Xiaoshu, a simple genetic model system for sweetpotato (Ipomoea batatas (L.) Lam.)

Abstract

Sweetpotato [Ipomoea batatas (L.) Lam.] is an important food and feed crop. However, the genetic study of sweetpotato is relatively lagging due to its complex physiological and genetic characteristics. Sweetpotato belongs to the Batatas section of the genus Ipomoea in the Convolvulaceae. It is a hexaploid with a high level of heterozygosity in the genome. Most sweetpotato varieties are poorly self-compatible. The complexity of sweetpotato genome makes high-quality genome assembly challenging. The above characteristics significantly complicate forward or reverse genetics, trait mapping, genetic engineering, and genome sequencing in sweetpotatoes. A model system is needed to facilitate the genetic breeding of sweetpotatoes. We discovered a variety of I. cordatotriloba, one of the closest diploid relatives of sweetpotatoes, in our genebank. We considered this variety to have potential as a model system for sweetpotato, and we named it “xiaoshu”.

Xiaoshu is a sprawling plant with heart-shaped or slightly lobed leaves and pink flowers (Figure 1a). Xiaoshu is day-neutral and normally bloom about 50–60 days after sowing. It is self-compatible and naturally pollinated, and each plant can produce sufficient seeds throughout the growth period (Figures 1b and S1a, Table S1). These performances were stable across generations under natural conditions. Unlike most of wild relatives, the root of xiaoshu is swollen (Figure 1c). We observed transverse sections of the swollen root of xiaoshu under a microscope. The results showed that there are irregularly arranged secondary cambiums (SCs) in the central part of the root. Around the SC, there are many parenchyma cells (PCs) filled with starch granules (SGs) (Figure 1d). When the environment is suitable, the swollen roots can germinate and grow seedlings without external nutrients (Figure S1b). This indicates that the swollen roots already possess the characteristics of storage roots.

Analysis of k-mer multiplicity showed that the genome size (1C) of xiaoshu was about 478.62 Mbp, the heterozygosity rate was 0.17%, and the proportion of repetitive sequences was 59.53% (Figure S2 and Table S2). These findings indicated that the genome of xiaoshu was highly homozygous, in line with the characteristics of self-pollinated plants. A total of 91.72 Gb PacBio HiFi reads and 31.17 Gb Oxford Nanopore Technologies (ONT) ultra-long reads were used to obtain the preliminary genome assembly. The contigs were corrected and polished three times with BGI short reads and PacBio long reads to generate the chromosome-level genome. The 96 Gb of Hi-C reads were mapped to polish the contigs. The polished contigs were scaffolded, ordered, and anchored into pseudo-chromosomes using filtered Hi-C data. Subsequently, the corrected ONT reads were utilized for gap filling on the genome. Eventually, a gap-free reference genome named xiaoshu-T2T was produced that consisted of 15 chromosomes with a total length of 454 814 757 bp (Figure 1e,f). This was the first gap-free genome for the genus Ipomoea. BUSCO analysis showed that, of the 1614 plant-specific orthologues, 1586 (98.3%) were identified in the assembly, and 1575 (97.6%) of these were complete (Table S3). The average mapping rate of the assembled genome reached 95.89% using RNA-seq data from different tissues (Table S4). The LTR Assembly Index (LAI) was 21.54. The consensus quality value (QV) was 45.2, and the error rate was 0.003%. We predicted a total of 40 238 protein-coding genes in the haplotype xiaoshu genome. A total of 37 834 genes (94.03%) were further functionally annotated in the five different databases (Figure S3). BUSCO analysis indicates 1568 (97.2%) complete BUSCO genes in our annotation (Table S5).

To track the evolutionary history of xiaoshu, we examined the genomic relationships between xiaoshu and eight other sequenced plant species/varieties (I. batatas var. Xushu18, I. trifida var. NSP306, I. trifida var. Mx23Hm, I. triloba, I. nil, I. pes-caprae, I. cairica, and Solanum tuberosum). The phylogenetic tree based on 2996 single-copy gene families showed that xiaoshu was most closely related to I. triloba. Both of them are the closest wild relatives to sweetpotato except for I. trifida (Figure 1g). The divergence between the branches of xiaoshu and sweetpotato occurred ~7.12 Mya (Figure 1g). A total of 38 778 genes of xiaoshu were clustered into 23 591 gene families. Among these gene families, 398 were expanded and 436 were contracted in comparison with the other eight species/varieties (Figure 1g). The genes within the expanded families of xiaoshu were enriched in pathways like starch and sucrose metabolism (Figure S4a). The expression analysis revealed that five genes in this pathway were significantly more expressed in roots than in stems and leaves (Figure S4b). All five genes were annotated as beta-glucosidase BoGH3B, a hydrolase that is involved in decomposing cellulose into small molecular oligosaccharides. It suggested that these genes may be involved in starch accumulation or root enlargement process of xiaoshu.

The number of gene families shared by xiaoshu and sweetpotato was 17 987, accounting for 86.8% of the gene families of sweetpotato. This proportion is higher than that in NSP306 (85.90%), Mx23Hm (78.59%), and I. triloba (80.46%) compared to sweetpotato (Figure 1h). This indicated that xiaoshu-T2T is the best representative of the sweetpotato genome among available assemblies of wild relatives, and most of the genes in sweetpotato had orthologues in xiaoshu. Analysis of syntenic gene pairs demonstrated high similarity between the two genomes (Figure 1i). The structure variations (SVs) between the genomes of xiaoshu and sweetpotato were compared. It was discovered that the genome of xiaoshu contains 2372 insertions and 5938 deletions (Figure 1i and Table S6). Because SVs located upstream or in exons of genes may affect gene expression or function, we performed an enrichment analysis of these genes (SV genes, SVGs). The results showed that some SVGs were significantly enriched in pathways related to carbohydrate metabolism, possibly related to the starch filling and sweet taste of sweetpotato tuberous roots (Figure S5a). Moreover, some SVGs were significantly enriched in the pollen recognition pathway, and this may contribute to the difference in fertility between xiaoshu and sweetpotato (Figure S5b).

The T2T gap-free genome assembly provided an unprecedented opportunity to analyse the architectures of centromeres. In total, 30 telomeres and 14 centromeres were identified, and only the position of the centromere on chromosome 5 was ambiguous (Figure 1f). The lengths of centromeres ranged from 114 631 to 1 712 130 bp. The centromere regions were rich in high-order repeats (HORs), consisting of a total of 80 types of HORs (Figure S6). The 14 centromeres encompassed a total of 134 genes. The gene density in the centromere region (1.36 genes/100 kb) is much lower than that of the entire genome (8.85 genes/100 kb). About 77.6% (104) of the genes were not expressed (FPKM < 0.1) in any of the tested tissues. This is significantly higher than the non-expression rate of the entire genome (40.9%–45.3% in different tissues; Table S7). About 16.4% (22) of the genes were moderately or highly expressed (FPKM > 3.75) in at least one tissue. These genes had diverse molecular functions, such as DNA methylation, DNA binding, and catalytic activity (Table S7). The results indicated that the expressed genes in the centromere are involved in the functions of centromeres, thereby confirming the reliability of the centromere region identified in this study.

The model system demands a stable and efficient genetic transformation process. We referred to the Cut–dip–budding method (Cao et al., 2023), and developed a genetic transformation procedure without tissue culture for xiaoshu. Red fluorescent protein (DsRed) driven by a CaMV 35S promoter was as a marker to test the efficiency of the genetic transformation process. Possibly due to the Agrobacterium rhizogenes, the infected explants formed pencil roots instead of swollen roots like those of xiaoshu (Figure 1j). However, the pencil roots still had the potential to germinate and grow into seedlings and mature plants (Figures 1k and S7). We performed the transgenic experiment three times. The results showed that the success rate of genetic transformation of xiaoshu can reach as high as 95.65% (Figure 1l). PCR also verified the authenticity of the transgenic events (Figure S8a). Eventually, positive seeds (T₁) were obtained through positive plants (Figure S8b,c). This indicated that exogenous genes could be stably transmitted to offspring in this model system. Thus, it facilitates the fixation of genotypes and preservation, an advantage that sweetpotatoes do not typically have. Gene editing in sweetpotato faces many challenges such as multi-copy genes. To determine whether the genes of xiaoshu could be easily edited by CRISPR-Cas9, phytoene desaturase (PDS) was chosen as the target for easy identification of gene-edited plants. Six transgene-positive roots were obtained from 10 infected explants. Among these, five generated transgene-positive shoots, and two of which produced albino shoots (Figure 1m). This suggested that xiaoshu is an efficient system for gene editing. As a model system for sweetpotato, it is important that the function of genes from sweetpotato can be identified in this system. The IbMYB1 from purple sweetpotato was cloned, driven by the CaMV 35S promoter, and overexpressed in xiaoshu. The transgenic roots were purple, including the skin and flesh (Figures 1n and S9). Furthermore, the anthocyanin content and IbMYB1 expression level in transgenic roots were examined. The results showed that the higher the expression of IbMYB1, the higher the anthocyanin content (Figure 1o). This indicated a strong correlation between target gene expression and phenotype in this system, suggesting that xiaoshu is a suitable system for the gene validation of sweetpotatos.

In conclusion, in this study, we have identified a wild relative of sweetpotato, xiaoshu, that has the potential to serve as a model system for sweetpotato. A T2T genome of xiaoshu was generated, and an efficient genetic transformation procedure for xiaoshu was developed. Our results will facilitate the genetic research and breeding of sweetpotatoes.

The authors declare no conflict of interest.

Q.C. and S.X. designed and conceived the study, and wrote the manuscript. S.X. and Y.W. led the studies related to transgenes. S.X. and P.X. led the studies related to the genome. Lu.Z., B.G., Li.Z., Z.Z. and X.D. participated in the transgenes, microscopic observation, planting and phenotyping of plant materials. All authors have read and approved the manuscript.