Background & Summary

Grass pea (Lathyrus sativus) is a cool-season legume crop cultivated for food mainly in the Indian subcontinent and Ethiopia and as a feed and fodder crop in other parts of the world. Lathyrus has various beneficial agronomic traits such as tolerance to drought, salinity, waterlogging, resistance to insects and biotic stresses, and growing well in semiarid and problem soils1,2,3. Furthermore, as a legume crop, it can fix nitrogen. These attributes make it an ideal crop for popularization to sustain agricultural productivity in the changing climatic conditions. Nutritionally, this pulse crop is very rich in proteins, second only to soybean, and provides a balanced amino acid diet in combination with cereals to poor people in countries where it is consumed. It is also a source of L-homoarginine with the potential to increase cardiac health4. Moreover, the genus Lathyrus belongs to the Vicieae tribe, of which important legume crops, Pisum, Lens, and Vicia, are other members. Therefore, research on Lathyrus and utilization of its genes underlying valuable agronomic traits like drought resistance, salt resistance, and biotic stress resistance in these closely related genera would be of considerable interest.

Despite many beneficial agronomic and nutritional traits, the major impediment in popularizing this crop is its association with neurolathyrism, characterized by irreversible lower limb paralysis in the affected individuals. Excessive Lathyrus consumption for prolonged periods lead to neurolathyrism, which has happened in famine-like situations when the grass pea seeds were consumed as a staple diet. Therefore, one primary research goal in this plant is to understand the mechanisms of neurotoxin β-N-oxalyl-L-α, β-diaminopropionic acid (β-ODAP) accumulation and thereby reduce neurotoxin content in the seeds.

Harnessing the vast diversity of germplasm and the gene pool of Lathyrus very much depends on the availability of high-quality genome sequence information. Pusa-24, a popular Indian cultivar, has a β-ODAP content of 0.3–0.6% in the seeds and is the parental plant line used in breeding programs of many low neurotoxin cultivars developed so far3. Here we report a high-quality reference assembly of Lathyrus sativus cv. Pusa-24.

Methods

Genome size estimation

Fresh leaf tissue (~100 mg) from 12–13 days old plants of Lathyrus, wheat, and pea (Pisum sativum) were taken in a pre-chilled Petri plate kept on ice. Thereafter, 1.5 ml ice-cold Galbraith’s buffer5 (45 mM MgCl2, 20 mM 3-(N-morpholino) propane sulfonic acid (MOPS), 30 mM sodium citrate, 0.1% (v/v) Triton X-100 and pH 7.0) was added to the plate and chopped the leaves using a new razor blade into very fine slices. Chopping of leaves was performed in four different combinations: Lathyrus leaves only, Lathyrus + wheat + pea leaves, Lathyrus + pea leaves, and Lathyrus + wheat leaves. Pea and wheat were used as standard reference samples with known genome sizes. The homogenate was mixed by up and down pipetting without trapping any air bubbles and was filtered through a 40 µm nylon filter. 0.5 ml filtrate was taken into a fresh tube, and 2.5 µl RNase was added and incubated on ice for 15 minutes. To stain the nuclei, propidium iodide (PI) was then added to a final concentration of 50 µg/ml, and samples were kept in the dark for 30 minutes on ice with occasional mixing. Flow cytometry was performed in a BD FACSAria Fusion flow cytometer (BD Biosciences). The genome size of Lathyrus was estimated using the known C value parameters of Pea (2 C = 9.09 pg) or wheat (2 C = 34.6 pg) as reference using the formula -

Sample 2C DNA content = [(sample G1 peak mean)/(Reference G1 peak mean)] x Reference 2C DNA content (pg DNA).

Sample collection, library construction and sequencing

Genomic DNA was extracted from leaves of L. sativus cv. Pusa-24 grown at 22 °C, 200 μmol m−2 s−1 light intensity, 16 /8 hours’ photoperiod and 60% relative humidity using the Qiagen Plant DNA kit as per the manufacturer’s description. The quality and integrity of the extracted DNA were evaluated based on its A260/A280 ratio and its electrophoretic run on an agarose gel. A total of three paired-end (300 bp, 500 bp, and 800 bp insert size), and 3 mate-pair (2–5 Kb, 5–8 Kb, and 8–10 Kb insert size) libraries were generated. The paired-end and mate-pair libraries were generated using the Illumina TruSeq DNA Nano Preparation Kit (Illumina, San Diego, CA, USA), and Nextera Mate Pair Library Preparation Kit (Illumina, San Diego, CA, USA) respectively. All libraries were sequenced on an Illumina HiSeq. 2500 platform following the manufacturer’s instructions. Additionally, for long-read sequencing, libraries were developed using the SMRTbell template preparation kit following the manufacturer’s instructions and sequenced on the PacBio Sequel (I) platform. Finally, ~625 Gb of short-read sequencing raw data and ~85 Gb of long-read sequencing raw data were generated (Tables 1, 2).

Table 1 Summary statistics of Lathyrus genome raw short-reads.
Full size table
Table 2 Summary statistics of Lathyrus genome PacBio reads.
Full size table

Preprocessing and genome assembly

The raw fastq files were pre-processed before performing assembly. We trimmed the adapters sequences and filtered out reads with an average quality score of less than 30 in any paired-end reads using Trimmomatic v0.366. De novo hybrid assembly was generated using MaSuRCA assembler v4.0.37,8. The cleaned paired-end reads, mate-pair reads, and PacBio long reads were configured as the input data for the hybrid assembly. The assembly was carried out using the default parameters in MaSuRCA. The contig-level assembly covered 3.8 Gb of the genome with a contig N50 value of 78.27 kb (Table 3). Further, the contig-level assembly was scaffolded with Pisum sativum as a reference9 using the reference-guided scaffolder RaGOO10. The scaffolded assembly contained seven chromosome-sized scaffolds and 25404 contigs. The N50 value of the scaffolded assembly was 421.39 Mb (Table 3).

Table 3 Summary statistics of the Lathyrus genome assembly.
Full size table

Repeat annotation

Repetitive regions of the Lathyrus genome were identified using RepeatModeler v1.01.11. A de novo repeat library was constructed using RepeatModeler. A combination of the Repbase1611 library and the de novo library was then used with RepeatMasker12 v4.0.715 to identify repeats in the Lathyrus genome. Overall, we identified 3.17 Gb of repetitive sequences, representing 83.31% of the Lathyrus genome assembly (Table 4, Fig. 1); of which the long terminal repeat (LTR) elements were the most abundant, accounting for 37.58% of the whole genome.

Table 4 Repeat summary statistics of the Lathyrus genome assembly.
Full size table
Fig. 1
figure 1

Genome features of Lathyrus genome assembly. The circos plot shows, from outside to inside, ideograms of the seven chromosome-sized scaffolds, gene density (blue-green scale), density of DNA transposons, density of LTR retrotransposons, density of simple repeats, Gene expression levels of 4-day and 7-day old seedlings, and position of SSPs on the scaffolds (Red square – Albumins, Blue circle – Legumins, Green square – Lathyrins, Black triangle – Convicillin, and Violet rhombus -Glutelin).

Full size image

Gene prediction and annotation

Ab initio and homology-based methods along with RNA-seq evidence were combined to predict protein-coding genes using the BRAKER2 v2.1.513 pipeline. For homology-based prediction, protein sequences of seven other legume species (Cajanus cajan, Cicer arietinum, Glycine max, Medicago sativa, Pisum sativum, Phaseolus vulgaris, and Vigna unguiculata) were downloaded from the Legume federation database (https://www.legumefederation.org/). The RNA-Seq data for Lathyrus was derived from a previous study14. A total of 50,106 protein-coding genes were predicted, out of which 45,632 were located on the chromosome-sized scaffolds (Fig. 1). The predicted genes were then annotated for their putative biological function by searching against the Uniprot and NCBI nr database. Approximately 96.21% of these genes were functionally annotated by at least one of the databases.

Data Records

The DNA sequencing data were submitted to the NCBI Sequence Read Archive (SRA) database under the SRA IDs: SRR1973230415, SRR1828632816, SRR1828632617, SRR1828632518, SRR1828632919, SRR1828632720, and SRR1828632421, which is associated with the BioProject accession number PRJNA813354. The genome assembly is available at NCBI22, and the protein sequences are publicly available at zenodo23. Additionally, we have constructed a web portal for the Lathyrus genome project (https://lathyrusgenome.nabi.res.in) to benefit the scientific community working on Lathyrus and other legume crops. The web portal offers multiple functionalities, including a BLAST search option (against the genome assembly, mRNA, CDS, and protein sequences), an ortholog search-retrieve option, Jbrowse2-based genome visualisation option, and download links to the genome assembly, mRNA, and protein sequences.

Technical Validation

Quality assessment of the genome assembly

The genome size of Lathyrus was estimated as 6.62 Gb using Pea (Pisum sativum) (4.3 Gb) as reference by flow cytometric analysis (Fig. 2, Table 5). Similar values were obtained when wheat was used as a reference (6.79 Gb, Table 5). The assembly presented here is the first Lathyrus genome to be available in the public domain. The contig N50 and scaffold N50 sizes were 78.27 Kb and 421.39 Mb, respectively, with the longest scaffold size 755.27 Mb. A preprint publication describing the draft genome of Lathyrus is available; however, the raw and assembled data is not available publicly. We compared the overall assembly statistics of our assembly with that of a draft genome of Lathyrus available in the preprint24. The draft genome assembly covered 6.2 Gb of the genome; however, it was highly fragmented and had a BUSCO v425 completeness score of 88.4% (Viridiplantae). We carried out BUSCO analysis of both the contig-level assembly and scaffolded assembly to assess the completeness of our assembly and to ascertain if the assembly covered the majority of the gene space. Both the contig-level assembly and scaffolded assembly had a BUSCO completeness score of 98.35% (Viridiplantae) (Table 6), which was higher than the BUSCO scores of the draft genome reported earlier. Additionally, we also subjected both the assemblies to BUSCO analysis with other databases like Eudicots and Fabales, which yielded completeness scores of ~97% and ~96%, respectively (Table 6). Therefore, the gene space coverage in our assembly is adequate and is suitable for various genic analyses involving protein content, β- ODAP metabolism, and drought hardiness.

Fig. 2
figure 2

Flow cytometry-based estimation of Lathyrus genome size. Propidium iodide fluorescence amplitude (in arbitrary units) is shown against event density (y-axis) which correspond to the G0/G1 DNA for each species. The pea, Lathyrus, and wheat nuclei events are denoted by red, blue, and green colors respectively.

Full size image
Table 5 Genome size estimation by flow cytometry.
Full size table
Table 6 Summary of BUSCO analysis of Lathyrus genome assembly against Viridiplantae, Eudicots, and Fabales databases.
Full size table

Gene prediction and annotation validation

Gene models in the Lathyrus assembly were predicted using the BRAKER2 pipeline, which used a combination of ab-initio gene prediction, homology-based, and RNASeq evidences. To enhance the quality of the gene prediction, we removed low-quality genes of short length (proteins with fewer than 30 amino acids) and/or exhibiting premature termination. The final gene set consisted of 50,106 genes, which was similar to the other legume species sequenced to date. Also, functional annotation of the predicted gene models indicated that 96.21% of them could be assigned to at least one functional term. Additionally, we also carried out orthology analysis of the Lathyrus gene models with the other legume species to validate the predicted genes in the Lathyrus assembly. In the orthology analysis, 49,331 genes (94.5%) of Lathyrus could be assigned to an orthogroup. A total of 13191 orthogroups contained genes from all the nine legume species, while 2100 orthogroups containing 13840 genes were specific to Lathyrus (Fig. 3). Furthermore, 488 single-copy orthogroups identified in the analysis were used to reconstruct a high-confidence phylogenetic tree, which was in concordance with previous studies that determined the phylogenetic relationship among the legumes (Fig. 3).

Fig. 3
figure 3

Ortholog analysis of nine legume species including Lathyrus. (a) A total of 401284 genes from nine species were grouped into 36014 orthogroups. The UpSet plot shows the overlap between orthogroups from each species and the size of overlap as bar charts. (b) A maximum likelihood tree representing the phylogenetic relationship between the nine legume species.

Full size image

Further, to confirm the validity and quality of gene prediction of the Lathyrus assembly, we searched for genes that contribute to β-ODAP biosynthesis in the Lathyrus genome. Since β-ODAP, an endogenous non-protein amino acid, is present exclusively in Lathyrus and not in the other legume species, identification of the biosynthetic genes of this non-protein amino acid in the current genome assembly will further affirm the quality of the gene prediction. β-ODAP biosynthesis is believed to occur in the mitochondria and chloroplasts and originate from precursors - asparagine and serine26. We identified most of the known genes associated with this pathway, viz. Serine O-acetyltransferase (SAT), Cysteine synthase (CS), cyanoalanine synthase (CAS), nitrilase, β-ODAP synthetase (BOS), oxalyl-CoA synthetase (OCS), and oxalate decarboxylase (ODC). The Lathyrus genome has five copies of the Serine O-acetyltransferase (SAT) gene; however, only two are expressed during the 4- and 7-day old seedlings (Fig. 4). Cysteine synthase (CS) is encoded by a multigene family in plants that includes cyanoalanine synthase (CAS) and other related enzymes. The Lathyrus genome encodes eight CS genes, out of which one may be a CAS. Previous studies reported only five CS isoforms, including a CAS26. Therefore, these results indicate that the predicted gene set of the Lathyrus genome is complete and of high quality and can be used for various gene discovery studies.

Fig. 4
figure 4

Biosynthesis of β-ODAP in Lathyrus sativus. The biosynthetic pathway of β-ODAP along with the genes/enzymes catalysing each step in the reaction is shown. The colored circles denote the expression levels of the corresponding gene in 4-day and 7-day old Lathyrus seedlings.

Full size image