Background & Summary

Nutritional symbiosis with bacteria has contributed significantly to the evolutionary success of insect taxa that feed on unbalanced diets such as phloem sap, blood or wood1. Indeed, in many insect species, the synthesis of nutrients (e.g. amino acids and/or vitamins) that are not present in sufficient amounts in the diet is ensured by an obligate nutritional symbiont, sometimes acquired tens of millions of years ago2. These symbionts are generally transmitted faithfully from generation to generation (i.e., by vertical transmission), and compartmentalized in specific host cells called bacteriocytes3. These cells mediate the metabolic exchanges between the insect and its bacterial partners, and regulate populations of obligate symbionts according to the insect’s nutritional needs throughout its life cycle4,5. However, this intracellular lifestyle causes Muller’s ratchet which, combined with severe population bottlenecks during vertical transmission and the relaxation of purifying selection on genes no longer needed in the context of interdependent association, leads to reductive genome evolution6. In some insect lineages, the ancestral nutritional symbiont has undergone such severe genomic erosion that it is no longer able to supply alone the compounds essential to its host’s physiology on its own, and is metabolically complemented by more recently acquired nutritional symbionts1,7. Thus, in many insect taxa, nutritional symbiosis does not rely on a single obligate symbiont but on a consortium of nutritional symbionts that evolve together in the same host, either within the same bacteriocytes8,9,10,11, or in distinct but anatomically connected bacteriocytes12,13,14,15,16.

These multi-partner symbiotic system have evolved in a wide range of hemipteran taxa, including several insect pests, including Psylloidae (psyllids13,17,18,19), Aleyrodidae (whiteflies20,21,22,23), Pseudococcinae (mealybugs8,24,25,26), Auchenorrhyncha (cicadas, leafhoppers, planthoppers and treehoppers12,27,28,29), Adelgidae (adelgids15,30,31,32,33,34,35) and Aphidoidea (aphids14,36,37,38,39,40,41). However, little is known about the development and functioning of these systems. This is largely due to the fact that their study is hampered by a range of constraints such as the difficulty of rearing some of these insects (e.g. cicadas), the absence of a clonal phase enabling individuals of identical genotypes to be obtained (e.g. cicadas, psyllids, whiteflies, leafhoppers), and their very small size (e.g. psyllids, whiteflies). A candidate species that overcomes all these difficulties is the cereal aphid S. maydis (Chaitophorinae) (Fig. 1A), a species that feeds on many species of grass (Poaceae) and is distributed in Europe, much of Asia and has recently reached North and South America where it is considered an invasive species that can damage cereal crops (Fig. 1B)42. S. maydis is easy to collect in the field, to rear and to reproduce clonally, making it an ideal species for experimental studies. Another advantage of this aphid species is that the genomes of the ancient obligate symbiont Buchnera aphidicola and the more recently acquired co-obligate symbiont Serratia symbiotica have recently been sequenced and annotated37. The two nutritional symbionts are compartmentalized in distinct bacteriocytes: S. symbiotica is confined to large syncytial secondary bacteriocytes sandwiched between uninucleate primary bacteriocytes containing B. aphidicola (Fig. 1C-D). This case of dual endosymbiosis is particularly relevant for studying how nutritional symbionts dwelling in distinct but contiguous bacteriocytes can collaborate metabolically with each other and with the host. However, the study of this valuable symbiotic system suffers from a lack of genomic information on the insect host. Assembling and annotating the complete genome sequence of S. maydis would be a highly informative and fruitful resource for deciphering multiple aspects of the species’ biology, and in particular the processes governing cooperative life between insects and anatomically integrated symbionts that form a metabolic unit in a three-way mutualistic symbiosis.

Fig. 1
figure 1

The cereal aphid Sipha maydis and its di-symbiotic system. (A) An adult surrounded by several nymphs, all feeding on bread wheat, Triticum aestivum. (B) Distribution of S. maydis worldwide. Red triangles represent collection locations reported in the literature. (C) Serratia symbiotica (red) is compartmentalized into syncytial secondary bacteriocytes (Sba) sandwiched between the uninucleate primary bacteriocytes (Pba) housing Buchnera aphidicola (green), forming a horseshoe-shaped bacteriome (green, red, and blue signals indicate B. aphidicola cells, S. symbiotica cells and host insect nuclei, respectively). (D) Close-up view of primary and secondary bacteriocytes showing their embedded layout.

Full size image

We present here the first complete assembly of the cereal aphid S. maydis using a PacBio high fidelity (HiFi) approach. The final assembly is 409.54 Mb in length, with a scaffold N50 of 187.22 kb and 95.5% completeness, providing an excellent genomic resource for further research on S. maydis. Structural annotation reveals that the genome contains 29% repeat sequences, and 24,453 protein-coding genes. Functional annotation focused on metabolism and metabolic pathway reconstruction, identifying the 2,541 enzymes of S. maydis involved in 273 metabolic pathways. These genomic and metabolic data provide a unique tool for studying the influence of bacterial symbiosis on insect genome evolution, and for exploring in depth the biology of S. maydis, an in particular the mechanisms underpinning an interdependent tripartite collaborative life between an insect and its prokaryotic partners.

Methods

Sample collection and genome sequencing

A colony of S. maydis sampled on Hordeum vulgare in Midelt (Morocco) in April 2016 was used to generate a clonal line from a single individual. Aphids were reared on Triticum aestivum (bread wheat) under long-day conditions (16 h light, 8 h dark) in a room maintained at a constant temperature of 20 °C to ensure parthenogenic reproduction. Thirty adult individuals were used for DNA extraction. Only the heads were used to minimize DNA contamination by the symbionts B. aphidicola and S. symbiotica that are present only in the abdomen, and whose genomes have been sequenced previously from the same aphid clonal line37. Whole insects were first surface sterilized with 99% ethanol, rinsed with sterile water and then immersed in 70% ethanol where the heads were dissected with microscissors before being stored directly in a sterile plastic tube at −80 °C prior to DNA extraction. DNA extraction was performed using phenol-chloroform. Briefly, tissues were homogenized in 500 ml STE buffer (100 mM NaCl, 1 mM EDTA, 10 mM Tris-Cl, pH 8.0) with a sterile pestle, then treated with 25 µl SDS 10% and 3 µl proteinase K (20 mg/ml). After a two-hour incubation at 55 °C with frequent mixing, the sample was treated with 6 µl RNase (10 mg/ml) and incubated at 37 °C for 30 min. Genomic DNA was purified by two successive extractions with phenol:chloroform:isoamyl alcohol (25:24:1, v/v/v) followed by extraction with 1 vol of chloroform:isoamyl alcohol (24:1, v/v/v). Genomic DNA was then precipitated by 0.7 volumes of isopropanol. After washing the pellet with 70% ethanol, genomic DNA was recovered in TE buffer (1 mM EDTA, 10 mM Tris HCl pH 8). DNA concentrations and quality were assessed using NanoDrop (Thermo Fisher Scientific, Waltham, MA, USA), agarose gel electrophoresis and Qubit fluorometer (Thermo Fisher Scientific). Sequencing libraries were prepared using the Express Template Prep Kit 3.0 (Pacific Biosciences, Menlo Park, USA) and whole-genome sequencing was performed on the PacBio Sequel IIe system at the Genomics Core Leuven (KU Leuven, Leuven, Belgium) using the Sequel® II Binding Kit 3.2 (Pacific Biosciences).

Genome assembly and evaluation

The complete S. maydis genome assembly and annotation workflow, including quality assessment steps, is shown in Fig. 2.

Fig. 2
figure 2

Flowchart highlighting the Sipha maydis genome assembly and annotation process, including quality assessment steps.

Full size image

The genome size of S. maydis was estimated using k-mer analyses from raw PacBio HiFi reads. A k-mer (k = 21) distribution was generated with Jellyfish43 (v2.2.10) using the PacBio HiFi reads and genome size was estimated using three different strategies: i) findGSE44 v1.94.R, ii) gce45 v1.0.2 and iii) the ratio of total distinct k-mers divided by the frequency mode of the k-mer distribution using R46 v4.2.1 as described in Hon et al.47. The S. maydis genome was assembled from the PacBio HiFi raw reads using hifiasm48 v0.18.9-r527 with default parameters. The primary assembly was then screened for contaminants using the NCBI Foreign Contamination Screen (FCS) and seven contaminant scaffolds corresponding to the two endosymbiont genomes (B. aphidicola and S. symbiotica) were removed from the assembly prior to the annotation step. The accuracy and completeness of the assembly were assessed using (i) QUAST49 v5.0.2 with the –large and -k options, (ii) BUSCO v5.4.650 using the Insecta ODB10 database, and (iii) KAT51 v2.4.2 to compute shared k-mers between PacBio HiFi reads and the assembly. A total of 3.70 Gb of PacBio HiFi reads with a mean read length of 6.89 kb were assembled to generate a 409.54 Mb draft genome assembly consisting of 3,570 contigs with a N50 length of 187.22 kb and a largest contig of 1.25 Mb (Table 1).

Table 1 Genome assembly and annotation statistics of Sipha maydis as compared to its close relative Sipha flava.
Full size table

The assembly size is comparable to the genome size estimate of ~433 Mb using k-mers (findGSE: 446.20 Mb; gce: 421.93 Mb; total distinct k-mers divided by the frequency mode of the k-mer distribution: 431.32 Mb). The genome assembly was found to have a high level of completeness (95.5%). Of the 1,367 Insecta BUSCOs, 91.0% were complete and single-copy, 4.5% complete and duplicated, 0.8% fragmented and 3.7% were missing (Table 2). The alternative haplotype-resolved assemblies produced by Hifiasm have however a reduced total length (347.06 Mb and 329.15 Mb), N50 (70.77 kb and 71.73 kb) and complete BUSCO scores (82.6% and 81.3%) (available at: https://doi.org/10.57745/6RYSBE52).

Table 2 BUSCO assessment of Sipha maydis genome assembly compared with its close relative Sipha flava (Insecta_odb10 scores, n:1,367).
Full size table

The mitochondrial genome was assembled using the MitoHiFi pipeline53 to generate a 16,379 bp genome consisting of 37 genes, including 13 protein-coding genes, 2 rRNAs, and 22 tRNAs, with a GC content of 15.43%.

Gene prediction and general functional annotation

A de novo repeat library was generated using RepeatModeler54 v1.0.11. RepeatMasker55 v4.1.2 was then used with the de novo filtered repeat library to identify and soft-mask repeats in the draft assembly prior to annotation. Ultimately, we identified 117.21 Mb of repetitive sequences, accounting for 28.62% of the assembled genome (Table 3).

Table 3 Repetitive sequences in the genome of Sipha maydis.
Full size table

After masking the repeat sequences, the structural annotation (i.e., gene prediction) was performed using the BRAKER356,57,58,59,60,61,62,63,64,65,66,67,68 pipeline v3.0.3 using ab initio prediction, homology searching and transcriptome-based approaches to predict protein-coding genes. For transcriptome-based prediction, the pipeline used 16 RNAseq libraries (PRJNA1031833)69 that were aligned to the soft-masked genome using HISAT270 v2.2.1. For the homology-based approaches, annotated proteins from Sipha flava genome annotation (GCA_003268045.1) and the Arthropoda protein dataset from OrthoDB71 v11 were downloaded. The final set of protein-coding genes was retrieved from the Augustus predictions56,57. The completeness of the annotated protein set was assessed using BUSCO50 v5.4.6 and the Insecta ODB10 database. A first general run of functional annotations of predicted proteins was carried out using EnTAP72 v0.10.8-beta. Comparisons were performed against UniProtKB/Swiss-Prot73 and NCBI RefSeq invertebrate annotated proteins (https://ftp.ncbi.nlm.nih.gov/refseq/release/invertebrate/) as reference databases. Hence, we identified and soft-masked 28.62% (117.21 Mb) of the S. maydis genome as repeated sequences. After masking those repeated sequences, a total of 24,453 protein-coding genes were predicted using a combination of ab initio, homology-based and transcriptome-based approaches. The completeness of the gene prediction revealed that 94.8% of BUSCO genes were successfully detected (90.0% are single-copied and 4.8% are duplicated).

Functional annotation of metabolism

We used several methods to perform a functional annotation of the S. maydis enzyme set: (i) the online KAAS – KEGG74 v2.1 automatic annotation server against both “For gene” and “For eukaryotes” representative sets, (ii) the v2 of the PRIAM75 tool, (iii) the Blast2GO76 pipeline v3.5 and (iv) the InterProScan77 v5.56 pipeline with a local installation for faster data generation. These methods generated information such as EC numbers, KEGG Orthology and Gene Ontology related to the protein sequences. All annotations were collected in a SQL database using CycADS78 and associated with the genomic information data. Default settings were used for software configurations and the BLAST alignments (prior to the Blast2GO analysis) were performed against the curated UniProtKB/Swiss-Prot79 protein sequence database.

Metabolic network reconstruction

The final step in assessing the quality of the S. maydis genome was the reconstruction of its metabolic network, which validated the functional annotation of this organism’s enzyme set. This expert reconstruction step also makes the network directly accessible to the scientific community through the exploration of the dedicated metabolic database we make publicly available, or as an additional dataset that can be uploaded by users in suitable formats (e.g., sbml and Biopax). The enriched gene records containing all annotations were extracted from the CycADS SQL to generate the corresponding BioCyc-like metabolic database using Pathway Tools v27.080,81, which we named SipmaCyc, according to current convention. In the summary section for each gene/protein resulting page, the information relative to the annotation results was recorded to allow the researchers to evaluate the confidence for each putative function assigned to a protein. Incomplete EC_numbers (i.e., classes and subclasses) and EC numbers inferred from a single Blast2GO annotation were excluded from the network, even though the annotation remains accessible for users in the gene description.

Since the functional annotation of metabolic pathways in an insect dependent on nutritional symbionts cannot be done without taking into account their metabolic contributions, we validated the S. maydis metabolic annotations by assessing their homogeneity and correct integration with those of its symbiotic partners, B. aphidicola and S. symbiotica, using the CycADS annotation system78,82. This led to the production of an integrated metabolic network of S. maydis and its bacterial associates, which we made publicly available on ArtSymbioCyc82 (http://artsymbiocyc.cycadsys.org/), a collection of metabolic databases dedicated to arthropod symbioses. S. maydis encodes 26,059 predicted proteins from its 24,466 protein-coding genes, of which 2,523 are enzymes involved in 273 metabolic pathways. The SipmaCyc database in ArtSymbioCyc provides a complete description of the central metabolism of S. maydis at the genome scale (Fig. 3) and enables users to visualize and explore individual metabolic networks at the level of compounds, reactions, or pathways.

Fig. 3
figure 3

Schematic overview of the S. maydis metabolic network. The figure contains all the 38 metabolic categories, of which only 21 are highlighted for a more reader-friendly representation. Users can explore this map in ArtSymbioCyc, also having access to all reactions and metabolites in each pathway.

Full size image

Data Records

The sequencing data that were used for the genome assembly and annotation have been deposited in the NCBI Sequence Read Archive with accession numbers SRP44391883 and SRP46826369 respectively. This genome assembly is available under accession number GCA_034509805.184.

Data on metabolic network reconstructions are available at Recherche Data Gouv (https://doi.org/10.57745/6RYSBE)52 as well as in the ArtSymbioCyc collection (http://artsymbiocyc.cycadsys.org/)82. Metabolic network reconstructions and the resulting BioCyc metabolism databases are available in the ArthropodaCyc collection82 (https://arthropodacyc.cycadsys.org) for S. maydis alone (organism database: “Sipha maydis”) and in the ArtSymbioCyc collection (http://artsymbiocyc.cycadsys.org/) for S. maydis associated with its two obligate nutritional symbionts (organism databases: “Sma-Sipha maydis” and “Sma-Sipha maydis holobiont”). A single repository (Recherche Data Gouv)52 was created to unite (i) the S. maydis genome assembly commands file (txt), (ii) the genomic files (fasta primary and haplotype-resolved assemblies), (iii) the gtf/gb structural annotations for both the genome and the mitochondrion and (iv) the functional annotations (tabular text), reactions (sbml) and network in Biopax format for the three partners composing the symbiotic system.

Technical Validation

The quality of the S. maydis genome assembly was assessed by computing several metrics: (i) comparison with the estimated genome size, which is ~433 Mb (see above); (ii) genomic BUSCO analyses, which identified 95.5% of Insecta BUSCOs in the S. maydis genome (91.0% are single-copied and 4.5% are duplicated, Table 2), and 94.8% of Insecta BUSCOs proteins in its predicted gene models (90.0% are single-copied and 4.8% are duplicated, Table 2); (iii) comparison with the PacBio HiFi reads using QUAST and KAT which showed that 99.95% of the k-mers (k = 27) of our assembly are covered by the k-mers from the PacBio HiFi reads and 99.73% of the PacBio HiFi reads could be mapped into the assembly. Despite a low sequencing yield (3.70 Gb) and small read lengths (6.89 kb), the aforementioned quality metrics indicated that the S. maydis genome assembly has a high level of completeness and is of high-quality.

The functional annotation of the S. maydis genome coding for central metabolism is supported by the use of our CycADS expert system and leads to the possibility of reconstructing the metabolic network, whose integrity and consistency can be tested using the comparative tools of the ArtSymbioCyc database interface82. As an example, Table 4 shows the distribution of S. maydis reactions in the top 6 levels of the Enzyme Commission classification, which is fully consistent with those of the other insects in the database.

Table 4 Distribution of S. maydis reactions across the 6 top-level categories identified by the Enzyme Commission.
Full size table