Obtaining a genome sequence for a species for which no related genomes are available is a long task that requires the involvement of a motivated and collaborative community. In the case of the pea aphid, the IAGC (founded in 2003) wrote a white paper describing the advantage of getting the genome sequence of an aphid and rapidly agreed to focus on one model species: A. pisum. The pea aphid is considered as a model aphid species for several reasons: a low number of chromosomes (n = 4) and a medium-size genome (525 Mb, six times less than the Homo sapiens genome, but three times that of Drosophila melanogaster [D. melanogaster]) (Table 1). Furthermore, the pea aphid has been often used for physiological studies – mostly for nutrition physiology and endosymbiosis interaction. A. pisum is a quite large aphid (several millimeters long for an adult) amenable to dissections. This is a species easy to rear; complete life cycle and genetic crosses can be performed in controlled room on a single plant species (Vicia fabae). Finally, A. pisum represents an extraordinary example of host-races adapted to different host plants (among the Fabacae) [9].

The financial support was obtained from the National Human Genome Research Institute (USA): it is noteworthy that a large majority of sequencing projects (not only for insects) are funded by the USA, a kind of new quest such as the moon project in the 1960s. The sequencing of the pea aphid genome was performed at the Baylor College of Medicine (USA). The strategy used was the Whole Genome Shotgun (WGS) which consists in:

• • extracting genomic DNA;
• • cutting it by different restriction enzymes and generating small DNA fragments;
• • cloning the cut fragments into different vectors (plasmids for small fragments, Bacterial Artificial Chromosomes for large fragments);
• • mass sequencing.

In the case of the pea aphid, three libraries were constructed corresponding to approximately 2, 10 and 100 kb fragments, respectively (Baylor College of Medicine, Houston, USA). Sequencing was performed by the regular Sanger technique that provides long reads (> 600 bp) with a high level of accuracy. In the WGS strategy, the following step is to reconstruct the genome. This is the assembly step, one of the most critical for a genome project. Genome assembly is performed by specialized algorithms that work by alignments of overlapping fragments from the different libraries. Because of the relatively small size of the sequences (600 pb) and the presence of numerous and long stretches of repeated sequences within a eukaryotic genome, it is difficult, even impossible in some cases, to assemble correctly all the raw sequences. To partially overcome this limitation of WGS, one approach consists in sequencing a very large number of DNA fragments, in order to increase the probability that each DNA fragment is sequenced at least once, and to get more overlapping fragments. That is why for a genome project, the final number of sequences is largely higher than the estimated size of a genome. This is the so-called “coverage”. The pea aphid genome has, for example, a 6X coverage, meaning that the final number of sequenced nucleotides corresponds to six times the number of nucleotides in the pea aphid genome. Thus, for the 525 Mb pea aphid genome, we obtained 525 × 6 = 3150 million sequenced bases.

The assembly (performed by the Baylor College of Medicine using the Atlas assembly pipeline) of the pea aphid genome ended with approximately 22,000 scaffolds; this means 22,000 different stretches of genomic DNA still with gaps among the different scaffolds. We are thus quite far from the four expected haploid chromosomes. More sequencings, as well as genetic maps, are required to improve the assembly. Among all the sequences, 25% were not assembled, probably corresponding to repeated regions of the genome. Nevertheless, despite these imperfections, the newly assembled pea aphid genome already provides a gold mine of information for gene and other features analyses (see below).

After sequencing and assembly, the third essential step for a genome project is the annotation. The aim is to describe the different features of a genome: genes encoding proteins, non-coding RNAs, repeats, transposons, telomeric and centromeric sequences… The strategy is based first on a scan of the genome with specialized algorithms able to detect these different features, and a second phase where experts check manually what the algorithms found. For the pea aphid genome, several algorithms were used to find genes encoding proteins (find a start and stop codon, intron/exon junctions…) and using homology search with other species. These programs also used other genomic resources that were available for the pea aphids, such as ESTs that correspond to sequences of mRNAs, thus putative protein encoding genes [10]. For the pea aphid, a first set of approximately 10,000 putative genes was defined, with strongly supported predictions due to their homology to well described genes (from Drosophila melanogaster for instance) or the detection of pea aphid ESTs. This constitutes a set of high-confidence genes, similar to the so-called “RefSeq” catalog provided by the NCBI [11]. A second set of approximately 24,000 other genes was defined by combining several of the prediction algorithms results using the GLEAN method, used for the first time to define the gene set of the Apis mellifera [12], but these predicted genes have less biological evidence supporting their existence and function. Thus, a total of approximately 34,000 predicted genes were found on the pea aphid genome and this is the Official Gene Set associated with the first version of the assembly.

This number seems high when compared to other insect species, nearly the double of what was described for Diptera, Lepidoptera, Coleoptera and Hymenoptera. At first, such a high number of predicted genes was thought to correspond to sequencing artifacts that could have affected the overall quality of the genome assembly. Furthermore, allelic polymorphisms might also cause the generation of different assembled scaffolds that in reality correspond to the same locus (a factor increased by the choice of a single generation of self-cross prior sequencing). These possibilities were carefully evaluated for the pea aphid genome, but it was shown that such potential errors could not account for the high number of genes. A detailed analysis of the genes in the pea aphid genome led to the identification of many gene duplications. Analysis of the evolutionary distances among duplicated genes clearly shows that duplication occurred at a unusually high level over a long evolutionary history (dating back to the diversification of the aphid group) rather than at a precise time. Even though the genome sequence is scattered with 22,000 scaffolds, which limits the physical description of gene expansions (i.e. some blocks of duplication involving many genes – as, for example, duplication of chromosomes or parts of chromosomes – could have occurred and not be detected), the very low frequency or synteny among duplicated genes (paralogons) strongly suggest that duplications did mostly arise as repeated events involving single genes. Why the frequency of duplication (or the propensity to conserve duplicated copies once they arise) is so much higher in aphids than in other insects still remains to be explained.

At the same time as the pea aphid genome sequencing, the genome of Daphnia pulex (http://wfleabase.org/) was released: this arthropod has the peculiarity to share with aphids both the capacity to express phenotypic plasticity towards local environmental changes, and the cyclical parthenogenesis ability [8]. It was puzzling to notice that the Daphnia genome also possessed a large number of predicted genes (approximately 40,000) with many duplications. Whether this similitude between the two organisms is a purely random event or it reflects the adaptation to phenotypic plasticity is an intriguing question [13]. This interesting comparison clearly shows one of the advantages of global genome comparisons between species with evolutionary clues and how this can open new research perspectives.

What are the functions of these 34,000 predicted genes? Nowadays, putative gene functions rely mainly on homology search; sequence-A of the pea aphid resembles sequence-B of Drosophila that is known to have function X. In the best-case scenario, function X in Drosophila has been experimentally demonstrated. But in most cases, the relation of homology is indirect: sequence-A of the pea aphid resembles sequence-B of Bombyx that resembles sequence-C of Drosophila that resembles sequence-D of mice that has been showed to have the same function as sequence-E from human. In such cases, one needs to keep in mind that gene function annotation based on homology, albeit essential, only provides putative functions that have to be demonstrated experimentally for a given species of interest (here, the pea aphid). Thus, most of the pea aphid annotated genes have “putative functions” since very little experimental data demonstrating a specific function are available for aphid proteins. That is the reason why a manual annotation step is critical: this is a long procedure by which researchers (i.e. human beings and not machines) carefully check the predictions made by different algorithms. Approximately 2000 of the 34,000 genes were manually examined for the pea aphid genome [14]. This manual annotation required a strong and large community of specialists. A list of nearly 30 annotation groups were defined within the aphid community, each specialized on a given gene family functional role (e.g. meiosis, neuropeptides, immune response…) and each annotation group defined the gene of interests to be analysed. Such careful analysis allowed the identification of several features in different gene families. Manual annotation never ends since new descriptions and demonstrations might become available over time: as an example, the fifth version of the annotated D. melanogaster genome has been released in 2009, 9 years after the first draft of the genome and further annotations efforts are still ongoing in the community.

The main characteristic of the gene functions of the pea aphid have been described and reviewed [1]. Briefly, large duplications of genes affect many protein families involved in different functions, including microRNA synthesis, chromatin modification [14] and sugar transport [15]. In all these cases, it is still too early to fully explain the adaptive processes behind these duplications, but further research will help understanding this intriguing phenomenon. Some duplicated copies are under positive accelerated selection, suggesting they are acquiring new functions that remain to be characterized. Besides gene duplications, gene losses were also identified in some other pathways, as for example the missing selenoproteins (necessary for processing rare codons encoding selenocysteine). This is also the case for several genes involved in immune function: some genes from the IMD (immunodeficiency) pathway were not detected in the pea aphid genome whereas they are present in genomes of other sequenced insects. Complementary expression data indicated that the immune response of the pea aphid to different biotic challenges is reduced compared to other insects [16]. Several hypotheses can be raised such as a relatively sterile feeding source (the phloem sap) or the presence of a cortege of bacterial symbionts that might have affected during evolution the capacity of aphids to reject bacterial intruders. In fact, symbiosis is central to aphids biology, in particular the primary symbiont of the pea aphid Buchnera aphidicola lives in specialised aphids cells the bacteriocytes, and it is vertically transmitted to the offspring. This obligatory association had been known to provide essential amino acids lacking in the phloem sap. Buchnera genome sequence is available [17] and the comparison with the aphid genome data has revealed a clear integration and complementarity between the two organisms in several pathways, in particular as expected in the amino acids synthesis and degradation [18].

All genomic resources require solid, efficient and stable tools to store, analyze, organize, distribute and display all the available data. This is the aim of AphidBase, a centralized bioinformatics resource that was developed to collect all genome related data and to facilitate community annotation of the pea aphid genome by the IAGC (http://www.aphidbase.com) [19]. This essential genomic repository is complemented for the pea aphid by two other resources: the PhylomeDB (http://phylomedb.org) containing the phylogenomic analysis of the pea aphid gene set [20,21] and the AcypiCyc database (http://pbil.univ-lyon1.fr/software/cycads/acypicyc/), a database reconstructing the metabolism of the pea aphid and its symbiont bacteria Buchnera aphidicola.

One of the most interesting aspects of obtaining a new genome is not so much its static description than its comparative study with the genomes from other organisms, as it may reveal major changes in gene repertoires and genome organization, which can be related to shifts in ecological traits and biological novelties. For example, the evidence discussed above of a massive number of duplicated genes in the pea aphid requires phylogenetic studies to characterize the relationships among the duplicated genes: are some of those ancient duplicates that were lost in other insects, or did most of these gene duplications arise in an ancestor of modern aphids, after they diverged from other insects? It is already clear that the second scenario prevails in general, as most gene expansions seen in A. pisum appear to be monophyletic (they are specific to aphids). One limitation of current comparisons between genomes of aphids and other insects is, however, the considerable evolutionary distance between the different insect orders (i.e. at least 300 My separate A. pisum and all other completely sequenced insect genomes, whereas only a few My separate humans and other vertebrates). It will be rapidly necessary to develop genomic data, if possible complete genomes from other aphid species to allow finer scale studies: this has already been undertaken using EST data sets obtained in different species (in particular in the peach potato aphid Myzus persicae, another pest species found on many different crops). The advantage is that the sequences of two not too distant aphid species are more easily comparable in many ways: assessing their status as orthologous genes, aligning them, and estimating their evolutionary rates (especially the rates of synonymous and non-synonymous mutations indicate the selective pressure that characterizes the gene, and can only be accurately estimated between not too distant organisms). The comparison between A. pisum and M. persicae partial gene sets have already shown that some genes appear to evolve at very high rates, possibly due to diversifying (or “positive”) selection [22]. In the near future, extensive genomic data will be collected in several other aphid species, which will provide a dynamic framework of the evolution of the pea aphid genome. It will, in particular, allow one to date much more precisely the many duplications seen in A. pisum and to determine if and how gene duplications affect sequence evolutionary rates (an important question in genome biology, [23]) and if these can be related with different adaptations of the different aphid species: for example M. persicae has been shown to have a recent large expansion of esterases associated with insecticide resistance. Many more species-specific expansions or losses of genes will likely be detected once more extensive genomic data is obtained from different aphids.

The pea aphid and Buchnera aphidicola constitute the first example of a vertically transmitted endosymbiosis where both the genomes of the host and its primary symbiont are available. This unique resource opens the way to functional genomics research to better understand the regulatory networks underlying symbiosis. Much is already known about the pea aphid – Buchnera symbiosis from different studies and several experimental approaches have been used to study this intimate relationship. Genomics level work have been performed in the past few years thanks to the availability of the bacterial symbiont genome sequence, using both in silico evolutionary studies [24] and experimental approaches [25]. At present, the genomic status of both organisms opens the way to global integrated functional genomics approaches that will generate data allowing a systems level analysis of this endosymbiosis.

As previously discussed, the relationship between the pea aphid and it primary symbiont Buchnera is known to be centered on the exchange of essential amino acids. Even if we know that this association is obligate and long lasting, little is known about the molecular cross talk underlying and regulating the symbiotic relationship. A first step towards a better understanding of the metabolic relationship has been achieved through the analysis of the reduced symbiont genome [17]; this is now complemented by the integration of knowledge about the host genome that it has allowed the full symbiosis metabolic network reconstruction in the AcypiCyc database. This BioCyc database was build using an ad hoc annotation management system (including a database and several software tools), as part of the pea aphid genome annotation effort. A comparison of the integrated annotation of AcypiCyc with manual annotation work showed very good consistency and has allowed a first characterization of the amino acid metabolism [18], as previously discussed. This example is a clear indication that thanks to the availability of both genomes global studies of the metabolic network of symbiosis will be possible in the future and these will complement the knowledge on the Buchnera metabolic network [26]. Modelling approaches will also lead to the design of better experiments to dissect the complex metabolic partnership. It is also clear that future integration of secondary symbiont sequenced genomes data in this kind of database are key to perform analyses to further our understanding of more complex and composite metabolic symbiotic relationships.

In recent years, several studies of the gene expression patterns for the pea aphid have been performed using both EST sequencing [10,27,28] and custom microarrays platforms [29–31]. Inference of regulatory genetic network methods, starting from high-throughput experimental “omics” data and transcription factors binding site mining, has developed in recent years [32]. The genome sequence of the pea aphid is a stepping-stone to transcriptome analysis leading to the reconstruction of the genetic networks underlying the fascinating biology of this organism. Beyond metabolism, for which both the metabolic and genetic networks could be now analysed at the same time, transcriptomic and proteomic experiments could be integrated in modelling studies to better understand other aspects of pea aphid biology, as for example the unique characteristics of its reduced immune system [16].

The genomic resources generated for functional analysis on the pea aphid may also serve studies on population and quantitative genetics in several ways. High-throughput sequencing allows developing, without spending time and money in the laboratory, polymorphic markers that could be used to analyze population structure, to construct genetic maps and to conduct genome-wide association studies aimed at identifying genes involved in a given biological function. Indeed, sequencing of a genome generally requires sequencing more than one copy for assembly (see above) and usually more than one specimen of the target species. Thus, sequencing traces contain allelic variants of the same individual (intragenomic variation due to heterozygosity) and of different individuals (intergenomic variation due to population polymorphism). This diversity can be extracted through automated procedure at the scale of the whole genome by blast comparisons and sequence alignements. Variation can be used to detect polymorphic loci such as microsatellites (stretch of di-, tri- or tetranucleotide repeats), SNPs or indels (insertion/deletion sites). Therefore, an almost unlimited number of polymorphic markers can be derived from high-coverage genome-sequencing projects. EST databases, that usually accompany whole genome sequencing projects, can be also mined for polymorphic loci.

Analysing nucleotide diversity across genomes of several to many individuals provides a deep understanding of the evolutionary forces acting either at whole genome level or at specific genomic regions or sites. It can notably help detecting loci under selection and involved in adaptive divergence as well as reconstructing historical demographic events such as colonization, extension and bottleneck events. Also, numerous polymorphic markers are needed to built genetic maps and search for QTL underlying complex traits, e.g. reproduction, development, immunity and defense.

So far, population genetic studies of the pea aphid have been conducted with a limited amount of markers (< 20) isolated in the laboratory and used to measure population differentiation at neutral loci among host-adapted races or biotypes [9,33]. Recently, population genomic approaches have been launched for resolving the genetic architecture of complex traits of the pea aphid such as reproductive mode variation [8], ecological specialization and dispersal [34]. QTL maps have been elaborated to locate chromosomal position of performance and preference loci underlying host adaptation [35].

We expect more dense genetic maps to be constructed in a very near future, facilitating the assembly of the current 22,000 scaffolds into four linkage groups corresponding to the four haploid chromosomes of the pea aphid. We also anticipate the genome sequencing of other pea aphid genotypes with distinct geographic or host plant origins. Combined with functional genomics and other post-genomic methodologies, population genomics and genome-wide association studies will allow deciphering the mechanisms and the evolutionary histories of some of the peculiar and complex biological adaptations of the pea aphid.

3.4.1 Genomics and behavioural ecology

3.4.2 A genome for aphid chemical ecology