis, and T. magnoliae all had different S-endosymbionts, which belong in the γ-subdivision. Only the P-endosymbiont was found in P. venusta (Spaulding and von Dohlen, 1998). The studies were limited to few taxa but were consistent with cospeciation of the P-endosymbiont and the psyllid host and multiple acquisitions of S-endosymbionts (Fukatsu and Nikoh, 1998; Spaulding and von Dohlen, 1998).

Psyllids contain bilobed bacteriomes made up of round uninucleate bacteriocytes and a multinucleate syncytial region (Buchner, 1965). Many psyllids have endosymbionts in both the bacteriocytes and the syncytium; some have endosymbionts only within the bacteriocytes (Buchner, 1965). Using electron microscopy, Chang and Musgrave (1969) and Waku and Endo (1987) found that two psyllid species have endosymbionts in the bacteriocytes which are distinguishable from those found in the syncytium and that both endosymbiont types have a Gram-negative cell wall. In general, the bacteriocyte-associated endosymbionts are more numerous than the syncytium-associated endosymbionts (Buchner, 1965). Fukatsu and Nikoh (1998), using an oligonucleotide probe specific for the P- or S-endosymbiont, showed by means of in situ hybridization that the former was localized in the bacteriocytes while the latter was in the syncytium (Fig. 10). There is currently no information on the requirement of the endosymbiont(s) by the psyllid host or on their function. The similarity in diet between psyllids and aphids raises the possibility that psyllid endosymbionts may provide nutrients as do Buchnera.

Whitefly Endosymbionts

16S rDNA sequences have been obtained for endosymbionts of Bemisia tabaci, B. argentifolii (previously B. tabaci B biotype), Siphonius phillyreae, and Trialeurodes vaporariorum (Clark et al., 1992). The P-endosymbionts of these insects are a lineage within the γ-subdivision of the Proteobacteria (Fig. 1). B. tabaci and B. argentifolii have an S-endosymbiont which is a member of the Enterobacteriaceae. The P-endosymbionts and the S-endosymbionts from these two species have identical 16S rDNA sequences, consistent with their close relationship (Brown et al., 1995; Clark et al., 1992).

The ultrastructure of the endosymbionts of B. tabaci, B. argentifolii, and T. vaporariorum has been studied by electron microcopy (Costa et al., 1993b, 1995). There is evidence for at least two morphological types. Whiteflies are unusual in that they transmit an entire bacteriocyte containing endosymbionts to the egg (Buchner, 1965; Costa et al., 1996). Some prokaryote-specific antibiotics affect the growth and development of whiteflies, indicating a requirement for the endosymbiont(s) by the host (Costa et al., 1993a, 1997).

Mealybug Endosymbionts

16S rDNA sequences have been obtained for endosymbionts of Pseudococcus longispinus, P. maritimus, and Dysmicoccus neobrevipes (Munson et al., 1992), and these organisms were found to be a distinct lineage within the β-subdivision of the Proteobacteria (Fig. 1). The morphology of endosymbionts from several mealybug species has been studied by means of electron microscopy (Tremblay, 1989). Within the bacteriocytes the endosymbionts appear to be embedded in mucous spherules of unknown composition. There is no information on the function or the requirement for the endosymbionts by the host.

Tsetse Fly Endosymbionts

Tsetse flies (genus Glossina) are important vectors of trypanosomes, which are causative agents of African sleeping sickness and various diseases of animals (Harwood and James, 1979). They have a somewhat unusual reproductive cycle in that the female gives birth to fully grown larvae. Only one larva is carried at a time within the uterus. During this stage the larva is fed nutritive fluids from special glands, commonly known as “milk glands.” The female requires several blood meals to complete the development period of each larva, and it is these blood meals which result in the transmission of trypanosomes (Harwood and James, 1979).

Tsetse may be associated with three prokaryotes: (1) Wigglesworthia (P-endosymbionts), 2) Sodalis (S-endosymbionts), and 3) Wolbachia. The last are parasites found in reproductive tissue and causing reproductive disorders (O’Neill et al., 1997) and will not be considered here.