Introduction

The sweetpotato [6X Ipomoea batatas (L.) Lam] is a member of the genus Ipomoea, the largest genus in the morning glory (Convolvulaceae) family. This family contains approximately 50 genera and more than 1,000 species. Over half of these species are concentrated in the Americas, where they are distributed as cultigens, medicinal plants and weeds1. Among the morning glories, I. batatas is the only species with an economic importance as a major food crop2, although I. aquatica is also cultivated and consumed as a leafy vegetable, mainly in South-East Asia. Series Batatas is a subdivision within the genus Ipomoea. This is a relatively young clade that diversified circa 12 million years ago3. This group includes the cultivated hexaploid sweetpotato [I. batatas (L.) Lam], the tetraploid (4x) sweetpotato I. batatas (L.) Lam4, and 13 other species considered to be the wild relatives of the cultivated sweetpotato. These wild relatives are I. cordatotriloba, I. cynanchifolia, I. grandiflora, I. lacunosa, I. leucantha, I. littoralis, I. ramosissima, I. splendor sylvae (previously named umbraticola), I. tabascana, I. tenuissima, I. tiliacea, I. trifida and I. triloba5,6. Members of the series Batatas are endemic to the Americas, except I. littoralis that is native to Madagascar, South and Southeast Asia, Australia, and the Pacific region5. The basic chromosome number of the series Batatas species is 2n = 2 ×  = 30. While most species are diploid (2x), several are tetraploid (4x) or hexaploid (6x)7. To avoid confusion, hereafter in the current text, the (6x) sweetpotato (I. batatas) will be referred to as Ib6x, the tetraploid form of I. batatas as Ib4x, and the combination of both as “the sweetpotato group”.

The sweetpotato is a crop native to the Americas and it was an important food crop for the Inca and Mayan cultures. Its origin and center(s) of genetic diversity have been proposed as somewhere between the Yucatan Peninsula of Mexico and the mouth of the Orinoco River in Venezuela8,9, Peru and Ecuador9. Papua New Guinea, Indonesia and the Philippines are suggested as secondary centers of diversity10. Today, sweetpotato is a major staple food in numerous tropical countries11. However, its botanical origin and details about its domestication remain under debate.

Several hypotheses have been put forward to explain the sweetpotato’s botanical origin. Nishiyama12 proposed, based on cytogenetical studies, that Ib6x could have originated from the diploid species I. leucantha, from which the tetraploid I. littoralis was derived through polyploidization. The hybridization between these two species could have produced I. trifida, which is suggested to have different ploidies. Further cross-pollinations between these wild species, followed by selection and domestication of interesting genotypes, could have produced the Ib6x. Based on morphological and cytogenetical data, two additional hypotheses were subsequently suggested. Shiotani13 suggested that I. trifida forms an autopolyploid complex, and that the cultivated Ib6x is derived from this group. Austin8 suggested that the cultivated sweetpotato was derived from a hybridization event between I. trifida and I. triloba. Other studies carried out using molecular markers (RFLP, RAPD and SSR)14,15,16, beta-amylase gene sequences17 and cytogenetic analysis18 supported a contribution of I. trifida to the cultivated sweetpotato genome.

Advances in DNA sequencing technologies have allowed the assembly of complex polyploid genomes, including that of the cultivated sweetpotato. Yang et al.19 identified six haplotypes based on the assembly of a monoploid genome (15 pseudo chromosomes). The phylogenetic analysis of these haplotypes permitted the authors to trace back the hexaploidization process of Ib6x giving rise to a new hypothesis on its origin. These authors19 suggested that the cultivated sweetpotato could have arisen from a cross between a tetraploid and a diploid progenitor. The most likely diploid progenitor is I. trifida, while the tetraploid progenitor is currently unknown. It is not unreasonable to suspect that Ib4x, described by Bohac et al.20; Jarret et al.16; Roullier et al.21, which are known to share haplotypes with Ib6x22, might be the tetraploid progenitor.

A more recent, but related, hypothesis about the origin of the cultivated sweetpotato has been proposed by Muñoz-Rodríguez et al.23. These authors, based on the phylogenetic analyses of nuclear and chloroplast DNA regions, have proposed that Ib6x has a monophyletic origin (by autopolyploidization) and suggested that I. trifida is its most probable progenitor. This hypothesis also indicated a second role for I. trifida in the origin of the sweetpotato. Once Ib6x arose from I. trifida, it expanded its distribution range further than I. trifida’s natural distribution. Over time, both species became reciprocally monophyletic and then hybridized, giving rise to two cultivated sweetpotato lineages.

These previous investigations suggest that a further study of Ib4x and their wild relatives in series Batatas is required since they are key in efforts to elucidate the botanical origin of the cultivated sweetpotato.

The discovery of Agrobacterium IbT-DNA1 and IbT-DNA2, inserted into the Ib6x genome constitutes a noteworthy example of an ancient HGT event in a domesticated crop24. IbT-DNA1 contains genes for auxin biosynthesis (TR-T-DNA like), while IbT-DNA2 contains RolB/C genes (TL-T-DNA like). The acquisition of these genes by the cultivated sweetpotato and other Ipomoea species opens the possibility that these sequences have played a role in the evolution of this crop and its related species25. However, whether the acquisition of one or both IbT-DNAs by the Ib6x genome occurred before or after its speciation remains unknown. To address this issue, it is necessary to evaluate the presence/absence of IbT-DNA1 and IbT-DNA2 insertions in members of the sweetpotato group and/or other members of the series Batatas. The resulting knowledge might be expected to shed light on the botanical origin of the cultivated sweetpotato and also provide critical clues related to the time of the ancestral Agrobacterium infection(s). Hence, the current study proposes to evaluate (i) the presence of IbT-DNA1 and IbT-DNA2 in the sweetpotato group and other Ipomoea (series Batatas) species and (ii) the use of IbT-DNA1 and IbT-DNA2 genes as markers to reconstruct the evolutionary history of the sweetpotato.

Results

Distribution of IbT-DNA1 and IbT-DNA2 in Ipomoea spp. series Batatas

The presence of Agrobacterium T-DNAs (IbT-DNA1 and IbT-DNA2) in the genome of Ib6x was demonstrated by Kyndt et al.24. Likewise, a limited number of wild relatives, including Ib4x and member species of the series Batatas, were evaluated in that work. Nine Ib4x and four representatives of the species I. triloba, I. tabascana and I. trifida were tested for the presence of IbT-DNA genes [Acs, C-prot, iaaH, iaaM and ORF13 (Open Reading Frame 13)] by PCR, using sequence-specific primers. None of IbT-DNA genes were detected in these samples except for the ORF13 gene (on IbT-DNA2) in I. trifida.

The current analysis was extended to include a total of 14 species representative of Ipomoea series Batatas, 2 species corresponding to other Ipomoea members (not in series Batatas) and 5 from related genera (Supplementary Data; Tables 1–4) using newly designed degenerate primers. IbT-DNA1 genes were detected in Ib4x (3 out of 15) and 3 other species in the series Batatas, including; I. cordatotriloba (1 out of 5), I. tenuissima (1 out of 1) and one ambiguous Ipomoea sp. (2 out of 2). The IbT-DNA2 gene was detected in 8 out of 15 Ib4x and 9 out of 28 I. trifida (Fig. 1). No other Ipomoea species outside of the series Batatas (0 out of 2) and no species from related genera (0 out of 5) examined in this study tested positive for the presence of IbT-DNA genes by PCR using the degenerate primers.

Figure 1
figure 1

IbT-DNA1 and IbT-DNA2 detected in the wild relatives.

Full size image

The presence of IbT-DNA1 was analyzed and confirmed by DNA blot analysis in two PCR positive Ib4x accessions (PI 518474 and CIP 403270) and the three PCR positive wild relatives (Ipomoea sp. and I. cordatotriloba). Ipomoea batatas (L.) Lam. var. apiculata (PI 518474) (Fig. 2A3) showed four bands - like Ib6x (Fig. 2B1); while CIP 403270 (Ib4x) showed only one (Fig. 2A2). Ipomoea sp. CIP 460250 (2x) displayed at least 1 band (Fig. 2B2), whereas Ipomoea cordatotriloba PI 518494 (2x) (Fig. 2C2) and Ipomoea sp. CIP 460814 (2x) (Fig. 2C1), appear to have at least four bands. The presence of IbT-DNA2 was only tested and confirmed in Ib4x PI 518474 (1 band – Fig. 2D).

Figure 2
figure 2

Southern blot with IbT-DNA1 (C-prot probe, A–C) and IbT-DNA2 (ORF17n probe, D) on Spe I digests of Ipomoea spp. series Batatas. (A1) DNA ladder; (A2) I. batatas (L.) Lam CIP 403270 (4x); (A3) I. batatas (L.) Lam var. apiculata PI 518474 (4x). (B1) I. batatas (L.) Lam cv. Huachano CIP 420065 (6x); (B2) Ipomoea sp. CIP 460250 (2x). (C1) Ipomoea sp. CIP 460814 (2x); (C2) I. cordatotriloba PI 518494 (2x). D1) I. batatas (L.) Lam PI 518474 (4x). In A3, sizes with ** were estimated from the DNA ladder. In B2, B3, C1, C2, D1, sizes with * were estimated from I. batatas (L.) Lam cv. Huachano CIP 420065.

Full size image

Characterization of wild Ipomoea species

Phenotypic characterization (using ~30 descriptors, compiled based on Austin26 and Huamán27) confirmed the identity of accessions (Supplementary Data, Tables 2–4), with some exceptions. CIP 460250, which was collected as I. trifida, lacks the correct fruit and flower characteristics for the species. CIP 460397, collected as I. tiliacea, possesses flowers suggesting I. trifida. CIP 460786, collected as I. grandifolia, was morphologically similar to I. cordatotriloba. Conversely, CIP 460814 and CIP 460815 were collected as I. cordatotriloba, but had the characteristics of I. grandifolia (I. grandifolia and I. cordatotriloba are very similar, differing only in the size of the corolla, and some authors consider them varieties of the same species). CIP 460002 was collected as I. leucantha, which is a hybrid species between I. trichocarpa and I. lacunosa and which has highly variable characteristics. CIP 460811 was collected as I. cordatotriloba, however its flower color is white rather than violet as is typical for I. cordatotriloba.

Phylogeny of IbT-DNA1 and IbT-DNA2 genes among Ipomoea species

Phylogenetic analyses were performed to determine how IbT-DNA sequences are related in the genus Ipomoea (Figs 3–7). Four phylogenetic trees were inferred using the IbT-DNA1 genes C-prot (827 nt; Fig. 3), Acs (792 nt; Fig. 4), iaaH (641 nt; Fig. 5), and iaaM (485 nt; Fig. 6). The results obtained consistently showed that the Ib6x and Ib4x accessions group together (bootstrap value 71–99%), with the wild relatives as a sister clade [Ipomoea sp. (2 out of 2), I. cordatotriloba (1 out of 5) and I. tenuissima (1 out of 1)]. Both groups, Ib6x and Ib4x and their wild relatives, form a monophyletic group as compared to homologous genes from other sequenced T-DNAs; suggesting that they belong to the same lineage with a common origin.

Figure 3
figure 3

Phylogenetic tree generated by Neighbor-Joining of C-prot (827 nt) alignment. Values at the nodes show percentage of bootstrap support (of 1,000 bootstrap replicates) and they are indicated if greater than 50. Accession numbers (CIP/PI) and ploidy level are indicated for Ipomoea spp. whereas plasmid names are indicated for Agrobacterium spp. GenBank accession numbers are provided between brackets when available. Tetraploids (4x) I. batatas (L.) Lam are highlighted in grey.

Full size image
Figure 4
figure 4

Phylogenetic tree generated by Neighbor-Joining of Acs (792 nt) alignment. Values at the nodes show percentage of bootstrap support (of 1,000 bootstrap replicates) and they are indicated if greater than 50. Accession numbers (CIP/PI) and ploidy level are indicated for Ipomoea spp. whereas plasmid names are indicated for Agrobacterium spp. GenBank accession numbers are provided between brackets when available. Tetraploids (4x) I. batatas (L.) Lam are highlighted in grey.

Full size image
Figure 5
figure 5

Phylogenetic tree generated by Neighbor-Joining of iaaH (641 nt) alignment. Values at the nodes show percentage of bootstrap support (of 1,000 bootstrap replicates) and they are indicated if greater than 50. Accession numbers (CIP/PI) and ploidy level are indicated for Ipomoea spp. whereas plasmid names are indicated for Agrobacterium spp. GenBank accession numbers are provided between brackets when available. Tetraploids (4x) I. batatas (L.) Lam are highlighted in grey. Ipomoea sp. CIP 430434 was previously labeled as tetraploid (4x) I. batatas (L.) Lam.

Full size image
Figure 6
figure 6

Phylogenetic tree generated by Neighbor-Joining of iaaM (485 nt) alignment. Values at the nodes show percentage of bootstrap support (of 1,000 bootstrap replicates) and they are indicated if greater than 50. Accession numbers (CIP/PI) and ploidy levels are indicated for Ipomoea spp. whereas plasmid names are indicated for Agrobacterium spp. GenBank accession numbers are provided between brackets when available. Tetraploids (4x) I. batatas (L.) Lam are highlighted in grey.

Full size image
Figure 7
figure 7

Phylogenetic tree generated by Neighbor-Joining of ORF13 (492 nt) alignment from IbT-DNA2. Values at the nodes show percentage of bootstrap support (of 1,000 bootstrap replicates) and they are indicated if greater than 50. Accession numbers (CIP/PI) and ploidy level are indicated for Ipomoea spp. GenBank accession numbers are provided between brackets for Agrobacterium spp. and Nicotiana spp. Tetraploids (4x) I. batatas (L.) Lam are highlighted in grey.

Full size image

In the case of the IbT-DNA2 ORF13 gene (492 nt; Fig. 7), the analysis indicates that Ib6x and Ib4x accessions grouped together in a well-supported clade (bootstrap value 99%) that includes one I. trifida accession PI 561544. The rest of the I. trifida samples formed a basal group and together with the sweetpotato group, they form a well-supported lineage (bootstrap value = 100). Nucleotide sequences from two species of the genus Nicotiana were included in the analysis of IbT-DNA2. The results show that those are phylogenetically closer to A. rhizogenes strains pRi2659 (AJ271050.1), K599 (EF433766.1) and MAFF03-01724 (AP002086.1) in comparison with the Ipomoea sequences.

IbT-DNA1 and IbT-DNA2 gene similarities among Ib6x and its wild relatives

Pairwise comparisons of identities of partial nucleotide sequences of IbT-DNA1 genes (C-prot, Acs, iaaH, iaaM) and IbT-DNA2 gene (ORF13) were estimated. Nucleotide sequence identity values are above 99% for all genes analyzed within the sweetpotato group; which includes both Ib6x and Ib4x. Of note is that ORF13 from Ipomoea trifida PI 561544 shows higher identity values (~99.9%) with the sweetpotato group than the rest of the Ipomoea trifida accessions (Supplementary Data, Tables 6 and 7). Among the sweetpotato group and its wild relatives, the identity values of all genes analyzed ranged from 96–98.8%. Previously, IbT-DNA1 was found to be inserted in two copies, in the form of a partial inverted repeat, in the genome of the Ib6x cv. Xu78124. In the present study, the nucleotide sequence identity between the two copies of IbT-DNA1 (Fig. 8) was calculated in Xu781, which corresponded to 98.8% (divergency 1.2%).

Figure 8
figure 8

IbT-DNA1 insertion in F-box gene. (A) A schematic representation of F-box gene (Taizhong 6) showing their 5 exons; a deletion (19 bp) in the target site is represented as dot lines among exon 3 and 4. (B) IbT-DNA1 (Xu 781); IbT-DNA1 and its inverted repeat are presented as interrupted black arrows. The region flanking IbT-DNA1, to be analyzed in the next section (Fig. 7), is indicated as red arrows and its size (687 bp) is placed between brackets.

Full size image

Ib6x and Ib4x share the same insertion site of IbT-DNA1

A phylogenetic analysis of the region flanking IbT-DNA1 (687 nt; F-box third intron) was performed in order to elucidate the evolutionary relationship among all accessions in the sweetpotato group carrying IbT-DNA1 (Fig. 9). The alignment included: F-box-IbT-DNA1 sequences of six Ib6x accessions and three Ib4x accessions; F-box gene (without IbT-DNA1) of two Ib6x and three Ib4x; and F-box gene of the wild relatives I. trifida, I. triloba, I. cordatotriloba and Ipomoea sp. CIP 460250. An F-box gene sequence from I. nil, cv. Tokyo-kokei, were included as an outgroup. The resulting tree shows that the Ib6x and Ib4x F-box genes carrying IbT-DNA1 group together in a well-supported clade (bootstrap value = 99%). Likewise, sequences corresponding to the F-box gene uninterrupted by IbT-DNA1 appear in a sister clade. This suggests that the F-box gene carrying IbT-DNA1 might have diverged from the original F-box gene (either before or after the T-DNA insertion or both) and that the Ib6x and the Ib4x belong to the same lineage with a common origin. The nucleotide sequence identity calculated between F-box intact and F-box-IbT-DNA1 was 96.9% (3.1% divergence). The regions flanking IbT-DNA1 from I. tenuissima, I. cordatotriloba and Ipomoea sp. could not be included in the analysis since we were unable to amplify them with the primers designed.

Figure 9
figure 9

Phylogenetic tree generated by Neighbor-Joining of the F-box gene intact and containing IbT-DNA1 (687nt, F-box third intron) Values at the nodes show percentage of bootstrap support (of 1,000 bootstrap replicates) they and are indicated if greater than 50. Accession numbers (CIP/PI) and ploidy level are indicated for Ipomoea spp. F-box gene interrupted by IbT-DNA1 is indicated as “F-box-IbT-DNA1”, whereas F-box gene without IbT-DNA1 is labeled as “F-box gene”. Tetraploids (4x) I. batatas (L.) Lam are highlighted in grey. The GenBank accession number is provided between brackets for I. nil.

Full size image

Analysis of IbT-DNA2 in cultivated sweet potato Taizhong 6

The region flanking IbT-DNA2 in the Ib6x genome has not been described previously. It was predicted based on whole-genome sequencing data from cv. Taizhong 6. This analysis indicated that IbT-DNA2 (cv. Taizhong 6) is inserted in chromosome 7 and has an estimated size of 11,187 bp (Fig. 10). It comprises seven open reading frames (ORFs) homologous to ORF18/ORF17n, ORF13, RolB/RolC family, ORF17n, ORF14 and a hypothetical protein with a “NADB Rossman” domain of Agrobacterium rhizogenes. Compared to IbT-DNA2 in cv. Huachano (KM052617), there is an insertion of 369 bp within ORF13 cv. Taizhong 6. The region flanking IbT-DNA2 was confirmed using PCR, and on the basis of significant homology (via tblastx) it was identified as the mitochondrial substrate carrier family protein UcpB - the highest score associated with Ipomoea nil (e-value = 6e-108; score = 1494). There is also an uninterrupted copy of the UcpB gene (without IbT-DNA2) on chromosome 7 of cv. Taizhong 6, that is 4,004 bp in size with nine exons. The insertion site of IbT-DNA2 was determined by comparing UcpB and UcpB-IbT-DNA2. On one side, the T-DNA is flanked by an intronic region with high A/T-content after exon 7 while the other side is located in an intronic region 24 bp upstream from exon 9. Linked to the T-DNA insertion, there is a deletion of 893 bp in the UcpB gene that includes exon 8 (Fig. 10).

Figure 10
figure 10

IbT-DNA2 insertion sites in UcpB gene. (A) A deletion of 893 bp is indicated as a grey box with dot lines between 7th and 8th introns. (B) IbT-DNA2 of cv. Taizhong 6, including ORFs with significant homology to ORF18/ORF17n, ORF13, RolB/RolC family, ORF17n, ORF14 and a hypothetical protein with a “NADB Rossman” domain. Insertion sites are indicated as red lines; DNA filler as dark blue boxes at both ends. The region flanking IbT-DNA2, to be analyzed in the next section (Fig. 10), is indicated as red arrows and its size (750 bp) is shown.   

Full size image

Ipomoea trifida, Ib6x and Ib4x share the same IbT-DNA2 insertion site

A phylogenetic analysis of the region flanking IbT-DNA2 (750 nt, sixth intron – seventh exon) was performed in order to elucidate the evolutionary relationship between UcpB genes, with and without IbT-DNA2 (Fig. 11). The alignment included: UcpB-IbT-DNA2 sequences from one Ib6x, two Ib4x, one I. trifida, and UcpB gene sequences (without IbT-DNA2) from one Ib6x, two Ib4x, one I. trifida and one I. triloba. A UcpB sequence from I. nil, cv. Tokyo-kokei, was included as an outgroup. The resulting tree shows that Ib6x, Ib4x and I. trifida UcpB sequences carrying IbT-DNA2, group together in a well-supported clade (bootstrap value = 100%). Likewise, sequences containing only the UcpB gene (without IbT-DNA2) appear in a sister clade. In addition, the nucleotide sequence identity between ucpB and UcpB-IbT-DNA2 was estimated 95.7% (divergency 4.3%).

Figure 11
figure 11

Phylogenetic tree generated by Neighbor-Joining of the UcpB gene intact and containing IbT-DNA2 (750 nt, six intron– seven exon) Values at the nodes show percentage of bootstrap support (of 1,000 bootstrap replicates) and they are indicated if greater than 50. Accession numbers (CIP/PI) and ploidy level are indicated for Ipomoea spp. UcpB gene interrupted by IbT-DNA2 is indicated as “UcpB-IbT-DNA1”, whereas UcpB gene without IbT-DNA2 is labeled as “UcpB gene”. Tetraploids (4x) I. batatas (L.) Lam are highlighted in grey. The GenBank accession number is provided between brackets for I. nil.

Full size image

Discussion

Our data demonstrate that the HGT event of Agrobacterium into series Batatas taxa is not confined to the hexaploid sweetpotato. It is present also in its wild relatives, which includes its tetraploid form, as well as other members of the series Batatas. We report here the detection of sequences homologous to IbT-DNA1 and IbT-DNA2 genes in at least ten accessions corresponding to Ib4x and fourteen accessions belonging to I. trifida, I. cordatotriloba, I. tenuissima, and a currently unidentified Ipomoea sp. from the series Batatas. Accessions belonging to the genus Ipomoea, but not members of series Batatas, and other related genera, were also analyzed. These included members of the Quamoclit group and species from the genera Calystegia, Xenostegia, Operculina and Merremia. The presence of IbT-DNA1 and IbT-DNA2 could not be confirmed in any of these samples. However, it should be noted we cannot exclude the possibility of false negatives in our analyses, and our findings likely represent an underestimation of the HGT events across the target species. This is because despite using degenerate primers and Southern blots, only regions corresponding to a few genes were tested, and remnants of (re-arranged) T-DNAs may exist that do not contain these complete regions. Also, we generally only tested one or two seedlings from each wild Ipomoea sp. accession (which are maintained as seeds) and if the accession was segregating for T-DNAs their presence could have been missed by chance.

The tetraploid form of I. batatas has been poorly characterized and its taxonomic status remains unclear. This taxon, collected from Ecuador, Colombia, Guatemala and Mexico, has been a subject of interest for over 50 years. The fact that these samples form thickened “pencil-shaped” storage roots has been considered as evidence that the tetraploids are primitive sweetpotatoes28. Some accessions were initially tentatively identified as I. trifida but later they were classified as wild I. batatas20. Subsequently, it was observed that the tetraploid form shared haplotypes (based on chloroplast and nuclear DNA markers) with the cultivated hexaploid21. These findings reinforced the hypothesis proposed by several authors, who suggested that tetraploid I. batatas are the closest wild relative of the cultivated sweetpotato21,29.

In the current study, nucleotide sequence analyses (pairwise comparisons) of IbT-DNA1 and IbT-DNA2 genes reveal high identity values (above 99%) among accessions from the sweetpotato group (Ib6x and Ib4x). These results were supported by the phylogenetic analyses of the regions flanking IbT-DNA1 and IbT-DNA2, which showed that Ib6x and Ib4x share the same insertion site (Figs 9 and 11). These findings reinforce previous taxonomic and molecular studies20,21 and suggest that I. batatas includes both hexaploid and tetraploid forms. However, there is also a possibility that the tetraploid form represents an interspecific hybrid between I. batatas and a close wild relative (I. trifida). We suggest the use of IbT-DNA1 and IbT-DNA2 genes as markers to further elucidate the origin of the sweetpotato in a manner similar to the use of Agrobacterium T-DNAs to reconstruct the evolution of Nicotiana and Linaria30.

The series Batatas contains the sweetpotato group and 13 other species considered to be its closest wild relatives5,6. Within this group, the species I. trifida has been identified as a potential wild ancestor in several studies based on morphological data, molecular markers and cytogenetic analyses14,15,16,17,18. Recently, two studies have reopened the debate about the role of I. trifida in the origin of the sweetpotato. Yang et al.19 analyzed a complete 6x I. batatas genome and proposed that the crop species could have resulted from a cross between a tetraploid and a diploid (most likely I. trifida) progenitor. Such a hybridization would have resulted in triploid progeny that, subsequently undergoing genome duplication, would result in 6x forms. In contrast, Muñoz- Rodriguez et al.23, based on genomic analyses of whole chloroplast and single-copy nuclear DNA regions, proposed that I. trifida played a dual role in the origin of the cultivated sweetpotato. Firstly, to form the first I. batatas lineage, as its most likely progenitor by autopolyploidization and, secondly, as the species that this autopolyploid (6x) later hybridized with to produce another independent sweetpotato lineage. Most recently, Wu et al.31 found through sequence comparison of the genome of hexaploid I. batatas with the genomes of I. trifida and I. triloba, that approximately one third of the hexaploid I. batatas genome shows higher similarity to I. triloba than to I. trifida. In relation to the data in the present study, the detection of IbT-DNA2 (ORF13 gene) only in the I. trifida accessions (9 out of 28) examined, and not in the other series Batatas species examined, provides additional evidence supporting the close relationship of this species (I. trifida) with the hexaploid and tetraploid forms of I. batatas. Furthermore, the phylogenetic analysis of IbT-DNA2 and its flanking region indicated that I. batatas (6x and 4x) and I. trifida originated from a common ancestor.

Similar to the cT-DNAs in Nicotiana species30, it is possible that IbT-DNA2 was acquired initially by I. trifida (or a common ancestor of I. trifida and I. batatas) and later transmitted across speciation events to the sweetpotato. This hypothesis is reinforced by the fact that I. trifida, together with Ib6x and Ib4x, share the same insertion site of IbT-DNA2. An alternative explanation for the presence of Ib-TDNA2 in the sweetpotato involves its transfer by interspecific hybridization that is known to occur between I. batatas and I. trifida21. Ipomoea trifida accessions carrying the ORF13 gene do not form a monophyletic group as PI 561544 appears in the clade of the sweetpotato group. This accession was collected in Venezuela and could represent the closest sweetpotato wild relative, in addition to the tetraploid form of I. batatas.

Species from the series Batatas other than I. trifida have also been proposed as potential contributors to the origin of the sweetpotato, albeit these hypotheses are less generally accepted within the community. Jarret et al.16 considered I. tabascana (4x), I. trifida and K233 (4x, suggested to be a hybrid between I. batatas and I. trifida) to be the closest relatives of the cultivated sweetpotato based on RFLPs, among the taxa examined (which did not include Ib4x). Recently, Eserman32 concluded, based on hybridization analysis, that Ib6x could have hybrid ancestry, with parentage from I. ramosissima and either I. triloba or I. cordatotriloba. The present study indicates the presence of IbT-DNA1 genes in accessions belonging to the species I. cordatotriloba, I. tenuissima and two as yet unclassified Ipomoea accessions (CIP 460250 and CIP 460814). Our phylogenetic trees of IbT-DNA1 genes indicate that the sweetpotato group, I. cordatotriloba, I. tenuissima and Ipomoea sp. form a strongly supported (~99% bootstrap) monophyletic clade as compared to their homologues in Agrobacterium spp., suggesting a common ancestry. The identity of the two Ipomoea sp. accessions containing IbT-DNA1 has not been elucidated. These accessions were initially classified as I. trifida (CIP 460250) and I. cordatotriloba (CIP460814). However, upon morphological re-evaluation, it became clear that they were not consistent with the recorded classification. The latter shows phenotypic characteristics consistent with I. grandifolia, whereas the formers’ characteristics are not consistent with any of the established species. This was also confirmed by molecular markers, which showed CIP 460250 formed a sister clade compared to other Ipomoea series batatas33. It is not clear to what extent, if any, mis-identification of plant materials may have clouded efforts to resolve relationships within this group of taxa.

The presence of IbT-DNA1 in Ib4x, I. cordatotriloba, and other Ipomoea spp. from the series Batatas was confirmed by southern blot analyses. Tetraploid I. batatas (CIP403270 and PI 518474) and wild relatives (Ipomoea sp. and I. cordatotriloba) show dissimilar banding patterns when compared to Ib6x. Additionally, the identity values of IbT-DNA1 genes, among the sweetpotato group members and the wild relatives, range between 96–98.8% which is lower than within the sweetpotato group (above 99%). Thus, if the T-DNAs found in the series Batatas spp. represent a single ancestral event, it indicates that IbT-DNA1 sequences have evolved and diverged since their acquisition by the sweetpotato’s ancestors. Recently, Ipomoea evolutionary trees have been calibrated, with an estimated mutation rate of 0.7% base pairs per million years19. The divergency between the repeats of IbT-DNA1 is 1.2%, which leads to an estimated age of IbT-DNA1 of 1.7 million years. Muñoz-Rodríguez et al.23 pointed out that the clade including the sweetpotato and I. trifida diverged from its sister clade at least 1.5 million years ago. Considering that IbT-DNA1 is estimated to be older than the clade containing the sweetpotato and its potential ancestor (I. trifida); it is possible that IbT-DNA1 might have been acquired early in the evolution of these species. Consequently, IbT-DNA1 was fixed in the course of the evolution of the sweetpotato; while in other wild relatives it became less common, and in I. trifida this region could have been lost completely. The fact that I. trifida samples analyzed in this study do not contain IbT-DNA1, supports this possible course of events.

Based on the current data, at least two hypotheses arise to explain the combined origin of IbT-DNA1 and IbT-DNA2 in the hexaploid I. batatas. Hypothesis I suggests that the HGT from A. rhizogenes (or an ancestral related species) may have occurred in a single event, transferring both IbT-DNAs into a common ancestor of the species I. trifida, I. tenuissima, I. cordatotriloba and I. triloba. Subsequently, both regions were passed (independently or in combination) to I. trifida, I. tenuissima, I. cordatotriloba and I. triloba (or primitive forms). Later, one of these potential progenitors passed IbT-DNAs to the tetraploid I. batatas (L.) Lam by speciation, which later became I. batatas (L.) Lam (6x). Hypothesis II proposes that the HGT from Agrobacterium spp. into the cultivated sweetpotato’s ancestor might have occurred via two or more independent events. It is possible that at least two species independently acquired IbT-DNA1 and/or IbT-DNA2 and then two of them combined in the common ancestor of I. batatas (L.) Lam (4x) and (6x). This hypothesis could explain the fact that the flanking region of IbT-TDNA1 in I. tenuissima and I. cordatotriloba could not be amplified, despite using various sets of primers. Future efforts to determine the flanking sequences in these accessions should be able to confirm or discard this hypothesis. Nevertheless, based on our current data, because HGT events that enter the host germline are relatively rare in nature, and because of the clear correspondence between the phylogeny of the T-DNA genes and the species taxonomy, hypothesis I seems the most likely.

Material and Methods

Plant materials

In total, 114 plant samples were included in the present study. Detailed information on the accessions is included in Supplementary data (Tables 1–4). The materials included 11 accessions of hexaploid Ipomoea batatas, 15 accessions belonging to tetraploid Ipomoea batatas (4x), 82 accessions encompassing 13 species of the series Batatas, 2 accessions from other Ipomoea sp. (not series Batatas) and 5 accessions corresponding to related genera. The series Batatas species were distributed (numbers within parenthesis are the number of accessions sampled, within species) as: Ipomoea trifida (28), I. triloba (14), I. cordatotriloba (5), I. grandifloria (5), I. tiliacea (8), I. ramosissima (7), I. leucantha (5), I. tabascana (1), I. tenuissima (1), I. littoralis (1) and I. splendor-sylvae (2), I. lacunosa (1), I. cynanchifolia (1), unverified Ipomoea sp. (2). The other Ipomoea spp. examined (not series Batatas) are I. heredifolia (1) and I. quamoclit (1). Ipomoea-related species (other genera) included Merremia quinquefolia (1), Merremia dissecta (1), Calystegia longipipes (1), Xenostegia tridentata (1) and Operculina aequisepala (1). The plant materials were provided by the germplasm collection of the International Potato Center (CIP, Lima, Peru) and The National Genetic Resources Program (NGRP, USDA, USA).

For taxonomic verification, tetraploid I. batatas accessions from CIPs Genebank (3 siblings per accession) were germinated in a petri dish and then transferred to planting trays (Jiffy 7) for 15 days after which they were transferred into screenhouses for characterization using 30–60 descriptors (Rossel et al., unpublished). To determine the ploidy levels, samples from young leaves were analyzed in an Accuri C6 flow cytometer (BD Biosciences) with propidium iodide and data were analyzed with BD Accuri C6 Software. This was supplemented by chromosome counting in squashed root-tips stained with aceto-orcein as required.

DNA sequences from other sources

Published DNA sequences from five Ipomoea spp. were added to our nucleotide alignment and analyses; including those derived from the genome browsers of cv. Taizhong 6 (http://ipomoea-genome.org/), I. trifida NSP306 and I. triloba NSP323 (http://sweetpotato.plantbiology.msu.edu/). The last two of these do not contain IbT-DNA genes. Genebank and BAC library (KM113766) nucleotide sequences (KM113766), belonging to cv. Xu 781 and I. nil (XM 019334701.1 and XM 019341879.1), were also aligned and analyzed. I. nil does not contain IbT-DNA genes and was used as an outgroup.

DNA extraction

Laboratory procedures detailed below were performed essentially as previously described24. DNA extraction from fresh leaf tissues of 115 samples was performed using the CTAB method34. DNA quantity and quality were measured using a Nanodrop spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA) and agarose gel electrophoresis, respectively.

Screening for IbT-DNAs in Ipomoea spp

Detection of IbT-DNA1 and IbT-DNA2 genes in Ipomoea samples was carried out by PCR using primers listed in Supplementary data (Table 5). The degenerate primers were designed for each gene (Acs, C-prot, iaaH, iaaM and ORF13) manually by examining multiple alignments of the target sequences from Agrobacterium spp. and 6x I. batatas (Supplementary Data Table 5.2). Part of the Ipomoea-specific malate dehydrogenase gene (MDH) was amplified from each DNA sample as a positive PCR control. The detection of the chromosome regions flanking IbT-DNA1 and IbT-DNA2 were carried out in Ipomoea spp. containing these regions by PCR. Likewise, uninterrupted F-box and ucpB genes were amplified by PCR. The PCR specific primers are listed in Supplementary data (Supplementary Data Table 5.3). PCR reactions were accomplished in 25-µl volumes containing 1x PCR buffer (Invitrogen, Carlsbad, CA, USA), 0.4 mM each of dGTP, dATP, dTTP, and dCTP; 0.3 µM of forward and reverse primer; 1 Unit of Taq DNA polymerase (Invitrogen); and 100 ng of genomic DNA. The PCR conditions were 94 °C for 5 min, followed by 35 cycles of 95 °C for 30 s, 50°–60 °C for 30 s, and 72 °C for 2 min, and then a final extension at 72 °C for 10 min. PCR products were separated on 1% agarose gels for visual detection of DNA.

Sequencing and sequence analyses

PCR products were recovered using the Wizard SV gel extraction kit (Promega) according to the manufacturer’s recommendation. The eluted DNA was ligated into plasmid vector pCR 2.1 (Invitrogen), according to the manufacturer’s instructions, and cloned in Escherichia coli strain DH5α. PCR products were sequenced by LGC genomics, using the Sanger method and then assembled using the software Seqman II (DNAstar, Inc. Madison, WI, USA). Sequence alignments, phylogenetic analyses and pairwise comparison were performed using the software MEGA 535.

IbT-DNA2 annotation

The flanking region of IbT-DNA2 in the Ib6x genome was predicted based on whole-genome sequencing data from cv. Taizhong 6. IbT-DNA2 in Taizhong 6 was annotated based on the top hits when performing blastn searches in the genome browser http://ipomoea-genome.org/.

Southern blot hybridization

Southern blot analyses were performed to confirm previous PCR data on selected Ipomoea samples. Two probes complementary to the ORF coding for C-protein and ORF17n were utilized for these assays.

A total of 30 µg of genomic DNA was digested with Spe I, separated on a 0.8% agarose gel under 25 eV for 18 h, and transferred to a Hybond-N+ nylon membrane (Amersham Pharmacia Biotech, Piscataway, NJ, USA) with transfer buffer (20x SSC).

Primers used to amplify the DNA probes C-prot and ORF17n are listed in Supplementary Data (Table 5.4). Probe labeling was performed using the PCR DIG Probe Synthesis Kit (Roche, West Sussex, UK). Pre-hybridization and hybridization steps were carried out using the buffer DIG Easy Hyb (Roche), according to the manufacturer’s instructions. Following hybridization, membranes were washed twice (5 min) at low stringency (2x SSC, 0.1% SDS) at room temperature and two additional times (15 min) at high stringency (0.1x SSC, 0.1% SDS) at 65 °C. The images were captured by chemiluminescence on photosensitive film (Fujifilm Life Science).