Introduction

Potato (Solanum tuberosum L.) and its tuber-bearing wild relatives (collectively classified in Solanum section Petota Dumort.) form a rich gene pool that is useful in breeding. Hawkes (1990) recognized seven cultivated and 226 wild species, which were divided into 21 taxonomic series. However, many species are highly polymorphic and their species boundaries are often obscure. Spooner et al. conducted molecular studies on species boundaries and revealed many synonyms, which significantly reduced the species number to less than half of the 233 species described by Hawkes (Spooner et al. 2009; Solanaceae Source—http://www.nhm.ac.uk/research-curation/research/projects/solanaceaesource/). Chloroplast DNA phylogeny revealed four clades: (1) Mexican (including the southern US and Central America) diploid species, exclusive of S. bulbocastanum Dunal, S. cardiophyllum Lindl., and S. verrucosum Schltdl.; (2) S. bulbocastanum and S. cardiophyllum; (3) all examined members of the South American series Piurana and some South American species classified to other series, such as S. andreanum Baker; and (4) all remaining South American species, Mexican polyploid species, and S. verrucosum (Spooner and Castillo 1997).

Almost all diploids in the tuber-bearing Solanum species are self-incompatible (SI), since they are controlled by the gametophytic SI locus (Pushkarnath 1942; Pandey 1962; Cipar et al. 1964; Thompson and Kirch 1992). An exception is S. verrucosum that can self-pollinate (Abdalla and Hermsen 1973). Solanum verrucosum can be crossed as a female, but not as a male, with SI species (Abdalla and Hermsen 1973; Jansky and Hamernik 2009). The Mexican polyploid species, classified into a tetraploid species group (series Longipedicellata) and a hexaploid species group (series Demissa), are highly self-fertile, but are reproductively isolated from the Mexican diploid species, except for S. verrucosum. These polyploid species and S. verrucosum can be crossed with the South American species, including cultivated potatoes, and these species have the A genome in common (Matsubayashi 1991). Solanum stoloniferum Schltdl. is a highly polymorphic tetraploid that belongs to the series Longipedicellata. Cultivars carrying S. stoloniferum-derived cytoplasm exhibit complete male sterility, called “tetrad sterility” (Abdalla and Hermsen 1971) or “lobed sterility” (Grun et al. 1962) because anthers shed small quantities of pollen that mostly cluster in tetrads and have a four-lobed appearance. Solanum demissum Lindl. is a hexaploid species that belongs to the series Demissa and is widely used in potato breeding as a source of resistance to the most serious disease, late blight [Phytophthora infestans (Montague) de Bary] (Ross 1986; Plaisted and Hoopes 1989). A pentaploid hybrid from S. demissum × S. tuberosum is readily obtained when S. demissum is used as the female parent. The resulting pentaploid F1 hybrids produce abundant normal-looking pollen, but are non-functional as males and usually produce seeds only if backcrossed with the pollen of S. tuberosum (Dionne 1961). Thus, male gametophytes from S. verrucosum, S. stoloniferum, and S. demissum or genotypes that have the cytoplasm of these species exhibit peculiar behaviors during crossing. The molecular basis for their male sterility remains unknown.

Previously, we found that one amplified fragment length polymorphism band was different between reciprocal hybrids of S. tuberosum × S. demissum. This band was 170 bp in size, derived from S. demissum, maternally transmitted to subsequent progeny, and designated Band 1 (Sanetomo and Hosaka 2011a). Although its sequenced region was extended to 1,032 bp, it did not show any homology to known sequences. We surveyed for the presence or absence of Band 1 for 164 accessions of 36 cultivated and closely related wild species belonging to the South American taxonomic series Yungasensa, Megistacroloba, Conicibaccata, Piurana, Tuberosa, and Acaulia, two accessions of S. pinnatisectum Dunal (series Pinnatisecta), two accessions of S. stoloniferum (series Longipedicellata), and three accessions of S. demissum (series Demissa). Band 1 was detected only in three accessions of S. demissum (Sanetomo and Hosaka 2011b). Due to its specificity, we used Band 1 as an indicator of S. demissum cytoplasm when we developed a rapid identification method for potato cytoplasm types (Hosaka and Sanetomo 2012). Using this method, we evaluated the cytoplasmic diversity in varieties and breeding lines of different countries (Hosaka et al. unpublished). On the basis of our analysis, we noted that Band 1 was frequently amplified very weakly by polymerase chain reaction (PCR) from many genotypes that should not have S. demissum cytoplasm.

In this study, we further sequenced Band 1 containing regions to reveal its cellular origin. In addition, we conducted a comprehensive survey for the presence or absence of Band 1 in the tuber-bearing Solanum species, and discuss on its evolutionary implications.

Materials and methods

Plant material

A S. tuberosum breeding line Saikai 35 and S. demissum PI 186551 (referred to as T and D, respectively), F1 plants of D × T and T × D, and BC1 plants (D × T) × T and (T × D) × T were used to detect Band 1. The presence or absence of Band 1 was surveyed for four accessions of tomato group and 172 of 38 Solanum species, including non-tuber-bearing species (series Etuberosa), Mexican diploid series (series Morelliformia, Bulbocastana, Pinnatisecta, and Polyadenia), South American series not previously examined (series Circaeifolia and Lignicaulia), Mexican polyploid series (series Longipedicellata and Demissa), series Piurana and Conicibaccata, and S. verrucosum (Table 1). Solanum species seeds were obtained from the US Potato Genebank (NRSP-6; Sturgeon Bay, Wisconsin). The classification system of Hawkes (1990) was used throughout this text for continuity with our previous publications, although Spooner’s most updated taxonomy is also given in Table 1.

Table 1 Materials used in this study
Full size table

Band 1 detection by PCR

Total DNA was extracted from fresh leaves using the method of Hosaka and Hanneman (1998). Overlapping sequences of Regions 1, 2, and 3 (Fig. 1) were used to detect Band 1. PCR was performed using 5 μl of the PCR mixture containing 1 μl of template DNA (~5 ng/μl), 2.5 μl of Ampdirect® Plus (Shimadzu Co., Japan), 0.125 U Taq DNA polymerase (BIOTAQ™ HS DNA Polymerase, Bioline Ltd., UK), and 0.3 μM primers (Table 2). For a wide survey, a granule-bound starch synthase I gene (GBSS) marker (Table 2) was included at 0.3 μM in the reaction as a positive control to check that PCR was conducted accurately. PCR consisted of one cycle of 10 min at 95 °C; 35 cycles of 30 s at 94 °C, 30 s at 60 °C, and 1 min at 72 °C; and finally, one cycle of 5 min at 72 °C on the Veriti® 96-well thermal cycler (Applied Biosystems). Gel electrophoresis and staining procedures were as described previously (Sanetomo and Hosaka 2011b).

Fig. 1
figure 1

Schematic representation of sequences harboring Band 1. Broken lines indicate recombination-derived sequences

Full size image
Table 2 Primer sequences used in this study
Full size table

Genome walking to extend the Band 1 sequence

The sequenced region containing Band 1 (1,032 bp; Sanetomo and Hosaka 2011b) was extended from both ends using the LA PCR™ in vitro Cloning Kit (Takara Bio Inc., Japan) and DNA Walking SpeedUp™ Premix Kit II (Seegene, Inc.) according to the manufacturers’ instructions and using inverse PCR (Green and Sambrook 2012). PCR products were purified using the LaboPass™ PCR kit (Hokkaido System Science Co., Ltd., Sapporo, Japan) and directly sequenced.

Illumina sequencing

A genomic DNA library of a monoploid clone of S. verrucosum (11H23, PI 160228) was sequenced by a commercial service provider (Hokkaido System Science Co., Ltd., Sapporo, Japan). A 100-base paired-end run was performed on the Illumina Hiseq 2000 platform. All 100 bases in each read were filtered to ensure sequence quality and complexity, and generated high quality sequences with a total of 47 gigabases. After adapter-trimming, read sequences were assembled using the Velvet program (Zerbino and Birney 2008) for de novo assembly. The hash length was optimized to 75-mer, with which the Velvet assembly performed the highest contig contiguity (longer contig length) and specificity (fewer spurious overlaps). Solanum verrucosum contigs were BLAST-searched for homology with the extended S. demissum Band 1 sequence.

Distribution maps

To visualize the geographic distribution of presence or absence of Band 1, wild species accessions, the locality data obtained from the NRSP-6 database were localized on maps using DIVA-GIS software (Hijmans et al. 2001).

Results

Amplification variation of Band 1

A PCR-amplified band from Region 1 of Band 1 has been used as an indicator of S. demissum-derived cytoplasm among potato varieties (D marker in Hosaka and Sanetomo 2012). We frequently observed weak bands of the expected size from various species and varieties that did not carry S. demissum-derived cytoplasm (Fig. 2). The possibility of accidental contamination was excluded by examining all solutions, enzymes, and plastic ware used in our procedures. We performed experiments also in China and Peru. Our colleagues also examined using the same experimental protocols in their laboratories. However, this weak band was amplified randomly among samples, repeated experiments, people, and laboratories in different countries. For example in Fig. 3a, D, D × T, and (D × T) × T, all of which had S. demissum cytoplasm, showed clear amplified bands from Regions 1 to 3 of Band 1, whereas T, T × D, and (T × D) × T, all of which had S. tuberosum cytoplasm, showed very weak and similarly sized bands randomly among samples with different Regions. These weak bands were sequenced, which demonstrated sequences that were completely identical to those of the corresponding clear bands (data not shown). When PCR amplification was extended from the usual 35 to 40 cycles, all samples produced a clear Band 1 (Fig. 3b).

Fig. 2
figure 2

Representative gel photograph for evaluating the presence of Band 1 among varieties and breeding lines. Five genotypes clearly showed the presence of Band 1, whereas weak Band 1-like bands were observed in other genotypes. The left-most lane contains λDNA HindIII digests

Full size image
Fig. 3
figure 3

Detecting Band 1 using Region 1–3 markers in S. tuberosum (T), S. demissum (D), five each of F1 (T × D) and F1 (D × T), and five each of BC1 [(T × D) × T] and BC1 [(D × T) × T]. a Even in T, F1 (T × D) and BC1 [(T × D) × T], weak bands were detected randomly among samples and regions. b Longer PCR amplification cycles generated the Region 3 marker band from all samples, which made it impossible to discriminate Band 1 carriers from non-carriers. The marginal lanes contain λDNA HindIII digests

Full size image

Band 1 originated from the mitochondrial genome

The S. demissum DNA sequence harboring Band 1 (1,032 bp; Sanetomo and Hosaka 2011b) were extended from both terminal ends by genome walking to 7,040 bp (Fig. 1). Whole-genome sequencing of the S. verrucosum genome generated 174,235 contigs with an average length of 3,633 bases. The extended S. demissum sequence of Band 1 was BLAST-searched against these S. verrucosum contigs. One 10,794-bp contig was found to encompass the 7,040-bp S. demissum sequence with only 3 base substitutions in the corresponding regions (Fig. 1). A further search indicated that the 1,034-bp 3′-terminal end sequence of S. demissum or 1,063-bp 3′-terminal end sequence of S. verrucosum had 100 % homology with segment 2 of the S. tuberosum mitochondrial genome (http://solanaceae.plantbiology.msu.edu/pgsc_download.shtml; Potato Genome Sequencing Consortium 2011). The 3′-terminal ends (54 bp of S. demissum and 83 bp of S. verrucosum sequences) were included in the rps19 gene coding region (Fig. 1).

An F2 and R primer pair (Fig. 1; Table 2) amplified a 749-bp band from both S. demissum and S. verrucosum, but not from S. tuberosum (Fig. 4). An F1 and R primer pair (Fig. 1; Table 2) amplified a 660-bp band from not only S. tuberosum but also S. demissum and S. verrucosum. When PCR was subsequently run with three primers together, S. demissum and S. verrucosum amplified both 749- and 660-bp bands with approximately the same intensities (Fig. 4). The 660-bp band of S. demissum was sequenced and compared with the sequences of S. tuberosum segment 2, S. verrucosum contig, and extended Band 1 (Fig. 5). The 660-bp band of S. demissum had a sequence completely identical to that of S. tuberosum segment 2. However, the S. verrucosum contig and extended Band 1 sequences proximal beyond the point shown by the arrow in Fig. 5 were completely different from those of the 660-bp band of S. demissum and S. tuberosum segment 2; these showed no homology to known sequences.

Fig. 4
figure 4

PCR amplification using a S. tuberosum breeding line Saikai 35 (T), S. demissum PI 186551 (D), and S. verrucosum PI 160228 (V) and a pair of F1 and R1 primers (F1), a pair of F2 and R1 primers (F2), or a set of F1, F2 and R1 primers (F1/F2). M denotes λDNA HindIII digests

Full size image
Fig. 5
figure 5

Comparisons of PCR product sequences from S. demissum using a pair of F1 and R1 primers (normal type) with the corresponding regions of extended Band 1 (recombination-derived type), S. verrucosum contig, and publicly available segment 2 sequences of S. tuberosum mitochondrial genome

Full size image

Distribution of Band 1 among species

The presence or absence of Band 1 was surveyed for 176 accessions, among which 10 accessions of seven species were previously examined for Region 2 of Band 1 (Sanetomo and Hosaka 2011b). In this study, three primer sets (Regions 1–3) were used for all samples. However, these three primer sets gave the same results; thus, these results are shown simply as the presence or absence of Band 1 in Table 1. Band 1 was present as a clearly amplified band in 83 accessions of S. verrucosum and Mexican polyploid species (Table 1; Fig. 6). Neither tomato, South American potato species, nor Mexican diploid species had Band 1 (Table 1). The species of series Conicibaccata are distributed from Central America to South America, but Band 1 was not detected among these.

Fig. 6
figure 6

Representative photographs to evaluate Mexican polyploid species using Region 2 (upper) and Region 3 (lower) markers. GBSS marker was included in the PCR run as a positive control. Different accessions of S. schenckii (lanes 1–3), S. iopetalum (lanes 4–7), and S. demissum (lanes 8–16). M denotes λDNA HindIII digests

Full size image

Band 1 was detected in all 24 accessions of S. demissum. When multiple accessions per species were examined, all accessions of S. hjertingii Hawkes (6 accessions), S. guerreroense Correll (2 accessions), and S. hougasii (7 accessions) exhibited Band 1. In S. verrucosum, however, 7 of 9 accessions exhibited Band 1. Similarly, in other Mexican polyploid species, the presence or absence of Band 1 was segregated within these species, although at least one accession in each species exhibited Band 1. Thus, the presence of Band 1 was not specific to S. demissum cytoplasm, but was widely distributed among S. verrucosum and Mexican polyploid species. Also, the presence or absence of Band 1 did not correlate with species or groups defined by Spooner’s updated taxonomy (Table 1).

Geographical distribution of Band 1

Of 176 accessions surveyed in this study, 129 represent species that grow in North and Central America, and among them, 116 accessions of 20 species had detailed passport data; thus, their collection sites could be localized on a map (Fig. 7). Band 1 carriers (shown by solid marks) were scattered widely and randomly on this map. Most accessions were collected in the Mexican highlands. Thus, the presence or absence of Band 1 did not correlate with the collection site altitudes (data not shown).

Fig. 7
figure 7

Distribution of North and Central American accessions used in this study. Accessions shown by solid triangles, circles, and squares denote Band 1 carriers, while those shown by empty marks and crosses denote non-carriers

Full size image

Species distributions of series Tuberosa and Longipedicellata (Fig. 8) and Demissa (Fig. 9) were further investigated. The accessions of S. verrucosum were collected in a northern region (2 accessions) and a southern region (7 accessions); only the latter exhibited Band 1. All accessions of S. demissum, S. hjertingii, and S. hougasii exhibited Band 1, partly because all accessions of each species were collected in nearby vicinities. For all the other species, accessions with Band 1 were widely distributed within the species habitats and even accessions with and without Band 1 were localized close to each other.

Fig. 8
figure 8

Distribution of S. verrucosum and tetraploid species accessions. Circled accessions carry Band 1

Full size image
Fig. 9
figure 9

Distribution of hexaploid species accessions. Circled accessions carry Band 1

Full size image

Discussion

The present results indicated that the weak bands of Band 1 were amplified from extremely low DNA copy number (substoichiometric), while the clear bands of Band 1 were amplified from high DNA copy number (stoichiometric) in the samples. High-frequency inter- and intramolecular recombination is detected within the mitochondrial genomes of higher plants. We found that Band 1 originated in a recombination-derived mitochondrial subgenome (sometimes referred to as sublimon), and that Band 1 non-carriers also contained this subgenome with extremely low DNA copy number. A dramatic and rapid change in the relative copy number of portions of the mitochondrial genome over time of one generation is known as substoichiometric shifting (Janska et al. 1998). This often accompanies ectopic recombination, which derives chimeric sequences of mitochondrial DNA molecules (Kanazawa et al. 1994; Janska et al. 1998; Kim et al. 2007). Substoichiometric forms have been estimated at levels as low as one copy per every 100–200 cells (Arrieta-Montiel et al. 2001; Feng et al. 2009). Thus, it is likely that the Band 1-containing subgenomic mitochondrial DNA molecules are retained in the manner of substoichiometric shifting in tuber-bearing Solanum species.

Since the first report in maize (Small et al. 1987) of stable subgenomic mitochondrial DNA molecules within the genome at nearly undetectable levels, a recombination-derived subgenome has been found to be associated with cytoplasmic male sterility (CMS) in many plants, such as common bean (Janska et al. 1998; Arrieta-Montiel et al. 2001), Brassica napus (L’Homme and Brown 1993; Chen et al. 2011), Arabidopsis (Abdelnoor et al. 2003), Solanum (Scotti et al. 2004), and pearl millet (Feng et al. 2009). Chen et al. (2011) reported that in every cultivar of Brassica napus, both Polima (pol) and nap mitochondrial genomes were present. CMS plants primarily had a pol mitochondrial genome, and the nap mitochondrial genome was the primary genome of normal (fertile) plants. Large variations in the copy number ratios of pol and nap mitochondrial genomes were found even among cultivars sharing the same type of cytoplasm, and these ratios appeared to be maintained from generation to generation, except under conditions of CMS when the substoichiometric form may be amplified (Janska et al. 1998; Feng et al. 2009; Chen et al. 2011). Solanum verrucosum, S. stoloniferum, and S. demissum are self-fertile, and contradictorily, their cytoplasms are known to aberrantly affect male gametogenesis, resulting in non-functional pollen during interspecific hybridization as well as in hybrids (Dionne 1961; Grun et al. 1962; Abdalla and Hermsen 1971, 1973; Jansky and Hamernik 2009; Sanetomo et al. 2011). Mitochondrial ectopic recombination events have often been reported in a region associated with the ATPase subunit 6 (atp6) gene as well as with CMS in Brassica napus (Singh and Brown 1991; L’Homme and Brown 1993), pearl millet (Feng et al. 2009), and pepper (Kim and Kim 2006). The 5′-terminal sequence of the S. verrucosum contig harboring Band 1 appeared to be associated with atp6 gene (data not shown). Thus, Band 1 or the present recombination-derived mitochondrial subgenome may be associated with male gametogenesis.

In this study, we used all Longipedicellata and Demissa species, except for odd-ploidy species [S. vallis-mexici (3x), S. edinense (5x), and S. semidemissum (5x)]. The Band 1 sequence was found in all of the species examined, although its presence or absence was segregated within these species. Together with the results of our previous survey for cultivated and wild species in South America (Sanetomo and Hosaka 2011b), we conclude that Band 1 is shared only among S. verrucosum and Mexican polyploid species. Based on classical genome analysis of chromosome paring in interspecific hybrids and polyploid species, Longipedicellata and Demissa species are strict allotetraploid and allohexaploid species, respectively, and they share the A genome with S. verrucosum and many South American species (Matsubayashi 1991). Recently, the allotetraploid nature of S. stoloniferum (AABB genome) was confirmed by DNA sequence data for the GBSS gene (Spooner et al. 2008) and the nitrate reductase gene (Rodríguez and Spooner 2009) and by genomic in situ hybridization (GISH) analysis (Pendinen et al. 2008). DNA sequence data (Spooner et al. 2008; Rodríguez and Spooner 2009; Cai et al. 2012) and GISH analysis (Pendinen et al. 2012) also supported a taxonomic division of series Demissa into an allopolyploid Iopetala group (S. guerreroense, S. hougasii, S. iopetalum (Bitter) Hawkes, and S. schenckii Bitter, with component genomes A, B, and P) and an autopolyploid Acaulia group (S. demissum, possessing two minor variants of the A genome). Solanum verrucosum is the sole A genome species in North and Central America, and DNA sequence results support that S. verrucosum is an A genome contributor to all Mexican hexaploids (Spooner et al. 2008; Rodríguez and Spooner 2009). Chloroplast DNA analysis indicated that S. verrucosum and Mexican polyploids were in the same maternal lineage (Hosaka et al. 1984; Spooner and Castillo 1997). Solanum verrucosum was also proposed as a maternal ancestor of all Mexican polyploid species (Spooner et al. 2008; Rodríguez and Spooner 2009). The species distribution data for Band 1 also supports S. verrucosum as a maternal ancestor, from which all Mexican polyploid species were derived.

The Mexican polyploid species S. demissum is now classified into the Acaulia group together with the South American tetraploid species S. acaule Bitter and a hexaploid species S. albicans (Ochoa) Ochoa (Spooner et al. 2004). Their close relationships are well documented, based on data from morphology, AFLPs, nuclear RFLPs, and flavonoids (Spooner et al. 1995, 2004; Kardolus et al. 1998; Nakagawa and Hosaka 2002). However, Band 1 does not support the inclusion of S. demissum in the Acaulia group, because at least all examined accessions of S. acaule (two accessions) and S. albicans (two accessions) did not have Band 1 (Sanetomo and Hosaka 2011b).

All Mexican polyploid species and S. verrucosum possessed Band 1. However, species, ploidy, geographical distance, and altitude were not associated with the presence of Band 1. Moreover, the presence or absence of Band 1 was not uniform within these species, except for S. demissum, S. hjertingii, and S. hougasii in which all accessions had Band 1. We have sometimes observed the non-association of specific DNA segments with evolutionary events. For example, a 241-bp deletion of chloroplast DNA was found in some accessions of unrelated species S. berthaultii Hawkes (=S. tarijense Hawkes) and S. neorossii Hawkes et Hjerting (Hosaka 2002). Ames et al. (2007) discovered a 41-bp deletion of chloroplast DNA among some accessions of the unrelated species S. chiquidenum Ochoa, S. chomatophilum Bitter, and S. jalcae Ochoa. Multivariate morphological analyses showed no morphological associations to this deletion. They suggested that there was extensive interspecific gene flow among these three species, or that there was a common evolutionary history among these species that had never been suggested to be interrelated (Ames et al. 2007). In the case of Band 1, a common maternal history can be proposed. However, this does not completely explain the intraspecific variation of Band 1 because assuming all Mexican polyploid species originated from a Band 1 carrier of S. verrucosum, all polyploids would probably become Band 1 carriers. To explain this intraspecific variation at least three evolutionary hypotheses can be proposed: (1) tetraploids and hexaploids were all derived from Band 1 carriers of S. verrucosum, and later, they lost the Band 1 subgenome (intraspecific evolution from presence to absence); (2) tetraploids and hexaploids were all derived from Band 1 non-carriers of S. verrucosum, and later, they acquired the Band 1 subgenome (intraspecific evolution from absence to presence); or (3) tetraploids were derived multiple times from both Band 1 carriers and non-carriers of S. verrucosum, and hexaploids were also derived multiple times from Band 1 carriers and non-carriers of S. verrucosum or tetraploids (multiple origin).

Substoichiometric shifting is known to be controlled by nuclear genes, at least in Arabidopsis (Martinez-Zapater et al. 1992; Abdelnoor et al. 2003; Zaegel et al. 2006; Shedge et al. 2007) and Phaseolus vulgaris (Mackenzie and Chase 1990). The co-existence of nuclear genes that control substoichiometric shifting and stably retain a recombination-derived subgenome appears to be adaptive in plants (Feng et al. 2009; Chen et al. 2011). Therefore, we propose a fourth, alternative hypothesis. The most ancestral diploid species, S. verrucosum, evolved the nuclear genes that could trigger predominant amplification of a recombination-derived subgenome to retain both Band 1 carriers and non-carriers with a balanced ratio toward an adaptive peak of these species, and these nuclear genes were preferentially transmitted to the polyploid species. Thus, the intraspecific frequency of Band 1 carriers has neither decreased nor increased; rather, it has stabilized. To examine these hypotheses for the mechanism that retains the Band 1 subgenome, it will be necessary to investigate functional differences, particularly the reproductive capabilities between Band 1 carriers and non-carriers of the same species or those with the same genetic background. It will also be necessary to determine the molecular functions of these Band 1 sequences.