Background & Summary

Some insects can induce abnormal growth and development of host plants and form highly specialized structures termed galls. Galls benefit insect inducers and their offspring by providing abundant nutrition and protection against natural enemies and adverse abiotic factors1. Gall formation at the early stage is generally induced by insect stimuli, including feeding and oviposition1. The molecular mechanisms of gall induction and development have been found to be inseparable from insect effectors2,3 and phytohormones4,5,6. Aphids (Hemiptera: Aphidoidea) are an important group of plant-sapping insects that comprise over 5,000 species, 10–20% of which can induce galls on their primary host plants7,8. Galling aphid species mainly belong to Adelgidae, Phylloxeridae, and several subfamilies of Aphididae, including Eriosomatinae, Hormaphidinae, Tamaliinae, Thelaxinae, and Aphidinae9,10. To date, reference genomes are available for more than 40 aphid species. Among them, only seven true gall-forming species have been sequenced and assembled11,12,13,14,15,16. The lack of genomic resources for galling aphids has greatly hindered our understanding of the genetic basis for adaptive evolution of gall induction in aphids.

The tribe Fordini of subfamily Eriosomatinae is a typical lineage of true gall-inducing aphids. Species from one subtribe, Melaphidina, induce galls on Rhus (Anacardiaceae). The Rhus galls usually contain high concentrations of tannins and have economic and medicinal values17. The genome of one representative species, Schlechtendalia chinensis, has been reported recently15. Aphids within the other subtribe, Fordina, induce galls on Pistacia (Anacardiaceae), which is an economically significant genus of plants. Pistacia vera (cultivated pistachio) produces edible nuts that are of great commercial importance. Seeds of other wild Pistacia species also possess economic value and can be utilized for local consumption, oil extraction, soap production, and most importantly, as rootstock sources for pistachio trees18. Extracts from Pistacia galls exhibit anti-inflammatory and antioxidant activities19,20. However, no genomic data are available for this galling group.

Producing more high-quality genome assemblies encompassing different lineages is foundational to the genomic study of galling aphids. In this study, we generated chromosome-level genome assemblies of two galling species from Fordina, Slavum lentiscoides Mordvilko and Chaetogeoica ovagalla (Zhang). Both species induce closed bag-like galls on the main veins of Pistacia spp. leaves8 (Fig. 1a,b). A total of 332.26 Mb and 289.72 Mb assembled sequences were anchored to 17 and 27 pseudo-chromosomes for S. lentiscoides and C. ovagalla, with 11,747 and 14,492 predicted protein-coding genes, respectively.

Fig. 1
figure 1

Aphid galls and genome features of Slavum lentiscoides and Chaetogeoica ovagalla. Galls and aphids living in galls of S. lentiscoides (a) and C. ovagalla (b). K-mer distribution plots for genomes of S. lentiscoides (c) and C. ovagalla (d). Hi-C interaction heatmaps for genomes of S. lentiscoides (e) and C. ovagalla (f).

Full size image

Methods

Sample collection

S. lentiscoides samples within galls were collected on a pistachio tree (P. vera) from Andijan, Uzbekistan (40.719°N, 72.436°E) in September 2019. The galls of C. ovagalla were obtained on Pistacia chinensis from Qingdao, China (36.193°N, 120.573°E) in July 2018. The aphids from fresh galls were rapidly frozen and stored in liquid nitrogen at the National Animal Collection Resource Center, Institute of Zoology, Chinese Academy of Sciences, Beijing, China.

Genome sequencing

Total genomic DNA was extracted from over 0.2 grams of aphids within a single gall using a standard CTAB method. For Oxford Nanopore sequencing, the concentration and quality of DNA were checked using 1% agarose gel electrophoresis, NanoDrop spectrophotometry (Thermo Fisher Scientific, Lafayette, USA), and Qubit fluorometry (Invitrogen, Darmstadt, Germany). Large-sized segment libraries were selected (≥30 kb) with the BluePippinTM System and processed by the ONT Template Prep Kit (SQK-LSK109, Oxford Nanopore Technologies [ONT], Oxford, UK) protocol. DNA fragments were end-repaired and 3′-adenylated using the NEB Next FFPE DNA Repair Mix kit (New England Biolabs [NEB], Ipswich, USA). The Nanopore sequencing adapters were ligated using the NEBNext Quick Ligation Module (E6056) (NEB). The final library was sequenced on R9 flow cells using a PromethION DNA sequencer (ONT) with an ONT sequencing reagent kit (EXP-FLP001.PRO.6). The raw signal data were called, and the FAST5 files were converted into FASTQ files using MinKNOW software v2.0 (ONT). Short reads (<2kb) and reads with low-quality bases and adapter sequences were removed. Shotgun sequencing reads were used to estimate genome size and to correct the genome assembly. Paired-end libraries were prepared following the instructions for sequencing with the NovaSeq 6000 Reagent Kit on a NovaSeq 6000 platform, with an insert size of 350 bp. For Hi-C sequencing, nuclear DNA from tissue cells was cross-linked and enzymatically digested with DpnII. The DNA fragments with interaction relationships were captured with streptavidin beads and prepared for sequencing. Hi-C libraries with 300–700 bp insert sizes were constructed on the PE150 Illumina platform. Total RNA was extracted from the same colony for RNA extraction and sequencing. RNA-seq libraries were constructed using the NEBNext Ultra RNA Library Prep Kit for Illumina (NEB) and were then sequenced on the NovaSeq 6000 platform with a 150-bp paired-end output.

We finally generated for the first time chromosome-level genome assemblies of S. lentiscoides and C. ovagalla using a combination of Nanopore long reads (44.14 Gb and 55.02 Gb, respectively), Illumina short reads (58.13 Gb and 36.40 Gb, respectively), and Hi-C sequencing data (40.35 Gb and 45.54 Gb, respectively).

Karyotype analysis

We dissected young embryos of aphids on slides. The embryos were kept in 0.7% sodium citrate for 30 min and then fixed in Carnoy’s fixative for 15 min. After fixation, a coverslip was placed on the slide and vertically pressed down to disperse the cell mass as much as possible. The treated slide was frozen at −80 °C for 10 min, followed by air drying. The air-dried slide was stained in 5% Giemsa solution for 15 min and then washed with a tiny stream of distilled water. The slide was observed and subjected to photomicrography using a Leica DM6B photomicroscope (camera magnification 100 × 10). The result showed that S. lentiscoides has a diploid chromosome number of 2n = 34 (Fig. S1), while the karyotype of C. ovagalla was not obtained due to sample limitations.

Genome assembly

Before genome assembly, we extracted Illumina paired sequencing reads with approximately 50× coverage to estimate genome size and heterozygosity. The 17 k-mer frequency spectra obtained with Jellyfish v1.1.1021 and GenomeScope22 suggested that the heterozygosity of the S. lentiscoides and C. ovagalla genomes was 0.99% and 0.70%, the estimated genome sizes were 293.69 Mb and 276.95 Mb, and the repeat sequence content was 41.0% and 40.5%, respectively (Table 1, Fig. 1c,d).

Table 1 Genome assemblies and annotation of Slavum lentiscoides and Chaetogeoica ovagalla.
Full size table

Different assembly strategies were used to assemble the genomes of these two aphid species. For S. lentiscoides, the Nanopore long reads were initially cleaned using Canu v1.523 for sequence error correction and were then assembled from the error-corrected reads with the parameters “useGrid = false, genomeSize = 300 m, minReadLength = 2000, minOverlapLength = 500, corOutCoverage = 135, corMinCoverage = 2”. We used the Redundans24 pipeline to remove bubble contigs from the draft assembly. For C. ovagalla, the long reads were analyzed using Nextdenovo v2.125 for correction with the parameter “-seed_cutoff = 30k”, and then the corrected reads were assembled by SMARTdenovo (https://github.com/ruanjue/smartdenovo) with the parameters “-k 21 -J 5000 -t 20”. We found that the Canu assembly pipeline run obviously more slowly than the latter methods. These two draft aphid genome assemblies were subjected to two rounds of error correction using RACON26 with Nanopore reads, followed by three rounds of additional polishing using Pilon27 with Illumina sequencing data. We further removed a few contigs with microbial contamination by BLAST searches against the NCBI nt database, and no contigs related to the mitochondrial genome were found. After assembly and correction, we produced draft genome assemblies of S. lentiscoides and C. ovagalla, comprising a total of 336.27 Mb and 293.08 Mb of sequences, with 313 and 128 contigs, in which the contig N50 lengths were 6.79 Mb and 5.38 Mb, respectively (Table 1).

The Hi-C paired-end reads were mapped to the Nanopore assembly using BWA aln v0.7.10-r78928. The unique read pairs around the DpnII site were determined. Then, we applied LACHESIS software29 to cluster, order, and orient contigs from the draft assembly. Invalid read pairs were filtered with HiC-Pro30 using default settings. The predicted chromosomal genome was divided into bins of equal length (500 kb), and the number of valid Hi-C read pairs between bins was then used to represent the interaction signals between bins. Finally, we constructed a heatmap based on these interaction signals by R. For C. ovagalla, considering the variability in chromosome numbers observed in Aphididae31, we initially established a range of 10 to 30 for the cluster number. After careful consideration and analysis, we determined that a cluster number of 27 was the most appropriate choice. As a result, 40.35 Gb and 45.54 Gb of Hi-C clean reads were generated to build chromosome-level assemblies. After clustering, ordering, and orientation of the contigs, we obtained two final assemblies with chromosome-anchored sizes of 332.26 Mb and 289.72 Mb and scaffold N50 lengths of 19.77 Mb and 11.85 Mb for S. lentiscoides and C. ovagalla, respectively (Table 1). Therein, 128 and 86 contigs could be anchored to 17 and 27 linkage groups, with 98.81% and 98.85% anchoring rates, respectively (Table 1, Fig. 1e,f). The chromosome circus plots were generated using Circos v0.69-932 (Fig. 2).

Fig. 2
figure 2

Circus plots describing genomic characteristics of Slavum lentiscoides (a) and Chaetogeoica ovagalla (b). From outer to inner circles: chromosome length, gene density, transposable element density, and GC density.

Full size image

Genome annotation

Before gene prediction, a de novo repeat library was built by using LTR_FINDER33 and RepeatScout34, and the library was classified by PASTEClassifier35. Transposable element sequences in the genome were predicted with RepeatMasker36 using the de novo library and Repbase library37. Consequently, we identified 87.71 Mb and 61.25 Mb of repeat sequences, accounting for 25.94% and 20.90% of the genome assemblies of S. lentiscoides and C. ovagalla, respectively (Table 1).

The protein-coding genes were predicted by integrating the evidence generated via the de novo, homology-based, and RNA-seq-based methods. For ab initio prediction, we used Augustus v2.438 to generate the de novo predictions with the default parameters. For the homology-based analysis, GeMoMa v1.3.139 was used to query the protein sequences against a database of four species (Myzus persicae, Aphis gossypii, Rhopalosiphum maidis, and Acyrthosiphon pisum). For RNA-seq annotation, RNA-seq reads were aligned to the genome assembly using HISAT2 v2.0.440 and StringTie v1.2.341 with the default parameters. TransDecoder v2.042, GeneMarkS-T v5.143, and PASA v2.0.244 were used to generate transcript predictions. Finally, we integrated three types of evidence for gene prediction by EVidenceModeler (EVM) v 1.1.145. By integrating ab initio-based, homology-based, and transcriptome-based evidence, we predicted 11,747 genes in the S. lentiscoides genome, and 14,492 genes in the genome of C. ovagalla. The predicted genes were then searched for homology-based functions by BLAST searches against the NCBI non-redundant protein sequences (nr), Eukaryotic Orthologous Groups (KOG), TrEMBL, Kyoto Encyclopedia of Genes and Genomes (KEGG), and the Gene Ontology (GO) databases with BLASTP v2.2.3146 (-evalue 1e-5). The GO terms were assigned using blast2go v5.047. A total of 99.22% (11,656 genes) and 89.73% (13,004 genes) of the predicted genes of these two genomes were supported by functional annotation from the above five protein databases. Among these genes of S. lentiscoides and C. ovagalla, 98.11% and 89.15% showed homology to proteins in NCBI nr, 98.89% and 67.84% in TrEMBL, 67.34% and 57.00% in KOG, 48.76% and 43.27% in KEGG, and 38.11% and 43.17% in GO, respectively (Table 2).

Table 2 Number of functionally annotated genes of Slavum lentiscoides and Chaetogeoica ovagalla genomes.
Full size table

Three types of noncoding RNAs (ncRNAs) were identified. The microRNA (miRNA) and ribosomal RNA (rRNA) genes were predicted by BLASTN searches against the Rfam database48. The transfer RNA (tRNA) genes were identified using tRNAscan-SE49. Finally, we identified 27 and 31 miRNAs, 66 and 68 rRNAs, and 200 and 172 tRNAs within the genomes of S. lentiscoides and C. ovagalla, respectively (Table 1).

Data Records

The genome raw data, RNA sequencing data, and genome assemblies are available at the National Center for Biotechnology Information (NCBI) under the BioProject accession numbers PRJNA765394 and PRJNA832539. The Illumina WGS data was archived with the accession numbers SRR1604696350 and SRR2399932551. The Nanopore WGS data was deposited with the accession numbers SRR1604696452 and SRR2399932653. The RNA-Seq data was archived with the accession numbers SRR1604696154 and SRR2399932355. The Hi-C data was archived under the accession numbers SRR1604696256 and SRR2399932457. The genome assemblies have been deposited at GenBank under the accession numbers GCA_032441835.158 and GCA_032441825.159. The genome annotation files have been deposited at the Figshare60.

Technical Validation

The quality and completeness of draft assemblies were assessed by three methods as follows. The Illumina reads were mapped to the draft assembly with BWA-MEM v0.7.10-r78961, and the mapping ratio was calculated with SAMTOOLS v1.362. In addition, the Core Eukaryotic Genes Mapping Approach (CEGMA v2.563) and BUSCO v3.1.064 were used to assess the completeness of the genome assembly.

The S. lentiscoides and C. ovagalla genomes contained 96.77% and 97.58% of highly conserved Core Eukaryotic Genes (CEGs), respectively. They showed a high representation of conserved complete Benchmarking Universal Single-Copy Orthologs (BUSCOs) (S. lentiscoides: 93.70% complete BUSCOs of insecta_odb10; C. ovagalla: 97.00% complete BUSCOs of insecta_odb10), mapped with 96.78% and 97.19% of Illumina short reads, respectively, which indicated that these genome assemblies were of high quality and near-complete.