Genomic insight into the amino acid relations of the pea aphid, Acyrthosiphon pisum, with its symbiotic bacterium Buchnera aphidicola

Authors


Alex Wilson, Department of Biology, University of Miami, Coral Gables, FL 33146, USA. Tel.: +1305 2842003; fax: +1305 2843039; e-mail: acwilson@bio.miami.edu

Abstract

The pea aphid genome includes 66 genes contributing to amino acid biosynthesis and 93 genes to amino acid degradation. In several respects, the pea aphid gene inventory complements that of its symbiotic bacterium, Buchnera aphidicola (Buchnera APS). Unlike other insects with completely sequenced genomes, the pea aphid lacks the capacity to synthesize arginine, which is produced by Buchnera APS. However, consistent with other insects, it has genes coding for individual reactions in essential amino acid biosynthesis, including threonine dehydratase and branched-chain amino acid aminotransferase, which are not coded in the Buchnera APS genome. Overall the genome data suggest that the biosynthesis of certain essential amino acids is shared between the pea aphid and Buchnera APS, providing the opportunity for precise aphid control over Buchnera metabolism.

Introduction

Many insects possess nutritionally-important symbiotic microorganisms, some of which contribute to the digestion of complex polysaccharides in wood or soil (e.g. in many termites), while others provide nutrients to insects feeding on nutritionally unbalanced diets, e.g. B vitamins to blood-feeding cimicid bugs, tsetse fly and anopluran lice, and essential amino acids to aphids, whiteflies and other plant sap feeding insects (Buchner, 1965; Douglas, 2009). Many of these symbiotic associations are difficult to study experimentally, but genomic data can contribute to defining the likely interactions (e.g. McCutcheon & Moran, 2007; Warnecke et al., 2007).

The availability of the complete genome sequence of the pea aphid, Acyrthosiphon pisum (International Aphid Genomics Consortium, 2010) provides a particularly valuable opportunity to address symbiotic interactions, for two reasons. First, the genome of its bacterial symbiont Buchnera aphidicola (Buchnera APS) is already available (Shigenobu et al., 2000), making this the first insect-microbial symbiosis for which the complete genomes of both partners have been sequenced; and second, there is already a considerable body of experimental data on this symbiosis, providing a route to link genomic and functional data.

Buchnera cells are intracellular, located in specialized aphid cells, known as bacteriocytes, in the haemocoel and they are obligately vertically transmitted to eggs or early embryos in the female reproductive tract (Buchner, 1965; Braendle et al., 2003). There is persuasive experimental evidence that these bacteria provide the insect with essential amino acids, supplementing the limited supply of these nutrients in the aphid diet of plant phloem sap (Douglas & Prosser, 1992; Liadouze et al., 1996; Gündüz & Douglas, 2009). [Essential amino acids are those amino acids that contribute to protein and have a carbon skeleton that cannot be synthesized de novo by the insect (Dadd, 1985).] Consistent with these experimental data, the genome of Buchnera APS includes most genes required for essential amino acid biosynthesis, but lacks the genes for amino acid degradation and the synthesis of most nonessential amino acids (Shigenobu et al., 2000, Prickett et al., 2006). These nutritional, metabolic and genomic observations suggest that aphid metabolism supports a substantial flux of nonessential amino acids from insect to Buchnera and of essential amino acids in the reverse direction. It has been assumed widely that the amino acid biosynthetic reactions encoded in the aphid genome but not in the extant Buchnera genome were lost from the ancestral Buchnera symbiont through genetic drift and relaxed selection, with the prediction that the two partners are functionally complementary (Moran, 1996; Shigenobu et al., 2000; Moran et al., 2008). The availability of the pea aphid genome provides a unique opportunity to test this hypothesis of genomic complementarity.

Here, we describe the pea aphid genes that contribute to the biosynthesis and degradation of the 20 amino acids contributing to proteins. We interpret our results in the context of the amino acid relations of this insect with its symbiotic bacteria, Buchnera APS, enabling us to generate specific hypotheses for future investigations by genetic and biochemical approaches.

Results

Overview of the genetic capacity of the pea aphid for amino acid biosynthesis and degradation

The genes in amino acid biosynthesis and degradation identified from the genome sequence of the pea aphid, Acyrthosiphon pisum, are summarized in Table 1 (for the full list, see Supporting Information Table S1). Overall, the pea aphid genome has 66 genes contributing to amino acid biosynthesis and 93 genes to amino acid degradation. As predicted (see Introduction), the capabilities of the aphid broadly complement those of its symbiotic bacterium Buchnera APS and sequence comparisons show that no amino acid biosynthetic genes encoded in the pea aphid genome are the result of horizontal gene transfer from Buchnera APS or other microbes (International Aphid Genomics Consortium, 2010).

Table 1.  Number of genes in the pea aphid and its symbiotic bacterium Buchnera encoding for enzymes in the synthesis and degradation of protein amino acids
 Number of Genes
Pea aphidBuchnera
  • *The carbon skeleton of isoleucine is derived principally from aspartate with a minor contribution from pyruvate. The genes contributing to isoleucine synthesis are allocated to the aspartate biosynthetic family.

  • Genes mediating sulphate reduction are shown in parentheses.

Biosynthetic families  
 Aspartate*2216
 Phosphoenolpyruvate612
 Glutamate278
 3-Phosphoglycerate243(9)
 Ribose-5-phosphate08
 Pyruvate*111
Degradative families  
 α-ketoglutarate270
 Succinyl Co-A230
 Oxaloacetate270
 Pyruvate90
 Acetyl-CoA and acetoacetyl Co-A390

Manual annotation and analysis of the pea aphid genome revealed five instances involving amino acid biosynthesis where: (1) the genetic capacity of the pea aphid differs from that of other insects with completely sequenced genomes or (2) pea aphid genes contribute to pathways that are also represented by Buchnera APS genes, leading to the potential for shared metabolic pathways between the two partners. These five instances relate to pea aphid aminotransferases, and biosynthesis of arginine, the branched-chain amino acids, the aromatic amino acids phenylalanine and tyrosine, and the sulphur-containing amino acids. They are considered in turn.

Transamination

Transaminases, also known as aminotransferases, play an important role in amino acid metabolism; by transferring the amino group from an amino acid to an alpha-keto acid, they generate new amino acids from existing ones. Sixteen genes encoding 10 different predicted transaminases were identified in the pea aphid (Table 2 and Supporting Information Table S2). Genes encoding nine of these 10 activities are also predicted to occur in Drosophila melanogaster and in most other insects with sequenced genomes. The additional enzyme present in the pea aphid is of undefined specificity (EC 2.6.1.-) and is also present in Tribolium castaneum and as two copies in Apis mellifera but in no other insects with sequenced and annotated genomes (Table 2). Most insect genomes examined include two genes encoding each of aspartate transaminase (EC 2.6.1.1) and alanine-glyoxylate transaminase (EC 2.6.1.44), with the ancestral gene for each reaction duplicated prior to the common ancestor of the insects. Uniquely, the pea aphid has three copies of aspartate transaminase 1 (GOT1) and two copies of one homolog of alanine-glyoxylate transaminase (Table 2 and Supporting Information Table S2). Our survey of transaminase genes among insects with completely sequenced genomes and examination of the phylogenetic relationships of the pea aphid transaminase genes using Phylome DB (http://phylomedb.org), suggests that duplication of transaminase genes is not unusual, but that only the pea aphid possesses more than two copies of any transaminase (Table 2).

Table 2.  The complement of transaminase genes in Acyrthosiphon pisum, other insects with sequenced genomes and the bacterial symbiont Buchnera aphidicola APS Thumbnail image of

Two transaminases are of particular note. The first is tyrosine transaminase (EC 2.6.1.5), which mediates key reactions in the metabolism of the aromatic amino acids, phenylanaline and tyrosine (specifically, intercoverting these amino acids with phenylpyruvate and 4-hydroxy-phenylpyruvate, respectively). The gene for tyrosine transaminase is apparently absent from the pea aphid, but has been identified in all other sequenced insect genomes (Table 2). Aspartate aminotransferase 2 (GOT2; EC 2.6.1.1) has been documented to catalyse the two tyrosine transaminase reactions in other eukaryotes (e.g. Lain-Guelbenzu et al., 1990; Yagi et al., 1990, and BRENDA database, see (Chang et al. 2009)) suggesting that these reactions are likely to be mediated by the pea aphid.

The second transaminase, one with an apparently unusual distribution, is the unidentified transaminase EC 2.6.1.-, present in one copy in the pea aphid and T. castaneum and two copies in A. mellifera (Table 2). Absence of this transaminase from all dipteran genomes sequenced to date suggests a possible Diptera-specific loss. Extensive manual annotation of this pea aphid gene highlighted uncertainties in the annotation of the A. mellifera and T. castaneum genes. Consequently, we take a conservative approach assigning EC 2.6.1.- function and leaving the transaminase unnamed.

There is some functional overlap between the transaminase gene complement of the pea aphid and that of Buchnera APS. Buchnera APS, with five genes encoding likely transaminases, has less than one third of the 16 transaminase genes in the related free-living bacterium Escherichia coli. Two of the five Buchnera APS transaminase genes have functional equivalents in the pea aphid genome: glutamine-fructose-6-phosphate transaminase (EC 2.6.1.16) coded by the Buchnera APS gene glmS and pea aphid gene ACYPI003246, and phosphoserine transaminase (EC 2.6.1.52) coded by the Buchnera APS gene serC and pea aphid gene ACYPI004666.

Synthesis and degradation of branched-chain amino acids

As with other insects, the pea aphid has the genetic capacity to degrade the three branched-chain amino acids to acetyl-coenzyme A (isoleucine and leucine) and succinyl-coenzyme A (valine) and also has two genes in the biosynthetic pathways. These two genes code threonine dehydratase (also known as threonine ammonia lyase, EC 4.3.1.19 ACYPI006784), which converts threonine into 2-oxobutanoate in the isoleucine biosynthetic pathway (Fig. 1), and the branched-chain-amino-acid transaminase (EC 2.6.1.42 ACYPI008372, Table 2), which catalyses the final reaction in the production of each of the three branched-chain amino acids. Buchnera APS lacks the genes that encode the enzymes mediating these reactions in related species, including E. coli (Shigenobu et al., 2000). This metabolic complementarity between the aphid and its symbiont raises the possibility that the biosynthesis of these amino acids is shared between the Buchnera APS and the aphid. First, for isoleucine synthesis, this would involve the transfer of the substrate and product of the threonine dehydratase reaction (threonine and 2-oxobutanoate, respectively) between the symbiotic partners. Second, the terminal reactions in the biosynthetic pathways of each of the branched-chain amino acids would require transport of their substrates from Buchnera to aphid (shown for isoleucine in Fig. 1). One important implication is that Buchnera APS would be dependent on the aphid for its supply of the three essential amino acids, isoleucine, leucine and valine, even though the Buchnera cells mediate most of the reactions in the pathway. In this way, the proposed enzymatic activity and associated metabolite exchange between aphid and Buchnera APS can account for absence of the genes for threonine dehydratase (ilvA) and branched-chain amino acid transaminase (ilvE) from Buchnera APS. We consider this interpretation and alternative possible routes by which the enzymatic functions encoded by the missing Buchnera APS genes might be maintained in the discussion section.

Figure 1.

Proposed pathway for synthesis of the branched-chain amino acid isoleucine in the pea aphid-Buchnera symbiosis. In this and subsequent figures, black arrows indicate reactions supported by genome annotation data and grey arrows represent transport of metabolites between symbiotic partners. The larger white box represents the pea aphid with Buchnera as the inner grey box. Threonine ammonia lyase is also known as threonine dehydratase (see text).

Phenylalanine and tyrosine metabolism

Insects generally have the capacity to mediate the terminal reaction in the synthesis of phenylalanine by transaminases (tyrosine transaminase and aspartate transaminase), and the subsequent transformation of phenylalanine to tyrosine via phenylalanine hydroxylase, also known as phenylalanine 4-monooxygenase. The pea aphid genome has the capacity for these reactions, possessing the gene for aspartate transaminase (Table 2) and phenylalanine 4-monooxygenase (EC 1.14.16.1 ACYPI007803), although (as discussed above) it is the sole insect with a completely sequenced genome that apparently lacks a gene for tyrosine transaminase (Table 2).

These enzymatic capabilities of the pea aphid complement the genetic capacity of Buchnera APS, which has the genes for all reactions in phenylalanine and tyrosine synthesis except tyrA and tyrB (Fig. 2). Buchnera APS might, consequently, be dependent on the aphid for its supply of phenylalanine and tyrosine.

Figure 2.

Proposed pathway for phenylalanine and tyrosine biosynthesis in the pea aphid-Buchnera symbiosis. The genes coding the reactions in tyrosine synthesis that are missing from Buchnera are tyrA and tyrB; tyrB also encodes the final ‘missing’ reaction in phenylalanine synthesis.

The pathway for the degradation of phenylalanine and tyrosine in animals involves the elimination of the amino group (transamination) from tyrosine, followed by degradation of the carbon core by oxygenases to form acetoacetate and fumarate, which enter the Krebs cycle. Unusually for an animal, the pea aphid apparently lacks the genes coding the reactions in this pathway (EC 1.13.11.27, EC 1.13.11.5 and EC 5.2.1.2). The absence of this capability is not complemented by Buchnera APS which also lacks these enzymatic activities.

Arginine synthesis

Unusual among sequenced insects, the pea aphid lacks all the genes encoding enzymes of the urea cycle (EC 2.1.3.3, EC 6.3.4.5, EC 4.3.2.1 and EC 3.5.3.1) and is, consequently, unable to synthesize arginine. In other words, arginine is an essential amino acid for the pea aphid. Buchnera APS retains a complete pathway for arginine synthesis (Shigenobu et al., 2000, Prickett et al., 2006), and is therefore likely to provide this nutrient to the pea aphid. Provision of arginine by Buchnera to the pea aphid is a topic considered further by Ramsey et al. (2010) in relation to other consequences of the lack of a functional urea cycle in the aphid.

Sulphur amino acid metabolism

Inorganic sulphate is transported in phloem sap, and the green peach aphid, Myzus persicae, containing Buchnera has been shown to incorporate sulphur from this source into both cysteine and methionine (Douglas, 1988). Although the Buchnera APS genome encodes enzymes for the biosynthesis of cysteine from serine and sulphate, the methionine biosynthesis pathway is incomplete, preventing the incorporation of sulphur from cysteine to make cystathionine, homocysteine, and methionine (Fig. 3). Further insight into cysteine and methionine biosyntheses comes from analysis of the pea aphid genome, which has orthologues of all D. melanogaster genes that contribute to sulphur amino acid metabolism. These include the genes annotated as encoding homocysteine S-methyltransferase (EC 2.1.1.10), cystathionine β-synthase (EC 4.2.1.22) and cystathionine γ-lyase (EC 4.4.1.1), which mediate the three sequential reactions in the synthesis of cysteine from methionine in animals. Therefore, based on the predicted enzyme activities, it seems impossible for the pea aphid-Buchnera APS symbiosis, like that of M. persicae, to synthesize methionine from sulphate.

Figure 3.

Proposed pathways for sulphur amino acid metabolism in the pea aphid-Buchnera symbiosis. (1) Homocysteine methyltransferases (e.g. EC2.1.1.10; ACYPI009247, ACYPI001309) catalyse the formation of methionine from homocysteine. (2) S-adenosylmethionine synthase (EC2.5.1.6; ACYPI005100, ACYPI004816) produces S-adenosylmethionine from methionine. (3) S-adenosylmethionine is used by DNA methyltransferase and other methylation reactions (e.g. EC2.1.1.37; ACYPI007944), resulting in the production of S-adenosylhomocysteine. (4) S-Adenosylhomocysteine is converted to homocysteine by S-adenosylhomocysteine hydrolase (EC3.3.1.1; ACYPI003908, ACYPI004267).

One resolution to the paradox of methionine biosynthesis could be that aphid proteins currently annotated as cystathionine β-synthase and cystathionine γ-lyase actually catalyse the reverse reactions. Specifically, substrate concentrations in the bacteriocytes might allow these enzymes to function in the opposite direction and, together with the enzymes of the S-adenosylmethionine cycle, allow synthesis of homocysteine and methionine from cysteine (Fig. 3). Alternatively, these pea aphid enzymes might catalyse similar cystathionine β-lyase (EC 4.4.1.8) and cystathionine γ-synthase (EC 2.5.1.48) reactions, respectively. In either case, the transfer of metabolic intermediates between the two partners would be as illustrated in Fig. 3. These hypotheses can be tested by enzyme assays and analysis of metabolic flux in the pea aphid-Buchnera APS system.

S-methylmethionine and glutathione (γ-l-glutamyl-l-cysteinyl-glycine) are key sulphur transport molecules in the phloem of many, perhaps most, plant species (Bourgis et al., 1999). This raises the possibility that aphids have efficient mechanisms for producing methionine and cysteine, respectively, from these phloem metabolites. One or more of the predicted homocysteine S-methyltransferases encoded in the pea aphid genome might use S-methylmethionine as a methyl group donor to make methionine, a hypothesis that is supported by the observation that labelled S-methylmethionine fed to M. persicae is incorporated into methionine in a homocysteine-dependent manner (M. Lee and G. Jander, unpublished observations). The pea aphid genome also encodes predicted enzymes for the degradation of free or conjugated glutathione. The sequential action of γ-glutamyltranspeptidase (EC 2.3.2.2) and aminopeptidase (EC 3.4.11.2), both of which are represented in the pea aphid genome (Ramsey et al. 2010) would catalyse the release of cysteine from this abundant phloem metabolite.

Discussion

Annotation of pea aphid genes involved in biosynthesis and degradation of the 20 protein amino acids offers biological insight into interactions between the aphid and its symbiotic bacterium Buchnera APS in two distinct ways. First, genome annotation highlights the likely contribution of both Buchnera and aphid enzymes to the synthesis of several amino acids (including: methionine; the branched-chain amino acids, valine, isoleucine and leucine; and the aromatic amino acids, phenylalanine and tyrosine). Second, we find that genome evolution in the pea aphid, like that of Buchnera APS, has been subject to gene loss. In particular, the pea aphid has apparently lost the genes responsible for tyrosine degradation and arginine biosynthesis. These two aspects are considered in turn.

It has been known for several years that Buchnera APS lacks some genes encoding essential amino acid biosynthesis enzymes (Shigenobu et al., 2000; Zientz et al., 2004) and yet paradoxically, there is strong experimental evidence that all essential amino acids are produced in the pea aphid (Douglas, 1988; Febvay et al., 1999). The initially favoured explanation for this paradox was that other Buchnera APS enzymes are promiscuous, mediating the ‘missing’ reactions. For example with regard to biosynthesis of the branched-chain amino acids isoleucine, leucine and valine, Shigenobu et al. (2000) proposed that the Buchnera APS protein HisC could functionally replace the missing branched-chain amino acid transaminase, catalysing the terminal reaction in these pathways. Our annotation of the pea aphid genome raises an alternative explanation. We found that the pea aphid specifically possesses the genetic capacity to mediate the ‘missing’Buchnera APS branched-chain amino acid biosynthesis reactions; threonine ammonia lyase (threonine dehydratase) and the branched-chain amino acid transaminase compensate for the reactions of the missing Buchnera APS ilvA and ilvE genes (Fig. 1). Similarly with regard to biosynthesis of the aromatic amino acids, phenylalanine and tyrosine, we hypothesize that aphid phenylalanine 4-monooxygenase and aspartate transaminase compensate for the missing Buchnera APS TyrA and TyrB enzymes (Fig. 2). It is important to note here two points. First, in the biosynthesis of each of these five amino acids, isoleucine, leucine, valine, phenylalanine and tyrosine, the aphid gene products mediate one or two reactions in otherwise complete metabolic pathways of Buchnera APS. Second, each of these aphid genes have orthologues in other insects with completely sequenced genomes, and thus the presence of these genes should not be interpreted as aphid adaptations to the symbiosis, but rather that the pre-existing capability of the aphid to compensate for genes lost from the Buchnera APS genome has facilitated genomic degradation in Buchnera APS. In contrast, the whole methionine biosynthesis pathway, with the exception of the final reaction mediated by metE, is missing from the Buchnera APS genome (Shigenobu et al., 2000). In this case, we hypothesize that the missing Buchnera APS metB and metC reactions may be mediated by adaptations of the pea aphid, specifically reversal of reactions catalysed by cystathionine β-synthase and cystathionine γ-lyase (Fig. 3).

The fact that Buchnera APS has lost genes with functions that are matched by genes already present in the pea aphid genome leads to the interpretation, developed in this paper, that Buchnera has lost genes through redundancy and that metabolic intermediates are transported between the bacterial cells and the aphid. Consistent with this hypothesis, the aphid bacteriocytes are known, from EST profiling, to express several of the candidate enzymes (Supporting Information Table S3; Nakabachi et al., 2005); and these EST data are supported and extended by on-going deep sequencing of bacteriocyte cDNA using next-generation sequencing approaches (Ashton et al., unpublished). A further important issue is that the hypothesized transfer of metabolites between the partners involves transit across three membranes, the inner and outer Buchnera membranes and an outermost membrane, known as the symbiosomal membrane that is believed to be of aphid origin (McLean & Houk, 1973). Very little is known about the transport processes across these membranes. Although a range of metabolites are transported, as indicated by experimental analysis (Whitehead & Douglas, 1993) and reconstruction of the Buchnera metabolic network (Thomas et al., 2009), it is not understood how the small complement of transporters coded by the Buchnera APS genome meet these known transport capabilities (Shigenobu et al., 2000). An alternative interpretation is that the aphid enzymes could be transferred to the Buchnera cells where they contribute to metabolism of the bacterial cells. On the available evidence, this latter explanation appears unlikely because the aphid genes for the enzymes of interest lack any recognizable signal sequence, indicating that the cognate proteins are trafficked to an intracellular compartment. Finally, the original hypothesis of Shigenobu et al. (2000), that there are promiscuous transaminases in the Buchnera APS, is also still possible and could be tested experimentally.

The proposed contribution of the aphid to the synthesis of several amino acids can potentially enable the aphid to control amino acid supply to the Buchnera cells and thereby, rates of Buchnera protein synthesis and growth. This can be illustrated by synthesis of the aromatic amino acids phenylalanine and tyrosine. Because Buchnera APS lacks TyrA and TyrB, it appears to be dependent on the aphid for its supply of phenylalanine and tyrosine (Fig. 2). Yet, the principal source of aphid-derived phenylalanine and tyrosine is interpreted as Buchnera-derived phenylpyruvate (the immediate precursor of phenylalanine), meaning that Buchnera APS access to these amino acids is dependent on its sustained supply of phenylpyruvate to the aphid. Because the aphid (and not Buchnera) controls the allocation of phenylalanine to different fates (protein synthesis, metabolism to tyrosine, distribution of phenylalanine and tyrosine between aphid and Buchnera), Buchnera must supply sufficient phenylpyruvate to meet the total aphid demand in order to receive its requirements for phenylalanine and tyrosine. Likewise, our hypothesized localization of the terminal branched-chain amino acid biosynthesis reaction to the aphid (Fig. 1) has the consequence that Buchnera APS can obtain these amino acids only by committing sufficient flux through this pathway to meet the requirements of the aphid and itself. To date, the mechanisms by which aphids might allocate amino acids have not been investigated. In principle, aphids could regulate supply to the total Buchnera complement of a bacteriocyte by controlling the cytoplasmic availability of metabolites in a bacteriocyte. Alternatively, it could regulate supply of amino acids to an individual Buchnera cell by controlling the number and activity of transporters on the aphid-derived symbiosomal membrane.

While we argue above that genetic compensation by the pea aphid for reactions lost from the Buchnera APS genome cannot necessarily be interpreted as adaptations to the symbiosis (i.e. the traits had evolved prior to the evolutionary origin of the symbiosis), the gene complement of the pea aphid does include some features that may be adaptive in the context of the symbiosis. For example, despite the arginine content of plant phloem sap frequently being insufficient to meet aphid requirements (Gündüz & Douglas, 2009), the pea aphid lacks the urea cycle, including all arginine biosynthesis genes. Arginine is thus, an essential amino acid for the pea aphid, with the aphid's arginine requirements met by Buchnera APS, which retains all the genes for de novo arginine synthesis (International Aphid Genomics Consortium, 2010). In this respect, arginine is exceptional because it is the sole amino acid that was probably synthesized by both the insect and bacterial ancestor of the aphid-Buchnera symbiosis and subsequently lost by the aphid. In contrast, the capacity to synthesize eight other amino acids has been lost from Buchnera APS (Shigenobu et al., 2000). To date, loss of tyrosine degradation is unique to the pea aphid amongst the insects with completely sequenced genomes. Tyrosine degradation may be redundant in the pea aphid because of high Buchnera demand for this amino acid that the bacterium can neither synthesize nor degrade (Fig. 2). Specifically, excessive production of free tyrosine may be minimal in an aphid because of high Buchnera demand and tight coupling of tyrosine synthesis by the aphid to phenypyruvate supply from Buchnera APS. However, since the pea aphid genome is the first non-holometabolous genome to be sequenced, we cannot exclude the possibility that the loss of the urea cycle and tyrosine degradation pathway is general to many hemimetabolous insects, and not related to specific aspects of aphid biology. Thus, information on the gene content of other aphid species and hemipteran insects will provide a test for our proposal that these traits of the aphid genome are linked to the evolution of the symbiosis with Buchnera.

In conclusion, our analysis of the gene complement of the pea aphid has yielded qualitative predictions of the amino acid relations of the pea aphid and Buchnera APS that are indicative of far greater intimacy and complexity than previously envisaged. Further, our analysis provides an explanation for the apparent paradox that aphids with Buchnera can synthesize essential amino acids even though Buchnera lacks some of the necessary genes. Our genomic analysis provides well-defined hypotheses that can be tested by quantitative in silico analysis of pea aphid-Buchnera APS metabolic networks (Thomas et al., 2009) and by empirical investigation of the nutritional physiology of the symbiosis.

Experimental procedures

Annotation of amino acid metabolism genes in the pea aphid genome was initiated with genes present in the D. melanogaster KEGG database (Kanehisa & Goto, 2000) and executed following standard pea aphid manual annotation guidelines (Legeai et al., 2010). Following the initial manual annotation, the AcypiCyc database (a BioCyc database for the pea aphid) became available (see http://pbil.univ-lyon1.fr/software/cycads/acypicyc/home), enabling the manual annotation to be complemented by the automated annotation pipeline of AcypiCyc (see Supporting Information Table S1).

Amino acid metabolism genes that were not identified by manual annotation were used in TBLASTN searches against the A. pisum EST databases on NCBI and AphidBase (http://www.aphidbase.com/aphidbase/) and A. pisum unassembled WGS reads on AphidBase. Genes with no significant results in any of these searches were interpreted as absent from the pea aphid.

For each pea aphid gene annotated as a transaminase, PhylomeDB (http://phylomedb.org, Huerta-Cepas et al., 2008, 2010) was used to investigate the orthology relationship of each gene with those of D. melanogaster, A. mellifera, Anopheles gambiae, Aedes aegypti and T. castaneum. We annotated clades taking a combined evidence approach, first considering annotations of Homo sapiens genes and then those of D. melanogaster. Pea aphid transaminases were determined to be orthologous when each orthologous cluster contained copies from most or all of the five insect species listed above (Table 2). Pea aphid transaminases were determined to be paralogous either when they were linked by a terminal node or, when multiple pea aphid genes were found within a single monophyletic cluster.

Acknowledgements

We thank Augusto Vellozo for providing data from the AcypiCyc database, UM undergraduates Anne Chen and Isabel Llanes for technical assistance. For financial support, we thank BBSRC and ANR (grant BB/F005342/1 to GHT and HC), USDA (grant 2005-35604-15446 to GJ and ACCW), University of Miami Start-up Funds (to ACCW), The Sarkaria Institute for Insect Physiology and Toxicology (to AED) and the University of Miami Honors Summer Research Program (to JFS).

Ancillary