Whole Genome Sequence of the Soybean Aphid Endosymbiont Buchnera aphidicola and Genetic Differentiation among Biotype-Specific Strains.

Endosymbiosis with microorganisms is common in insects, with more than 10% of species requiring the metabolic capabilities of intracellular bacteria for their nutrient acquisition. Aphids harbor an obligate mutualism with the vertically transferred endosymbiont, Buchnera aphidicola, which produces key nutrients lacking in the aphid's phloem-based diet that are necessary for normal development and reproduction. It is thought that, in some groups of insects, bacterial symbionts may play key roles in biotype evolution against host-plant resistance. The genome of Buchnera has been sequenced in several aphid strains but little genomic data is currently available for the soybean aphid (Aphis glycines), one of the most important pests of soybean in North America. In this study, DNA sequencing was used to assemble and annotate the genome sequence of the Buchnera A. glycines strain and to reconstruct phylogenetic relationships among different strains. In addition, we identified several fixed Buchnera SNPs between Aphis glycines biotypes that were avirulent or virulent to a soybean aphid resistance gene (Rag1). The results of this study describe the genetic and evolutionary relationships of the Buchnera A. glycines strain, and begin to define the roles of an aphid symbiont in host-plant resistance.

Introduction

Insects are the most abundant and diverse animal class on earth, and are associated with a remarkable variety of symbiotic microorganisms. Their codiversification with intracellular bacterial symbionts (also called endosymbionts) has been well documented, including nearly all groups of phloem sap-sucking insects 1-3. Sap sucking insects often rely on their obligate mutualistic bacteria to synthesize nutrients that are deficient in their purely phloem diet but necessary to sustain normal development and reproduction 4, 5. One of the best studied cases of such symbiosis is between aphids and their obligate bacterial endosymbiont, Buchnera aphidicola (Proteobacteria: Enterobacteriaceae) 6-8.

The mutualism between aphid and Buchnera dates back over 200 million years and neither the insect nor its endosymbiont can survive independently 9. These vertically transmitted bacteria live within specialized bacteriocytes, aggregated to form the organ-like bacteriome located near the aphid midgut 6, 8. Aphids are dependent on Buchnera to produce some of the essential amino acids, vitamins, and sterols that are necessary for normal development and reproduction 10-15. In turn, the aphid provides Buchnera with nutrients, including nonessential amino acids and carbohydrates that are abundant in their phloem-based diet or produced by the host. Genomic evidence suggests that several amino acid biosynthetic pathways are shared between aphid and Buchnera, providing the aphid the ability to regulate the endosymbiont's metabolism 16.

No aphid species poses a greater threat to soybean yields in North America than the soybean aphid, Aphis glycines Matsumura (Hemiptera: Aphididae). Native to southeastern and eastern Asia, the species was first detected in Wisconsin in the summer of 2000 and has quickly spread throughout much of the North Central United States and Eastern Canada 17, 18. Host-plant resistance (referring to the plant's ability to resist damaging insect invasions) is often an effective method for controlling soybean pests, and at least five soybean aphid resistance genes (Rag1, Rag2, Rag3, Rag4, and Rag5) have been reported 19-26. However, biotypes of the soybean aphid have been identified in Midwest growing regions; biotypes are based on the ability to overcome host-plant resistance provided by one or more of these Rag genes. Biotype 2 can override the Rag1 gene 27, while biotype 3 can deride Rag2 resistance 28. Biotype 4 is capable of overcoming Rag1 and Rag2 defenses when expressed both singly and in concert 29. The mechanisms underlying biotype resistance are not well understood; however, recent work indicates Rag1 resistance may be associated with changes to the plant nutritional quality 30 or secondary plant metabolites 31, and elicits a transcriptional response typical of xenobiotic challenge in A. glycines 32. While most ongoing studies continue to focus on the plant and/or insect, one possibility is that the virulent phenotypes may be associated with aberrations in the insect microbiome, particularly the obligate symbiont, Buchnera 33. Recent work on Russian wheat aphids (Diuraphis noxia) and greenbugs (Schizaphis graminum) have revealed correlations between the symbiont profiles and biotypes 34, 35 but their genetic underpinnings have not yet been explored.

Genome sequences are currently available for several Buchnera strains derived from aphids of different tribes and subfamilies 36-39. The genome is small (<1 Mb) in comparison to other bacteria 40, having undergone reductive evolution involving the irreversible loss of genes and regulatory capabilities 37, 41. Many of the genes lost at the beginning of the symbiotic association functioned in gene transfer and/or recombination; thus, Buchnera has evolved independently in different host aphids, with little to no genetic exchange among strains 42. Consequently, phylogenetic relationships between aphids and Buchnera closely mirror one another across deep evolutionary divergences 43-46.

Transcript data has been accumulating for the Buchnera A. glycines strain 47, 48; however, the genome sequence and phylogenetic reconstructions are not yet available. Since Buchnera is not culturable, we performed next-generation DNA sequencing on whole body A. glycines to indirectly generate Buchnera sequence information. We assembled and functionally annotated the Buchnera genome, and reconstructed phylogenetic relationships. Under the premise that genetic differences in Buchnera may contribute to host-plant resistance, we carried out whole-genome resequencing of the Buchnera A. glycines strain in biotypes 1 and 2 to identify single nucleotide polymorphisms (SNPs) between them. We speculated that if SNPs were detected, they may be associated with genes involved in the aphid's nutrient synthesis and/or the counter-response to plant antiherbivory defense mechanisms (e.g. detoxification genes) - processes known to be mediated by microbial symbionts in the insect host. The results of this study characterize evolutionary and genetic associations of the Buchnera A. glycines strain and begin to define the roles of the aphid microbiome in host-plant resistance.

Materials and methods

Isolation of Buchnera DNA

Buchnera are obligatory symbiotic bacteria that have never successfully been cultured. Thus, we took advantage of next-generation sequencing technologies capability of generating sequence data for not only the aphid but the various associated microorganisms, including Buchnera. The A. glycines biotype 1 colony was established in 2008 at Ohio Agricultural Research and Development Center (OARDC) in Wooster, OH, using aphids from the University of Illinois' Soybean Aphid Biotype Stock Center. The biotype 2 colony was founded in 2005 from field collections in Ohio. Both colonies were periodically re-infested with aphids collected from natural populations. Subsequent to aphid collections, host plant resistance conferred by each biotype was validated using the Rag1 expressing soybean line 'A08-123074', and aphid susceptible soybean line 'SD01-76'. Biotype 2 was capable of surviving on the Rag1 plants (i.e. virulent), whereas biotype 1 was not (e.g. avirulent).

Whole body apterous aphids were simultaneously collected from each biotype and genomic DNA was extracted from the pools of 50 aphids using the E.Z.N.A® Tissue DNA Kit (Omega Bio-Tek, Norcross, GA). DNA (1 µg per sample) was used to generate an adaptor-ligated double-stranded DNA library for DNA sequencing using the TruSeq DNA PCR-Free Sample Prep Kit (Illumina, San Diego, CA) following the manufacturer's protocol. Quantification of DNA was done using the Qubit® 2.0 Fluorometer (Life Technologies, Carlsbad, CA) and quality was assessed using the Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA). The samples (biotypes 1 and 2) were diluted to 18 nM and pooled to generate the multiplexed DNA library.

Illumina library synthesis and sequencing

The DNA library (36 fmoles) was sequenced on three flow cell lanes using the Illumina MiSeqTM2000 platform at the Ohio State University Molecular and Cellular Imaging Center. The mean library insert sequence length was 475 bp and both ends of the library were sequenced to generate 250 nt raw paired-end reads. Illumina Analysis Package CASAVA 1.8.2 was used to perform bcl conversion and demultiplexing. Image deconvolution and quality value calculations were performed using the Illumina GA pipeline v1.6.

Genome assembly and gene prediction

The 88.29 million raw paired-end reads of ≥240 bp were imported into CLC Genomics Workbench (v6.5.1, CLC Bio) and trimmed for quality and adapter indexes using the default settings (Ambiguous limit = 2, quality limit = 0.05). The remaining processed reads (87.53 million) were de novo assembled into scaffolded contigs of at least 1,000 bp using an algorithm based on de Bruijn graphs and the optimized parameters defined in Table S1. The final set of 40,643 contigs (mean length of 7,188 bp and N50 of 5,782 bp) was obtained by collapsing reads with ≥90% sequence similarity into clusters and retrieving the longest contig using CD-HIT 49. By mapping contigs against the nr database using BLAST2GO software 50, three contigs were identified as Buchnera aphidicola in origin (Buchnera Ag strain, BAg). The contigs corresponded to the circular chromosome and two circular plasmids. The raw sequence reads can be retrieved from the NCBI short sequence read archive under the accession number SRP045572 (Table 1).

Table 1

NCBI accession numbers for the Buchnera short read data and genome sequence

NCBI Resource	Accession Number
Small Read Archive (SRA)	SRP045572
BioProject	PRJNA256101
BioSample	SAMN02937548 (biotype 1)SAMN02937549 (biotype 1)SAMN02937550 (biotype 2)SAMN02937550 (biotype 2)
Genbank-Genomes	CP009253 (chromosome)CP009254 (pLeu)CP009255 (pTrp)

To assess genome completeness, the preprocessed reads were mapped to each BAg contig using the resequencing function in CLC and the parameters described in Table S1. Since the BAg contigs are all circular, we were able to fill in any missing nucleotide sequence at each end by merging the overlapping sequence. Sequence alignment was also corroborated using this approach. Coverage analysis of the genome was conducted using the create statistics for target regions resequencing functions.

The Rapid Annotations based on Subsystem Technology (RAST) Genome Annotation Server v2.0 was used to annotate the identified Buchnera contigs. Functional annotation was done using the classic RAST scheme and GLIMMER-3 (Release70) gene caller. For comparison purposes, RAST was also used to annotate the Buchnera genome of the greenbug, Schizaphis graminum strain (BSg) 37.

Additional sequence of bacterial origin was identified in our de novo assembly. This included four contigs of the Wolbachia endosymbiont, as well as contigs of Escherichia, Halomonas, Lactobacillus, Rhodospirillum, and Streptomyces species. However, these contig sequences represented only a small fraction of their respective genomes.

Phylogenetic analysis

Single gene matrices and the concatenated sequence of 60 protein-coding genes shared by the Buchnera A. glycines strain and thirteen Buchnera strains from seven aphid species and deposited in the NCBI Bacterial Genomes FTP (accessed February 18, 2014) (Table 2) were used to reconstruct the phylogenetic associations among them. Maximum-Likelihood (ML) trees were constructed from both the individual and concatenated alignments using the CLC Phylogeny Module. The starting tree algorithm implemented the Unweighted Pair Group Method with Arithmetic Mean (UPGMA) and rate variation (4 substitution rate categories, Gamma distribution parameter = 1.0). The 'Model Testing' tool in CLC was used for selection of the best-fit model of nucleotide substitution to be used for maximum likelihood phylogeny tree reconstruction. The tool employs four different statistical analyses to test the substitution models. The reliability of tree topologies was evaluated using 1000 bootstrap replicates. For each phylogeny, the relative substitution rates between pairs of Buchnera strains was calculated, using E. coli as an outgroup. Tree topologies were validated using only the first and second codon positions for tree inference, indicating mutations in the third codon positions were not saturated.

Table 2

The thirteen Buchnera strains from seven aphid species and four tribes deposited in the NCBI Bacterial Genomes FTP (accessed February 18, 2014)

Host Species	Common name	Subfamily	Tribe	Buchnera Straina	# Sequencesb
Acyrthosiphon kondoi	blue alfalfa aphid	Aphidinae	Macrosiphini	BAk	559
Acyrthosiphon pisum	pea aphid	Aphidinae	Macrosphini	BAp (TLW03)	573
				BAp (JF98)	477
				BAp (JF99)	590
				BAp (5A)	555
				BAp (LL01)	577
				BAp (Tuc7)	553
				BAp (APS)	564
Baizongia pistaciae	N/A	Eriosomatinae	Fordini	BBp	504
Cinara cedri	cedar aphid	Lachninae	Eulachinini	BCc	357
Cinara tujafilina	cypress pine aphid	Lachninae	Eulachinini	BCt	360
Schizaphis graminum	greenbug	Aphidinae	Aphidini	BSg	546
Uroleucon ambrosiae	brown ambrosia aphid	Aphidinae	Macrosiphini	BUa	529

aThe Acyrthosiphon pisum strain is denoted in parentheses.

bThe number of sequences were derived from the .ffn files in the NCBI FTP database.

Sequence polymorphism analysis

Single nucleotide polymorphisms (SNPs), insertions, and deletions in the Buchnera genome (see above) were identified between Aphis glycines biotypes 1 and 2 using the quality variant detection re-sequencing function (based on the Neighborhood Quality Standard (NQS) algorithm) in CLC Bio. Only polymorphisms between biotypes with read coverage of ≥25 valid reads were considered. Mutations in coding regions that putatively result in non-synonymous changes were identified using the GLIMMER-3 predicted protein-coding sequences.

Results

Genome assembly and annotation

In this study we assembled the complete genome of Buchnera Ag strain (BAg). To do this, we carried out DNA sequencing of its host aphid species (Aphis glycines). De novo assembly of preprocessed reads and subsequent resequencing analyses produced a complete BAg genome of 638,852 bp, with 25.6% GC content. It consisted of a 628,164 bp chromosome, pLeu plasmid (7,639 bp), and pTrp plasmid (3,049 bp). The chromosome and plasmid sequences have been deposited in the NCBI GenBank archive under the accession numbers CP009253-CP009255 (Table 1).

Average coverage of the chromosome (x = 846.9) and pLeu (x = 777.3) were similar, whereas pTrp had roughly twelve-fold higher coverage (x = 10,244.1). The BAg chromosome and plasmid sizes and sequence alignments were coanalyzed and corroborated using de novo assembly algorithms of CLC Bio and Velvet 51 in the MCIC-Galaxy automated pipeline 52.

The RAST Prokaryotic Genome Annotation Server was used to annotate the genes, which implements a 'highest confidence first' propagation strategy (see methods) and the GLIMMER-3 gene prediction scheme. A total of 593 putative protein-coding genes and 35 RNAs were identified on the chromosome, as well as 7 and 3 protein-coding genes on the pLeu and pTrp plasmids, respectively (no RNAs) (Table S2). Thirteen of the chromosome genes had predicted amino acid sequences of <50 aa and could not be assigned function. Functional categorization of the protein-coding genes is displayed in Fig. 1. Approximately 83% (n = 503) of genes could be assigned to one or more functional groups, based on current Buchnera subsystem annotation.

Fig 1

Functional categorization of the protein-coding genes in the Buchnera aphidicola genome of A. glycines

Table 3 shows the comparison of the BAg genome with selected genomes of Buchnera strains of other aphid species, which was adapted from the KEGG genome database 53 and BuchneraBASE 54. The genome is comprised of 88.5% coding sequence and 25.6% GC content. The BAg genome contained the largest number of protein coding genes, with 603 predicted for the chromosome and plasmids combined. It also contained the smallest number of RNAs, even in comparison to the significantly reduced BBp and BCc genomes. This was primarily due to an absence of sRNAs in BAg, whereas all of the other strains had a minimum of two sRNAs. While it remains difficult to identify sRNAs by sequence inspection, pairwise analyses of the BAg genome with sRNA sequences derived from the other strains was unable to uncover any putative sRNAs. It is unclear whether there are any functional implications for the bacteria or aphid host related to the absence of sRNAs but it could be related to the extreme specialization of A. glycines on two seasonal hosts.

Table 3

Comparisons of Buchnera aphidicola genomes derived from A. glycines (BAg) and selected other aphid species. The data presented was adapted from the KEGG genome database (Kanehisa and Goto 1999) and BuchneraBASE (Prickett et al. 2006)

Feature	BAg	BBp	BAp (APS)	BSg	BCc
Genome
Genome Size (bp)	638,852	618,379	655,725	653,001	422,434
GC content (%)Protein-coding genes	25.6603	25.3507	26.4574	26.3555	20.2362
Chromosome
Size (bp)	628,164	615,980	640,681	641,454	416,380
Protein-coding genes1	N/A (593)	504 (510)	564 (578)	546 (600) 37	357 (371)
RNAs	35	37	36		37
Plasmids
Number	2	1	2	2	1
Total size (bp)	10,688	2,399	15,044	11,547	6,054
Protein-coding genes	10	3	10	9	5

1The number in parenthesis indicates the predicted number of protein-coding genes using the Rapid Annotations based on Subsystem Technology (RAST) Genome Annotation Server v2.0.

Functional comparison between BAg and BSg

Since the BAg and BSg belong to the same tribe and are the most closely related Buchnera strains currently available in the NCBI genome database, we implemented pairwise functional analyses between strains using the RAST annotation for both BAg and BSg. Roughly 85% of the BAg and BSg chromosome sequence overlapped and these regions were ~84% similar. The chromosome sequence of all other Buchnera strains had less than 70% overlap and 82% similarity. The BAg plasmid sequences had slightly less overlap (78%) and similarity (81%) with BSg plasmids. The gene order was identical but gene composition differed. A subset of genes was present in either BAg or BSg but absent in the other strain (present only in BAg: bioH, cmk, cysQ, hemD, lig, mltE, yfaE, ydgO, ydiC, and yhiQ; present only in BSg: YidD, YjeK, SohB, topA, mutS, mltA, rpmJ, ygfA, and MutL). There were also a two instances where a single-copy BSg gene (mrsA, and murE) was found in two adjacent copies in BAg; however, one gene copy was dramatically reduced in size and likely reflects remnant sequence of the functional gene.

Phylogenetic reconstruction of BAg

Phylogenetic relationships between Buchnera strains were explored previously 45, 46, but these studies did not incorporate sequence data from BAg. As expected, the topologies derived from the maximum likelihood analysis of 60 concatenated protein-coding Buchnera genes placed the strains into distinct clades corresponding to current aphid tribe classifications (maximum log likelihood of -485470.99; transition/transversion ratio of 1.58) (Fig. 2). Phylogenies from the single-gene datasets were consistent with the concatenated analysis, with the Fordini, Eulachinini, and Macrosphini tribes placed in monophyletic clades. For most genes, the Aphidini tribe formed a distinct clade. There were six instances (rpmI, rpmF, csrA, metE, rplT, and groeS) where BSg or BAg clustered more closely with the Macrosphini clade than to its fellow Aphidini member, which points towards potential issues in reconstructing symbiont phylogenies based on only one or a few genes.

Fig 2

Phylogenetic reconstructions of Buchnera aphidicola A. glycines strain, derived from the maximum likelihood analysis of 60 concatenated protein-coding genes

BAg sequence differentiation between aphid biotypes

Genome size and gene composition was identical for the BAg genomes of A. glycines biotypes 1 and 2. The quality variant detection re-sequencing function in CLC Bio was used to search for nucleotide polymorphisms, insertions, and deletions between the biotypes in the BAg genome. A total of 19 fixed homozygous polymorphisms were identified (Table 4). All of the polymorphisms were SNPs and most were detected on the chromosome (n = 17). The SNPs were spread throughout the chromosome, with 16 of the SNPs located in 15 protein-coding genes. Of interest, nine of the SNPs resulted in putative non-synonymous mutations, which are more likely to affect protein structure and/or function than synonymous changes 55, 56. Moreover, four of the SNPs putatively changed the chemical composition (i.e. polarity and/or pH) of the encoded amino acids. Several heterozygous sites were also identified on the chromosome (n = 14) and plasmids (n = 4), which may be characteristic of the within biotype diversity.

Table 4

The 19 fixed homozygous nucleotide polymorphisms identified in the Buchnera aphidicola A. glycines genome between A. glycines biotypes 1 and 2

Genomic Location	Position (bp)	Biotype 1a	Biotype 2b	Gene	Mutation	AA (B1/B2)c
Chromosome	31,438	A	C	metE	NS	Q/H
Chromosome	45,040	G	A	rplA	S	G
Chromosome	79,764	G	T	fliI	S	G
Chromosome	131,494	G	A	pheS	NS	V/I
Chromosome	176,777	T	G	N/A	N/A	N/A
Chromosome	254,880	C	A	N/A	N/A	N/A
Chromosome	311,966	A	C	bioD	NS	S/Ac
Chromosome	362,217	A	G	pyrC	NS	T/Ac
Chromosome	381,331	A	G	ycfH	S	A
Chromosome	448,664	A	G	ispD	NS	K/Ec
Chromosome	455,674	G	A	cysI	NS	T/S
Chromosome	496, 810	G	A	thiL	S	L
Chromosome	502,106	G	T	yajR	S	G
Chromosome	538,357	A	G	RpsD	S	D
Chromosome	584,826	C	A	N/A	N/A	N/A
Chromosome	598,886	T	C	amiB	NS	I/Tc
Chromosome	612,816	A	C	hemD	NS	I/L
pLeu	4,731	C	T	leuA	S	E
pLeu	7,289	C	A	leuA	NS	E/D
pTrp	----------	----------	----------

aAvirulent biotype.

bVirulent biotype for RAG1 resistance.

cAmino acid changes to different polarities and/or pH levels.

Discussion

Symbiosis with microorganisms occurs often in insects, with members of nearly all groups of sap-sucking insects requiring the metabolic capabilities of intracellular bacteria for normal development and reproduction. This includes the soybean aphid (A. glycines), which is one of the most important pests in soybean growing regions of the Midwest. All aphid species harbor an evolutionarily ancient obligate symbiosis with Buchnera aphidicola, which produces some of the nutrients that are deficient in the aphid's purely phloem diet. The genomes of several Buchnera strains have been sequenced 36-39, and their resulting small size strongly suggests that the bacteria have undergone degenerative evolution from its free living ancestral form. We were able to isolate and assemble de novo the chromosome and plasmid sequences of the Buchnera A. glycines strain (BAg) from whole body aphids, highlighting the effectiveness of next-generation sequencing in sequencing host-associated microorganisms.

The genome size of BAg was ~639 Kb, which is smaller than the fellow Aphidinae A. kondoi, A. pisum, and S. graminum, but considerably larger than members of the Lachninae subfamily (e.g. C. cedri and C. tujafilina). Most of the 603 predicted BAg genes could be assigned a function, and over half are involved in amino acid and protein metabolism, indicative of Buchnera's integral role in the nutrient biosynthetic pathways. The BAg genome contained the most protein coding genes of any Buchnera genome currently available. However, the number of predicted genes is likely a bit overestimated since 1) our annotation scheme did not detect pseudogenes; and 2) the predicted amino acid sequences of several hypothetical genes, absent from other Buchnera strains, were short (<50 aa) and probably not functional. Genome coverage of pTrp was substantially elevated, suggesting a greater copy number of the plasmid relative to the chromosome and pLeu in Buchnera cells. In Buchnera of other Aphidini members, anthranilate synthase is the limiting enzyme of the tryptophan biosynthetic pathway, and its coding genes (trpEG) are localized on pTrp 57-59. The remaining genes in the pathway are located on the chromosome 9. Buchnera is consequently capable of controlling the ratio of trpEG relative to chromosomal genes by modulating the pTrp copy number 57, 60. The BAg plasmid contained two copies of trpEG, which comprises the majority of the sequence. In addition, genes encoding the entire leucine biosynthetic pathway (leuABCD) are localized on pLeu 61, and BAg has one copy of each gene. The copy number of these plasmids differs among aphid species 57, 62-64, presumably due to adaptive modifications related to the varying nutritional requirements of the insect hosts. However, the disparity in the ratios of pTrp and pLeu to chromosome copy number is considerably greater for BAg than for the other Buchnera strains, and may suggest that A. glycines shows greater specificity for endosymbiont-derived essential amino acids.

The BAg genome was substantially diverged from the other sequenced strains, with <85% nucleotide similarity to Buchnera strains of its fellow Aphidini tribe member and the most phylogenetically related strain currently available in the NCBI Bacterial Genomes FTP, Schizaphis graminum (BSg). Gene order was identical between strains but there were differences in chromosomal gene composition. Most notable were genes encoding proteins involved in DNA replication and repair that were present only in BAg (topA, mutS, and mutL) or BSg (lig), indicating significant differences exist at the cellular level. The BAg genome also contained several genes functioning in building block biosynthesis (bioH, cmk, cysQ, and hemD), sulfur metabolism (ydiC and yfaE), and oxidative-reduction (ydgO), which were missing in BSg, suggesting BAg may retain some metabolic pathways that have since lost functionality in BSg. Conversely, BSg retained genes involved in macromolecule degradation (SohB) and posttranscriptional regulation (YjeK) that were no longer found in BAg. Also of interest, the BAg and BSg genomes contained different isoforms of the murein transglycosylase encoding gene (mltA and mltE, respectively). The considerable differences in nucleotide sequence and gene composition between Buchnera strains was expected since they have likely had little to no genetic exchange since the time of divergence of their common Aphidini ancestor 42.

Associations between biotype and host-plant resistance are common for species within the family Aphidiae 65, 66. A key attribute of this group, which may enhance its biotype adaptability relative to other insect pest species, is their obligate mutualism with Buchnera. Endosymbionts can be important mediators of direct and indirect interactions between herbivorous insects and their host plants, and could potentially contribute to biotype evolution or host-race formation of some insects due to their intricate role in nutrient synthesis 67, 68. A 2010 study found constitutive differences in amino acid composition between Rag1 and non-Rag1 soybean leaves, suggesting the resistance gene impacts nutritional quality of the plant in some capacity 30. In turn this may select for Buchnera A. glycines strain genotypes capable of overcoming the deficiencies in amino acid content in Rag1 plants. Proteome analysis revealed considerable differences in Buchnera expression patterns between potato aphid clones (Macrosiphum euphorbiae) with differing levels of virulence reared on resistant and susceptible tomato plants, suggesting the endosymbiont may influence the insect adaptation to host plant resistance 69. Most of the putative BAg nonsynonymous mutations identified between A. glycines biotypes 1 (avirulent) and 2 (virulent) bring about amino acid changes in proteins involved in amino acid activation (pheS) and the biosynthesis of amino acids (metE and leuA), cofactors (bioD, hemD, and ispD), and macromolecules (amiB), all of which could conceivably alter the nutrient profiles afforded by the different biotypes. Moreover, the genes bioD and ispD contain SNPs which change the chemical composition (i.e. polarity and pH) of the encoded amino acids, and could produce major changes to protein structure and stability 70.

Endosymbionts represent an important but often overlooked trophic level in aphid-plant interactions, and may contribute to the insect's differential adaptation to host plant resistance. Several A. glycines resistance genes have been reported and multiple aphid biotypes have been identified based on the ability to overcome host-plant resistance provided by these Rag genes. In this study, we carried out next-generation DNA sequencing to isolate, assemble, and annotate the genome of BAg from whole body aphids, as well as reconstruct the phylogenetic associations among different Buchnera strains. As expected, the BAg genome is small due to reductive evolution, comprised of ~639 Kb on one circular chromosome and two plasmids. While the genome is genetically distinct and differs in gene composition from other Buchnera strains, it forms a distinct clade with the other member of the Aphidini tribe. We detected several fixed, nonsynonymous BAg mutations between avirulent and virulent biotypes for Rag1 resistance, most of which could potentially impact the nutrient profiles afforded to the different biotypes. Future population genomics studies are needed to determine whether the identified SNPs represent fixed differences between biotypes that are consistent across multiple geographical locations. Additional targeted genetic, metabolomics, and proteomic studies will be needed to determine whether the genetic changes influence nutrient synthesis by Buchnera and contribute to aphid biotype evolution to overcome Rag1 resistance. Regardless, our results demonstrate that sequence divergence among Buchnera populations can occur quite rapidly, and could dramatically alter aspects of the aphid adaptability.

Table S1.

Click here for additional data file.