Chromosome-Level Genome Assembly of Solanum carolinense

1. Introduction

Solanum carolinense L. is a perennial herbaceous species in the Solanaceae family, native to the southeastern United States and widely recognized as a noxious weed [1]. It was first detected in Taizhou and Wenzhou, Zhejiang Province, China, in 2006 [2]. Currently, S. carolinense has spread throughout the United States, Canada, and at least 36 countries and regions across Oceania, Europe, and Asia [3,4]. As a troublesome weed in major food crops such as corn, potato, and peanut, as well as in vegetables and forage crops, S. carolinense also serves as a host for key pests, including the Colorado potato beetle and carmine spider mite, and for several Solanaceae-infecting viruses, thereby increasing pest and disease pressure in cropping systems [5]. In 2017, it was included in China’s Catalog of Imported Plant Quarantine Pests [6].

Beyond its agricultural impact, S. carolinense is also an important model for studying the plant-herbivore interactions [7]. It exhibits distinctive genetic traits, including a plastic self-incompatibility system [8], diverse stress resistance and defense mechanisms, and strong asexual reproductive capacity, all of which contribute to complex coevolutionary relationships with multiple herbivorous insects. Its defense strategies include both constitutive and inducible mechanisms: physical defenses such as non-glandular trichomes and prickles [9], and chemical defenses including polyphenols, glycoalkaloids, and trypsin inhibitors [10,11]. Its plastic self-incompatibility system [12], andromonoecious floral sex expression [13], and tolerance to inbreeding depression may influence the evolutionary trajectory of its mating system [12], providing flexibility in reproductive strategies.

S. carolinense shows remarkable ecological adaptability, thriving in diverse soil types including sandy and gravelly substrates, and reproducing through seeds, roots, and even small root fragments. Its extensive rhizome network can extend several meters and regenerate from minute segments [6,14], making management extremely difficult. Although it poses a significant threat to global food security, current control tactics remain inadequate, mostly relying on herbicide applications that could have negative side effects on crop seed set such as picloram alone in autumn to suppress root growth or combined with glyphosate to control fruiting plants [14]. Although prevention is regarded as critical and effective [6], manual removal before flowering and excavation of underground rhizomes are labor-intensive and insufficient measures due to the species’ strong vegetative regeneration. With many regions of China now suitable for its establishment [2,15], thorough studies on its genetic differentiation, allelopathy, and dispersal capacity are urgently needed.

Genome sequencing has become a core tool for elucidating plant genetic architecture, uncovering molecular mechanisms underlying key traits, and supporting crop improvement and sustainable agriculture [16]. In the genus Solanum, chloroplast genomes have been particularly informative. The chloroplast genome of Solanum elaeagnifolium Cav. is 155,049 bp in length and shows close affinity with species in section Melongena [17]. Species-specific SSR markers developed for S. elaeagnifolium provide valuable tools for assessing genetic diversity and tracing population origins in this species and its relatives [18]. Similarly, the chloroplast genome of Solanum dulcamara (155,580 bp) [19] exhibits a typical angiosperm quadripartite structure and low genetic polymorphism, with only weak geographical and ecological structuring among populations [20]. The chloroplast genome of S. carolinense, at 155,315 bp with a GC content of 37.6%; is highly conserved relative to congeners, particularly in coding regions [21]. However, no genetic characteristics associated with its invasiveness have yet been identified in the chloroplast genome.

While chloroplast genomes offer useful insights, whole-genome sequencing provides the complete nuclear sequence information required to investigate species evolution, gene function, regulatory networks and adaptive mechnisms [22,23].

Among Solanum weed species to date, only Solanum rostratum and Solanum torvum have undergone high-quality chromosome-level genome assembly. The final assembled genome of S. rostratum spans 869.69 Mb, with a BUSCO completeness score of 99.5% for pseudochromosomes [24]. The assembled genome of S. torvum is approximately 1.25 Gb in size, achieving a BUSCO completeness of 98% [25]. To date, only the chloroplast genome of S. carolinense has been assembled and characterized [21] with no nuclear genome assembly available. A high-quality chromosome-level nuclear genome would enable systematic exploration of the genetic basis of invasive adaptability, reproductive biology, herbicide target sites, non-target herbicide-resistance loci, and structural variants that may facilitate rapid evolution. Such resources are essential for deciphering the molecular mechanisms driving their invasiveness and for developing evidence-based management strategies.

This study employed a multi-platform sequencing approach to generate a high-quality chromosome-level reference genome for S. carolinense. We integrated Pacific Biosciences (PacBio) High-Fidelity (HiFi) long-read sequencing (renowned for long read lengths and high base accuracy), Illumina next-generation short-read sequencing (offering high throughput and cost effectiveness) and High-throughput/resolution chromosome conformation capture (Hi-C) (which facilitates chromosome-scale scaffolding and integration of 3D genomic architecture with multi-omics data) [26]. The resulting assembly was comprehensively annotated for protein-coding genes, non-coding RNAs, and repetitive elements. We also identified candidate genes associated with defense responses and reproduction. These genomic resources will establish a robust molecular foundation for future research on adaptive evolution, defense regulation, reproductive system evolution, and invasive dispersal mechanisms in S. carolinense.

2. Results

2.1. Sequencing Data Processing and Genome Characterization

Sequencing data for the S. carolinense genome assembly consisted of next-generation sequencing (NGS) short reads, and third generation (HiFi) sequencing. After quality control and filtering, a total of 396.96 Giga bases (Gb) of clean data were obtained (Table 1). K-mer analysis based on the NGS data estimated the genome size at approximately 842 Mb, with a heterozygosity rate of 1.13% and a repeat content of 49.64%. These results indicate that the S. carolinense genome is highly repetitive and with low heterozygosity. The K-mer distribution is shown in Figure 1.

Note: The abscissa represents the depth at kmer=19, and the ordinate represents the k-mer frequency at different depths multiplied by the corresponding depth.

2.2. Preliminary Genome Assembly Using HiFi Sequencing Data

A total of 99.89 Gb of high-quality HiFi reads were generated, with an average read length of 18,660.8 bp and an N50 of 18,384 bp. Genome assembly using hifiasm produced a draft genome of 915.39 Mb, consisting of 128 contigs with a contig N50 of 51.06 Mb (Table 2). To assess assembly accuracy, the NGS short reads were remapped, resulting in an alignment rate of 99.90% and a coverage of 98.45%, indicating excellent consistency between the assembled genome and the raw sequencing data.

2.3. Hi-C-Assisted Chromosome-Level Scaffolding and Quality Evaluation

Following Hi-C scaffolding and manual refinement, 879.26 Mb of contig sequences (96.05% of the assembled genome) were anchored to 12 pseudochromosomes. The final chromosome-level genome assembly of S. carolinense is 915.40 Mb in length, with a GC content of 35.61% and a Scaffold N50 of 73.17 Mb (Table 3), indicating high continuity and completeness of the assembly that meets the requirements for subsequent gene annotation and functional analyses. The chromosome-level assembly of the S. carolinense genome is visualized in a normalized Hi-C contact map (chromosome interval interaction plot) in Figure 2. The contact heatmap shows strong intra-chromosomal interaction signals along the diagonal and weak inter-chromosomal interactions, confirming the correct anchoring of scaffolds to individual chromosomes with minimal misjoins or chimeric assemblies.

Note: The color scale from light to dark red represents the log-transformed Hi-C interaction frequency, reflecting low to high interaction intensity. Twelve pseudochromosomes are designated as chr1 to chr12.

Genome completeness was evaluated with Benchmarking Universal Single-Copy Orthologs (BUSCOs) (v5.4.7) [27] in genome mode against the embryophyta_odb10 database, yielding a completeness value of 94.7% (Table 4), indicating excellent completeness of the assembly. A Circos plot integrating chromosome length, gene density, repeat sequence density, GC content, and sequence collinearity was constructed to provide an overview of the genomic characteristics (Figure 3).

Note: From outer to inner circles: gene density, repeat sequence density, GC content.

2.4. Genome Annotation and Quality Evaluation

2.4.1. Structural Annotation

A total of 674.12 Mb of repetitive sequences were identified in the S. carolinense genome, accounting for 73.64% of the assembly (Table 5).

Repeat sequence annotation was performed using a combined approach of “public database alignment + ab initio annotation”, with various types of non-coding RNAs identified in the genome, and their statistics are summarized in Table 6.

A total of 32,206 protein-coding genes were annotated and predicted in the S. carolinense genome, with an average length of 5,209 bp with a coding sequence (CDS) length of 1,149 bp, 4.78 exons per gene, an average exon length of 312 bp, and an average intron length of 982 bp (Table 7).

2.4.2. Functional Annotation

Of the predicted protein-coding genes, 97.95% were functionally annotated (Table 8). GO enrichment analysis (Figure 4) revealed that the annotated genes are involved in a wide range of biological processes, including phosphorylation, regulation of DNA transcription, and defense response. These genes are primarily localized in the cell membrane, nucleus, and cytoplasm, and exhibit molecular functions such as ATP binding, metal ion binding, and DNA binding. KEGG pathway analysis (Figure 5) revealed that the genes are predominantly associated with global metabolism, signal transduction, and environmental adaptation pathways. KOG classification (Figure 6) indicated a high proportion of genes involved in general function prediction, post-translational modification, protein turnover, molecular chaperones, and amino acid and carbohydrate metabolism. The abundance of post-translational modification and molecular chaperone genes highlights the importance of protein homeostasis in S. carolinense. BLAST searches against the NR database (Figure 7) showed that the top three species with the highest sequence similarity to S. carolinense genes were Solanum dulcamara (37.11%), Solanum tuberosum (20.12%), and Solanum verrucosum (10.01%). Among the top 10 hits, all species except Capsicum annuum belonged to the genus Solanum, confirming the close phylogenetic relationship of S. carolinense within the Solanaceae family.

3. Discussion

We annotated 674.12 Mb of repetitive sequences, accounting for 73.64% of the assembled S. carolinense genome (Table 5). This proportion is significantly higher than the 49.64% estimated by NGS-based K-mer profiling. This discrepancy stems from inherent limitations of the K-mer approach, which primarily detects high-copy and highly diverged repeats while systematically underestimating short repeats, low-complexity regions, and partially homologous sequences. Furthermore, heterozygous peak interference introduces a conservative bias that further depresses K-mer-derived estimates. In contrast, our annotation pipeline integrating de novo prediction and homology alignment effectively overcomes these limitations by capturing diverse repeat families, including LTR retrotransposons, other transposons, tandem repeats, and dispersed repeats. This dual strategy provides broader coverage and a more comprehensive characterization, ultimately yielding a more accurate estimate of the genomic repetitive content.

The S. carolinense genome is highly repetitive, predominantly long terminal repeat (LTR) retrotransposons. Such high repetitive sequences content poses significant challenges for chromosome anchoring, often resulting in contig misassemblies and reduced chromosome mounting rates. In this study, these challenges were effectively overcome through an integrated sequencing strategy combining PacBio HiFi long-reads and Hi-C chromatin conformation capture. The high accuracy and length of HiFi reads enabled reliable spanning of repetitive regions, while Hi-C long-range interaction signals facilitated accurate chromosome-scale scaffoliding. Additionally, precise annotation of repetitive elements pior to assembly further minimized their interference during genome scaffolding.

The estimated genome size of S. carolinense is approximately 915 Mb, which is relatively compact for a Solanum species. This compactness likely confers advantages such as shorter DNA replication cycles and faster cell division rates, reducing energetic costs associated with replication and transcription. Consequently, more resources can be allocated to key invasive traits, including enhanced stress tolerance, seed dormancy, and allelopathy, thereby increasing the species’ competitive ability and invasiveness.

Gene annotation revealed a substantial proportion of functionally uncharacterized genes based on KOG classification. These represent promising candidates for future functional genomics studies. NR-based functional annotation, combined with evolutionary analysis, provides a foundation for investigating adaptive divergence, secondary metabolism, and ecological interactions between S. carolinense and its closely related species.

Effective chemical control of S. carolinense remains challenging despite the evaluation of several herbicides, such as triclopyr and glyphosate for long-term suppression of root sprouting [28]; 2,4-DB for reducing seed set when applied pre-flowering [4], and picloram for limiting root biomass [14]. The species’ deep-rooted architecture and extensive horizontal rhizomes (often exceeding one meter from the mother plant) severely limit chemical control [29]. Moreover, the ability of rhizomes to overwinter and produce new shoots in spring complicates management [30]. By generating a high-quality chromosome-level genome assembly, tis study provides a critical resource for elucidating genomic structural characteristics, gene family composition, and evolutionary patterns underlying these traits. Future research can now focus on identifying key genes regulating rhizome growth and development, root initiation, elongation, stress acclimation, and colonization. Such insighhts will clarify the molecular mechanisms governing seed germination, root growth, plant establishment, and invasive spread, paving te wav for targeted genetic interfventions, novel control technologies, and precise management strategies. Building on this reference genome, priority should be given to developing species-specific molecular markers. These tools would enable rapid, accurate detection of seeds and root fragments at ports of entry, grain processing facilities, and nurseries, without requiring taxonomic expertise or intact plant material.

Despite these advances, current genomic understanding of S. carolinense is constrained by limited sampling. Future studies should address this by (1) expanding phylogeographic sampling across native and invasive ranges through whole-genome resequencing to reconstruct invasion routes and global population structure; (2) integrating transcriptomic and metabolomic datasets to dissect regulatory networks underlying invasiveness and stress resistance; and (3) translating genomic findings into practical applications, such as standardized quarantine detection kits based on population-specific markers integrated with field monitoring. Uitimately, a multidisciplinary framework combining molecular biology, ecology, and quarantine science will be essential for developing a comprehensive, effective management system for this globally invasive weed.

4. Materials and Methods

4.1. Sample Preparation and Sequencing

Fruits of S. carolinense were intercepted from soybean imports originating from the United States. Seeds were germinated in Petri dishes, then transplanted into nutrient soil, and grown in a controlled plant growth chamber for 50 days under a 12 h (35 °C) / 12 h dark 25 °C) cycle. Collect root, stem, and leaf tissue from a single individual for nucleic acid extraction. The extracted DNA was used for PacBio HiFi, Illumina next-generation, and Hi-C sequencing.

Sequencing libraries for PacBio HiFi, Illumina, and Hi-C platforms were constructed following standard protocols, and raw sequencing data were generated. Quality control and standard filtering were applied to all data sets to remove low-quality and contaminant sequences. Illumina reads were processed using fastp (v0.21.0) [31] and assessed with FastQC (v0.11.9) [32]; reads containing >5% N bases, >50% low-quality bases, adapter contamination, or PCR duplicates were discarded [33] to obtain clean reads. PacBio HiFi data were filtered by removing polymerase reads shorter than 50 bp, reads with quality scores <0.8, and sequences with adapter contamination or self-ligation (SMRTlink, v13.0 https://www.pacb.com/smrt-link/). The remaining subreads were processed into HiFi reads using SMRTLink (v13.0). Hi-C data were first filtered with fastp to remove adapters and low-quality reads, then aligned to the reference genome using HiCUP software. Non-uniquely mapped read pairs, invalid reads, and PCR duplicates were removed to obtain high-quality Hi-C data.

4.2. Genome Characterization and Assembly

Genome characteristics were estimated using K-mer analysis. Filtered reads were decomposed into k-mers and their frequency-depth distributions were obtained using Jellyfish (v2.2.1) [34]. Genome size, heterozygosity, and repeat content were estimated with GenomeScope (v2.0) [35]. To assess potential contamination, the first 50,000 sequencing reads were aligned against the NT database using BLASTn (https://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.11.0/, v2.11.0), and taxonomic classification was performed with the taxdump tool to identify possible exogenous sequences.

Geneme assembly was performed using high-quality HiFi reads with hifiasm software (v0.19.8) [36] generating contig sequences and constructing a draft genome. Hi-C data were subsequently used for chromosome-level assembly. Clean Hi-C reads were aligned to draft assembly, and uniquely paired-end aligned reads were screened to evaluate valid and invalid data volumes. Contigs were clustered into chromosome groups using ALLHIC software [37], followed by ordering and orientation within each group. Subsequently, files were converted using the 3D-DNA [38] software and Juicer (v1.6) [39], followed by manual ordering and orientation via Juicebox (v1.11.08) [40]. Redundant contigs were removed, and assembly quality was evaluated in terms of continuity, consistency, and completeness.

4.3. Genome Annotation and Visualization

4.3.1. Repeat Sequences and Non-Coding RNA Annotation

A de novo repeat sequence library was constructed using RepeatModeler (v2.0.6) [41], LTR_FINDER [42] and LTRharvest (v1.6.5) [43], followed by redundancy removal with LTR_retriever (v3.0.1) [44] and library establishment via teclass (v2.1.3) [45]. The resulting library was merged with RepBase, and repeat annotation was performed using RepeatMasker (v4.1.7) [46] and associated tools, to generate a final combined transposable element (TE) dataset. Transfer RNA (tRNA) genes were identified using tRNAscan-SE (v2.0.12) [47], while ribosomal RNA (rRNA) genes were predicted with RNAmmer (v1.2) [48]. Other non-coding RNAs (ncRNAs) were identified using INFERNAL (v1.1.4) [49] based on the Rfam database [50].

4.3.2. Protein-Coding Gene Annotation

Gene structure was predicted using a combination of transcriptome-based, homology-based, and ab initio methods. Next-generation and third-generation transcriptomes were filtered, aligned, and assembled to reconstruct transcripts. Protein sequences from closely related species were used for homology-based prediction with miniprot (v0.13) [51], and ab initio prediction was conducted using Augustus (v3.5.0) [52], Genscan (v1.0) [53], and glimmerhmm (v3.0.4) [54]. The final gene set was generated by interating all data using Maker (v3.0103) [54] and EVidenceModeler (v2.1.0) [56]. Functional annotation of predicted proteins was performed by alignment against Uniprot [57], NR database [58], KOG, and Pfam databases using Diamond (v2.1.8) [59]. Gene functions and associated pathways were further assigned based on Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases. InterProScan (v5.55-88.0) [60] was used to identify protein conserved sequences, motifs, and domains for gene function annotation.

4.3.3. Visualization Analysis

Chromosome-level Hi-C interaction heatmaps were generated using HiCExplorer (v3.7.5) [61] to visualize genomic spatial interaction characteristics based on contig interaction intensity and position. Protein sequences were self-aligned using BLASTp software, and collinear blocks were identified with MCScanX (v0.8) [62]. Finally, a circos plot was generated using the R package circlize (v0.4.16) [63] to integrate and visualize genome structure and collinearity.

5. Conclusions

In this study, we report a high-quality, chromosome-level genome assembly for S. carolinense, generated by integrating PacBio HiFi long-read sequencing, Illumina next-generation sequencing, RNA-seq, and Hi-C chromosome scaffolding technologies. The final assembly spans 915.40 Mb with a contiguity (Contig N50: 51.06 Mb; Scaffold N50: 73.17 Mb) and a chromosomal anchoring rate of 96.05%. Despite a high repeat sequence content (73.64%), and a heterozygosity of 1.13%, the assembly achieved a BUSCO completeness score of 94.7%. Of the 32,206 predicted protein-coding genes, 97.95% were functionally annotated.

This genomic resource offers a critical foundation for investigating the adaptive mechanisms, defense strategies, and reproductive evolution driving the invasiveness and dispersal of S. carolinense. Furthermore, it provides the molecular insights necessary to identify novel herbicide targets and develop precision management strategies. This work serves as a valuable tool for future comparative genomics and transcriptomics analyses aimed at mitigating the ecological impacts on this invasive species.