Prunus sensu lato (s.l.), a genus in the Rosaceae family, contains ten subgenera with approximately 250–400 flowering tree or shrub species (Chin et al., 2014). This genus comprises numerous species of significant economic value, including peach, plum, and cherry, along with oilseed species like almond (Govaerts et al., 2021). Members of Prunus share several conserved morphological traits (Rehder, 1927), such as simple leaves bearing basal or petiolar glands, a superior single ovary, and drupaceous fruits. Furthermore, all species in this genus possess a basic chromosome number of eight (Lin et al., 1991). Historically, subgenera within Prunus have been categorized into three major groups based on inflorescence structure: (1) racemose inflorescence in subgenera Padus Mill., Laurocerasus Tourn. ex Duh., Pygeum Gaertn., Prinsepia Royle., and Maddenia Hook. f. et Thoms.; (2) corymbose inflorescence in subgenus Cerasus Mill.; (3) solitary flowers in subgenera Amygdalus L., Prunus L. sensu stricto (s.s.), Armeniaca Mill., and Emplectocladus Torr. (Potter et al., 2007; Hodel et al., 2021).

The Prunus subgenus Laurocerasus represents a basal lineage among these subgenera (Hodel et al., 2021), characterized by a high frequency of tetraploidy (Meurman, 1929; Zhao et al., 2016). It comprises approximately 80 species, including Prunus spinulosa Siebold & Zucc., an evergreen species distributed in southeastern China, the Philippines, and parts of Japan (Deng et al., 2024). P. spinulosa typically inhabits moist, forested environment adjacent to large rivers, at elevations from 300 to 1500 m (Govaerts et al., 2021). Its glossy leaves superficially resemble those of holly (Ilex bioritsensis Hayata). In traditional Chinese medicine, seeds of P. spinulosa have been used to treat dysentery (Zhang et al., 1994).

At the time this study was conducted, 18 genome assemblies of Prunus species were publicly available (Table S1). However, only two of them, Prunus padus L. and P. fruticosa Pall., are polyploid and belong to subgenera Padus and Cerasus, respectively (Table S2). This imbalance in available genomic resources has limited our understanding of Prunus, particularly among polyploid species and underrepresented subgenera. Here, we report the first chromosome-level haplotype-resolved genome assembly of P. spinulosa (the first reference genome for subgenus Laurocerasus) with PacBio HiFi and Hi-C sequencing, which contains 87,838 genes. This high-quality tetraploid genome provides a valuable genetic resource for resolving the complex phylogenetic relationships and genome polyploidization evolution of Prunus.

Fresh leaves of Prunus spinulosa were collected from an individual in Hangzhou Botany Garden, Zhejiang Province, China (30°24′96″N, 120°12′20″E) for genome sequencing. A voucher specimen (No. ZSN240501-PSpin-001) has been deposited at the laboratory of systematic and evolutionary botany and biodiversity, Zhejiang University. Fresh young leaves, roots, and twigs were also collected from a seedling in Wuyun Mountain, Hangzhou (30°19′11″N, 120°09′41″E) for gene annotation with transcriptome sequencing. All samples were immediately frozen in liquid nitrogen and then stored at −80 ℃ before further use. Plant materials were delivered to Novogene (Beijing, China) for whole genome sequencing, while transcriptome sequencing was performed by Anoroad (Beijing, China).

We generated 98.32 Gb of Hi-C data, 21.28 Gb of HiFi long-read data, and 235.49 Gb of Illumina short-read resequencing data with the Illumina NovaSeq X Plus and Illumina HiSeq 2500 platforms, corresponding to an overall sequencing depth of approximately 318×. Transcriptome sequencing produced 2.44 Gb, 2.33 Gb, and 2.55 Gb from leaf, twig, and root, respectively (Table S3). Sequencing adapters were trimmed and clean reads were quality-filtered using fastp v.0.23.12 (Chen et al., 2018). Reads with low-quality bases (Q ≤ 5) or with more than 10% ambiguous bases (N) were removed.

Flow cytometry analysis was conducted with the maize (Zea mays L.) inbred line B73 as the internal standard, which gave an estimate of the genome size of Prunus spinulosa approximately equal to 1.14 Gb. This estimate is substantially larger than those of known diploid Prunus species, which typically range from 227.6 Mb to 344.3 Mb (Table S1). Additionally, to assess the heterozygosity and ploidy of P. spinulosa, we conducted a genome survey using Illumina short reads following a two-step k-mer based protocol. First, Jellyfish v.2.3.0 (Marçais and Kingsford, 2011) was employed to compute the 21-mer frequency, which was further analyzed in GenomeScope v.2.0 (Vurture et al., 2017). The estimated haploid genome size was 219 Mb, with 59% unique k-mer (Fig. S1A). Results of GenomeScope also illustrated that the k-mer distribution of P. spinulosa was highly consistent with typical patterns of heterozygous polyploid genomes (Ranallo-Benavidez et al., 2020). Next, Smudgeplot v.0.4.0 (James et al., 2020; Ranallo-Benavidez et al., 2020) was used to estimate the heterozygosity and ploidy of P. spinulosa, given the same 21-mer histogram with the lower and upper thresholds of k-mer set to 15 bp and 285 bp, respectively. The results suggested the genomic composition of P. spinulosa most likely match that of either an AB-type diploid (48% probability) or an AABB-type allotetraploid species (36% probability). The probabilities of being other ploidies were negligible, including AAB (7%), AAAB (6%), and AAAABB (3%) (Fig. S1B). Collectively, all above analyses support an allotetraploid origin for P. spinulosa.

All Pacbio HiFi and Hi-C reads were first de novo assembled into contigs using Hi-C integration mode implemented in hifiasm v.0.19.8 (Cheng et al., 2021) with default parameters. The initial assembly gave a full genome size of 1131.54 Mb, which was consistent with the estimate of flow cytometry. Hi-C reads were aligned to contigs by bwa v.0.7.17 (Li, 2013) with parameters ‘-mem -5SP -S -h -b -F 3340’. Reads with mapping quality lower than 30 were discarded using SAMtools v.1.6 (Li et al., 2009). The contigs were then scaffolded into 32 pseudochromosomes with HapHiC (Zeng et al., 2024), which were further ordered and oriented based on Hi–C interaction signals using Juicebox (Durand et al., 2016). Ultimately, 90.23% of the contigs was anchored to 32 pseudochromosomes with a final size of 1021.06 Mb and a contig N50 of 31.9 Mb (Table S2). The reference genome assembly comprised two representative subgenome pairs, with sizes of 501.9 Mb and 519.16 Mb, respectively. Hi-C contact maps showed strong intra-subgenomic interactions with clear diagonal patterns (Fig. 1B). The completeness of the P. spinulosa genome assembly was evaluated with BUSCO v.5.4.7 (Manni et al., 2021) based on the eudicots_odb10 dataset, which contained 1614 conserved single-copy genes. The result showed an overall completeness of 99.4% (Fig. S1C and Table S2).

To resolve the subgenome structure, synteny analysis was used to compare the genome of Prunus spinulosa to that of P. campanulata, which had the highest assembly quality among available closely related species. A total of 16 pseudochromosomes showing higher mapping percentages (46.46–81.18%) than P. campanulata were designated as subgenome A using RagTag (Alonge et al., 2022). The other 16 pseudochromosomes with lower mapping percentages (15.75–37.52%) were designated as subgenome B (Fig. 1C). The final genome assembly comprised four haplotypes with each size ranging from 249.82 Mb to 259.69 Mb (Fig. 1A). Scaffold N50 values were calculated by QUAST v.5.2.0 (Gurevich et al., 2013) and varied from 31.35 Mb to 33.25 Mb depending on haplotypes. The LTR Assembly Index (LAI) scores (Ou et al., 2018) varies from 17.25 to 17.83, suggesting a high level of repeat sequence continuity across all haplotypes (Table 1).

Given the biological significance of tandem repeats, we constructed a repeat sequence library using RepeatModeler v.2.0.5 (Flynn et al., 2020) for Prunus spinulosa genome, which was combined with Repbase v.21.12 (Jurka, 2000), and then used to identify repetitive elements by RepeatMasker v.4.1.5 (Tarailo-Graovac and Chen, 2009). In total, 129.89 Mb–145.99 Mb of repeat contents were identified for P. spinulosa four haplotypes, taking up to more than 55% of the whole genome (Table 1).

Subsequently, a comprehensive protocol integrating transcriptome-assisted prediction, homology-based prediction, and ab initio prediction, was performed to annotate the protein coding genes for Prunus spinulosa genome. For transcriptome-assisted prediction, RNA-seq reads were aligned to the genome using HISAT2 v.2.2.1 (Kim et al., 2015) as well as de novo assembled into transcripts using Trinity v.2.15.1 (Grabherr et al., 2011). Gene models were then predicted using PASA v.2.3.3 (Haas et al., 2003) based on both read alignments and assembled transcripts. Next, homology-based gene prediction was conducted using GeMoMa v.1.9 (Keilwagen et al., 2018) with protein sequences from P. campanulata as the reference. Later, ab initio gene prediction was performed by GeneMark-ET v.4.68_lic (Lomsadze et al., 2005) and AUGUSTUS v.3.3.3 (Stanke et al., 2004), respectively. The results were combined by BRAKER v.3.0.3 (Hoff et al., 2016). Finally, gene models predicted by all three approaches were integrated using EvidenceModeler v.1.1.1 (Haas et al., 2008) with weighted contributions set to ‘PASA: BRAKER: GeMoMa = 10: 5: 1’. The longest transcript containing both start and stop codons was retained for each gene model. In total, 87,838 protein-coding genes were annotated across the four haplotypes, with gene counts equal to 22,599 in A1, 21,272 in A2, 22,668 in B1, and 21,299 in B2. Correspondingly, 115,922, 105,134, 114,848, and 105,811 exons were identified in A1, A2, B1, B2 haplotype, respectively (Table 1). The completeness of protein coding gene annotation was assessed with BUSCO, which suggested 99.3% of conserved genes were complete, only 0.2% were fragmented and 0.5% were missing.

To assess the collinearity of genome structure within Prunus, we performed synteny analyses using NGenomeSyn v.1.41 (He et al., 2023). Analyses were performed both among the four haplotypes of P. spinulosa (Fig. S1E) and across Prunus genomes from different subgenera (Fig. 1D). Although P. kanzakura and P. yedoensis were included in the initial analyses, they were not shown in the final visualization due to their relatively fragmented assemblies. Overall, most syntenic blocks were well preserved across subgenera, suggesting genome structure is highly conserved in Prunus despite differences in ploidy. In particular, P. spinulosa exhibited high synteny with closely related species from subgenera Cerasus and Padus, further supporting the overall quality and structural accuracy of the P. spinulosa genome assembly.

Phylogenetic relationships among the Prunus species were reconstructed to provide an evolutionary framework for interpreting synteny patterns (Fig. 1D). Orthologous genes were identified with OrthoFinder v.2.5.5 (Emms and Kelly, 2019), focusing on single-copy orthologs (SCOs) and low-copy orthologs (LCOs), the latter defined as genes present in more than 80% of species with less than 5 copies per species. LCO sequences were aligned using MAFFT v.7.505 (Katoh et al., 2002) with ‘-gt 0.8 -st 0.001 -cons 60’, trimmed with trimAL (Capella-Gutiérrez et al., 2009), and concatenated via AMAS.py (Borowiec, 2016) under ‘concat -f phylip -d dna’. The maximum likelihood tree was inferred in IQ-TREE v.2.4.0 (Nguyen et al., 2015) with Fragaria vesca (Shulaev, 2011) as outgroup. ModelFinder Plus (‘-m MFP’) tested 1223 models, and node support was assessed with 1000 ultrafast bootstrap replicates and SH-like aLRT tests (‘-bb 1000 --alrt 1000’). The resulting topology was consistent with previous classifications of Prunus (Hodel et al., 2021; Su et al., 2023), confirming the close relationship of P. spinulosa with subgenus Padus (Fig. 1D).

In summary, the chromosome-level, haplotype-resolved reference genome of Prunus spinulosa presented in this study is of high quality and shall serve as an important genetic resource for future phylogenomic analysis and evolutionary research within Prunus.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grant No. 32570239) and Key Technology Research and Development Program of Zhejiang Province (Grant No. 2023C03138). We thank Dr. Junjie Wu (Zhejiang University) for his help during the genome annotation of this study.

Data availability

The genome assembly, annotation, and raw reads of HiFi, Hi-C, Illumina resequencing data, and RNA-seq of Prunus spinulosa that support this study have been deposited into the China National Center for Bioinformation with bioproject accession number: PRJCA040363.

CRediT authorship contribution statement

Sining Zhang: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Formal analysis, Data curation, Investigation. Jun Chen: Methodology, Writing – review & editing, Conceptualization, Supervision, Resources, Project administration. Pan Li: Methodology, Writing – review & editing, Funding acquisition, Conceptualization, Supervision, Resources, Investigation, Project administration.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2025.09.007.