Chromosome evolution in Solanum traced by cross-species BAC-FISH

Authors


Author for correspondence:
Yuling Bai
Tel: +31 317 482136
Email: bai.yuling@wur.nl

Summary

  • Chromosomal rearrangements are relatively rare evolutionary events and can be used as markers to study karyotype evolution. This research aims to use such rearrangements to study chromosome evolution in Solanum.
  • Chromosomal rearrangements between Solanum crops and several related wild species were investigated using tomato and potato bacterial artificial chromosomes (BACs) in a multicolour fluorescent in situ hybridization (FISH). The BACs selected are evenly distributed over seven chromosomal arms containing inversions described in previous studies. The presence/absence of these inversions among the studied Solanum species were determined and the order of the BAC-FISH signals was used to construct phylogenetic trees.
  • Compared with earlier studies, data from this study provide support for the current grouping of species into different sections within Solanum; however, there are a few notable exceptions, such as the tree positions of S. etuberosum (closer to the tomato group than to the potato group) and S. lycopersicoides (sister to S. pennellii). These apparent contradictions might be explained by interspecific hybridization events and/or incomplete lineage sorting.
  • This cross-species BAC painting technique provides unique information on genome organization, evolution and phylogenetic relationships in a wide variety of species. Such information is very helpful for introgressive breeding.

Introduction

Chromosome painting based on fluorescent in situ hybridization (FISH) can detect individual chromosomes in nuclei and cell complements, and represents one of the most common cytogenetic methods for establishing structural and numerical chromosomal variants in all eukaryotic model species, including yeast, Arabidopsis and man (Lysak et al., 2001; Schubert et al., 2001). The method is also very powerful in demonstrating large-scale chromosomal rearrangements that may be responsible for or accompanied by unique events leading to evolutionary divergence (e.g. Müller et al., 2003). In plants chromosome painting has been a big challenge for a long time because repeats in pooled DNA probes from isolated chromosomes paint all chromosomes equally and thus do not allow such probes to detect individual chromosomes (Schubert et al., 2001). This issue was overcome by Lysak et al. (2001) who selected repeat-poor bacterial artificial chromosomes (BACs) covering whole chromosome arm euchromatin regions as probes for FISH detection of individual Arabidopsis thaliana chromosomes. Later comparative painting studies of Arabidopsis BAC probes on chromosome complements of related Brassica species under lower stringency allowed comparative chromosome painting among related species, revealing the evolution of their karyotypes (Lysak et al., 2003, 2007; Lysak & Lexer, 2006). Recently, the production and allocation of various tomato and potato BAC libraries has allowed adaptation of the cross-species BAC-FISH mapping technique to Solanum crops and wild species, and revealed cytogenetic evidence of known and novel chromosomal rearrangements between tomato, potato and related Solanum species (Iovene et al., 2008; Tang et al., 2008; Lou et al., 2010).

The study of the genetic and cytogenetic relationship between tomato and potato has a long history. Gottschalk’s (1954) pioneering work in chromosome morphology revealed a surprising similarity in overall chromosome morphology among tomato, potato and several other Solanum species. Later studies of the genetics and genomics of Solanum crops revealed large-scale synteny (Bonierbale et al., 1988; Tanksley et al., 1992; Grube et al., 2000; Doganlar et al., 2002; Fulton et al., 2002; Wu & Tanksley, 2010), but also demonstrated varying numbers of translocations and inversions between tomato (Solanum lycopersicum), potato (S. tuberosum), eggplant (S. melongena) and pepper (Capsicum spp.) (Table 1). Such major chromosomal rearrangements, which are supposedly relatively rare and independent events, are one of the dramatic processes that shape the genome and the karyotype, and thus can be used as phylogenetic markers for the study of chromosome evolution in the Solanaceae family.

Table 1.   Overview of chromosomal rearrangements in Solanum and Capsicum based on comparative genetic mapping
Chromosome (Tomato (T))Wild relatives of tomatoPotato (P)Eggplant (E)Pepper (C)Citation1
  1. Tomato (S. lycopersicum) is used as the reference. Chromosome arms are named with numbers (chromosome number) followed by either S (short arm) or L (long arm).

  2. 1a, Doganlar et al. (2002); b, Tanksley et al. (1992); c, Bonierbale et al. (1988); d, Livingstone et al. (1999); e, Seah et al. (2004); f, Canady et al. (2006); g, Stamova & Chetelat (2000); h, Pertuzéet al. (2002); i, van der Knaap et al. (2004); j, Albrecht & Chetelat (2009).

  3. 2Description of the chromosome part is according to Doganlar et al. (2002).

5 5S inversion5L + 12 L
5S and 5L inversion
5S and 5L inversiona, b, c, d
6Mi-homologues on 6S inverted in S. Peruvianum PI 128657; 6S Inversion in S. Juglandifolium LA2788 Upper2
6 inversion
Co-lineara, f, e, j
77S Inversion in S. Pennellii LA716 Two inverted segmentsUpper2
7 scattered
a, d, i
9 9S inversionNine inversionsNine inversions;
additional rearrangements
a, b, c
1010S Inversion in S. Juglandifolium LA2788
10L inversion in S. sitiens LA1974 and S. Lycopersicoides LA2951
10L inversion5S + 12S + 10L
Lower2
10 inversion
Lower2
10 inversion
a, b, c, f, h
11 11S inversion11S inversion
Lower2
11 inversion
4S + 11S
T11S = C12L
11S inversion (indication)
a, b
1212S proximal inversion in S. Chilense LA0458
Reciprocal translocation with 8S in S. ochranthum LA3650 and SJuglandifolium LA2788
12S inversionUpper T = E = C
12S + 11L
Lower2
12 inversions with T and P
Upper2
T = E = C
Translocation 12S 11S
a, b, g, j

Phylogenetic relationships between Solanum species have been the subject of various studies (e.g. Bohs & Olmstead, 1997; Weese & Bohs, 2007). Many studies have focused on the economically important species of section Lycopersicon (wild and cultivated tomatoes, e.g. Peralta & Spooner, 2001; Spooner et al., 2005; Peralta et al., 2008), section Petota (wild and cultivated potatoes, e.g. Spooner & Raul Castillo, 1997; Jacobs et al., 2008) and the interrelationships between these groups (Spooner et al., 1993; Rodriguez et al., 2009). The different types of data used in these studies (morphology, AFLPs, sequences of chloroplast DNA, internal transcribed spacer (ITS), the GBSSI gene and COSII markers) have often resulted in conflicting phylogenetic reconstructions.

In this paper we present the analysis of inversion events as a parameter of chromosomal evolution in Solanum species, with an emphasis on species related to tomato and potato. We detected chromosomal rearrangements by cross-species BAC-FISH using tomato and potato BACs as probes on chromosome complements of various Solanum species of the tomato and potato clade. With S. melongena as the outgroup representative, we constructed a phylogenetic tree that was then compared with a generalized tree derived from a number of the studies mentioned above. We discuss the power of this chromosome painting technique in relation to other techniques such as comparative genetic mapping (Wu & Tanksley, 2010), DNA sequences comparison (Wang et al., 2008; Wu et al., 2009a,b) and chromosome paring analysis of F1 hybrids (de Jong et al., 1993; Anderson et al., 2010).

Materials and Methods

Plant materials

The plants used in this study were from section Lycopersicon (six species), section Lycopersicoides (one species), section Juglandifolia (one species), section Etuberosum (one species) and section Petota (five species) (Table 2). In addition, two eggplant cultivars were included (Table 2). Plants were grown in a glasshouse and young flower buds were collected in the morning and immediately fixed in freshly prepared acetic acid/ethanol (1 : 3) at 4°C. The next day, the flower buds were transferred to 70% ethanol for storage at 4°C. In total, 18 tomato and 17 potato BACs (Fig. 1) were selected covering seven chromosome arms (chromosomes 5, 6, 7, 9, 10, 11 and 12; five BACs per chromosome arm), where previous papers on comparative genetics suggested inversions. All BACs were repeat-poor, except for H146I19 that hybridized to the heterochromatin of several chromosomes of all species of the subsection Lycopersicon. This BAC was used because no repeat-poor BAC was available in the middle of the euchromatin on the short arm of chromosome 12.

Table 2.   Plant materials used in this study
SpeciesAccession/introgression line/cultivarTaxonomy1
  1. 1The taxonomic classification is according to Peralta et al. (2008) except for S. melongena that belongs to subgenus Leptostemonum, section Melongena.

Solanum peruvianumLA2172Section Lycopersicon
S. peruvianumLA2157
S. habrochaitesG1.1290
S. habrochaitesG1.1560
S. pimpinellifoliumG1.1554
S. habrochaites f.  glabratumCGN.1561
S. lycopersicumHeinz 1706
S. chilenseLA1969
S. pennelliiLA716
S. lycopersicoidesCGN90124 (PI255549 or PI365378)Section Lycopersicoides
S. ochranthumLA2166Section Juglandifolia
S. etuberosumPI558054Section Etuberosum
S. bulbocastanumPI275198Section Petota
S. tarijenseCGN22729
S. megistracrolobumPI458396
S. pinnatisectumGLKS2268
S. tuberosumRH-89-039-16; van der Voort et al. (1997)
S. melongenaHalf Lange Violette and MM738Section Melongena
Figure 1.

Chromosomal location of tomato (Solanum lycopersicum) and potato (Solanum tuberosum) BACs on the seven studied chromosome arms, based on BAC-FISH signals on the tomato chromosomes.

Slide preparation

Slides were prepared according to the protocol of Szinay et al. (2008) with the following minor modifications. The standard enzyme mix containing 1% pectolyase Y23 (Sigma P-3026), 1% cellulase RS (Yakult 203033; Yakult Pharmaceutica, Tokyo, Japan) and 1% cytohelicase (Bio Sherpa 24970-014) was diluted 10 times with 10 mM sodium citric buffer (pH 4.5) for most of the species. For species that have pollen mother cells with thick callose walls the dilution was less: Solanum melongena (three times dilution); S. pennellii, S. lycopersicoides and species of section Petota (two times diluted stock).

BAC labelling

BACs were isolated as described in Szinay et al. (2008). In some cases we used the High Pure Plasmid Isolation Kit (Roche 11754785001) (Szinay et al., 2008). BAC DNA was labelled by nick translation following the manufacturer’s protocol (Roche (http://www.roche.com). The following direct labelling systems with dXTPs were used: Cy3-dUTP (Amersham, http://www.gelifesciences.com/), Cy3.5-dCTP (Amersham) and Diethylaminocoumarin-5-dUTP (DEAC; Perkin Elmer, http://www.perkinelmer.com), and two indirect labelling systems of dUTPs labelled with biotin and digoxigenin, respectively. For painting BACs on chromosome 12, we used Cot-100 DNA to block labelled repetitive sequences in BAC H146I19 from hybridization (Peterson et al., 1998; Chang et al., 2008). Cot-100 DNA was isolated according to Szinay et al. (2008).

FISH procedure and data analysis

FISH was performed according to the protocols of Rens et al. (2006) and Szinay et al. (2008) with the following modifications. Hybridization was carried out for 2 or 3 d, followed by a post-hybridization wash in a series from 82% to 64% formamide at 42°C (Schwarzacher & Heslop-Harrison, 2000) for 3 × 5 min. The biotin-labelled probes were amplified three times for 45 min with streptavidin conjugated with Cy5 (Invitrogen, http://www.invitrogen.com/site/us/en/home.html) and biotinylated-anti-streptavidin (Vector laboratories, http://www.vectorlabs.com/). The digoxigenin-labelled probes were amplified twice with anti-digoxigenin-FITC (Roche, http://www.roche.com) and anti-sheep-FITC (Invitrogen, http://www.invitrogen.com/site/us/en/home.html). Microscopy and FISH data interpretation were carried out as described by Szinay et al. (2008).

Phylogenetic analysis

The order of the BAC FISH signals was established on selected chromosome arms (Fig. 1), except for chromosome 5 and 12 on S. melongena. Doganlar et al. (2002) pointed out that the short arms of chromosome 5 and 12 of tomato translocated to chromosome 10 of S. melongena. We thus used missing values for BACs on chromosome 5 and 12 on S. melongena. In addition, BAC H037D07 on chromosome 7 and RH162O21 on chromosome 11 were not included in the phylogenetic analysis as a result of their insufficient hybridization on S. etuberosum and S. melongena.

We explored a number of coding strategies and methods of analysis. We used SoRT2 (Huang et al., 2010) to infer a phylogenetic tree based on pairwise genome rearrangement distances. Treeview was used to visualize the output files of SoRT2 (text files in Newick format) and to root the tree with S. melongena. In another approach, we coded different BAC-orders as (unordered) character states for each chromosome and performed a maximum parsimony (MP) analysis with PAUP v4.0 (Swofford, 1999) (MP, chromosome coding). In a third approach, we adopted the method of Müller et al. (2003) to derive discrete characters by coding the presence or absence of adjacent chromosomal segments in a binary data matrix. A MP analysis on this data matrix was performed in PAUP v4.0 (Swofford, 1999) and Jackknife analyses were performed with 10 000 replicates (MP, segment coding). The most parsimonous evolutionary history of the events on the investigated chromosome arms was reconstructed and illustrated as a phylogenetic tree in Fig. 4 (left panel).

The topology of our phylogenetic reconstruction was compared with a generalized tree derived from published phylogenies of a similar set of materials (Spooner et al., 1993, 2005; Peralta & Spooner, 2001; Rodriguez et al., 2009). The material investigated in those studies included representatives of section Etuberosum (usually S. etuberosum, sometimes also S. palustre), section Petota (represented by a varying number of species), section Lycopersicoides (with S. lycopersicoides and S. sitiens), section Juglandifolia (S. juglandifolium and S. ochranthum) and section Lycopersicon (S. lycopersicum and related species). The trees in the papers by Peralta & Spooner (2001, Fig. 4), Spooner et al. (2005, Fig. 7) and Rodriguez et al. (2009, Fig. 4) show an identical topology, while the tree in the paper from Spooner et al. (1993) deviates slightly in placing S. sitiens (section Lycopersicoides) closest to S. ochranthum (section Juglandifolia), separated from the other species of section Lycopersicoides and S. juglandifolium. Although these branches have bootstrap support of only 51% and 52%, we give an informal ‘consensus topology’ in Fig. 4 (right panel).

Results

Chromosomal rearrangements revealed by FISH

For establishing chromosomal rearrangements in the Solanum species we selected BACs on seven chromosome arms (Figs 1, 2) that were known from comparative genetic studies of Solanum crops to contain inversions (Table 1). We use here the term ‘syntenic species’ to indicate those species with identical BAC-FISH patterns on all studied chromosome arms (Figs 2, 3). The first group of syntenic species (syntenic sp. A) includes potato (S. tuberosum) and its wild relatives S. bulbocastanum, S. tarijense, S. megistacrolobum and S. pinnatisectum; the second one (syntenic sp. B) comprises tomato (S. lycopersicum) and its wild relatives S. peruvianum, S. habrochaites and S. pimpinellifolium. These two syntenic species groups show different BAC orders on chromosomes 5, 6, 9, 10, 11 and 12 (see Fig. 3 for schematic representations, namely 5a, 6a, 9c, 10b, 11a and 12a).

Figure 2.

BAC-FISH images showing an overview of BACs on pachytene of selected chromosomes. The leftmost column shows information on the chromosome arms with their number followed by S (short arm) or L (long arm). The other columns show BAC FISH images on chromosomes of Solanum melongena (S.m), S. tuberosum (P), S. etuberosum (S.e), S. ochranthum (S.o), S. lycopersicoides (S.ly), S. pennellii (S.p), S. lycopersicum (T, represented by Heinz 1706), S. chilense (S.c). BAC names and their hybridization images on a specific chromosome arm are labelled with different colours. There are no images for 5S and 12S of S. melongena due to hybridization difficulties.

Figure 3.

Schematic overview of chromosomal inversions in Solanum. Arrows indicate likely directions of inversions. Below the arrows the inversion events are numbered with a small letter in alphabetic order distinguishing the different inversions within one chromosome. Lines indicate inversions with unknown directions. Question marks are weak or missing FISH signals. Signals of BACs are shown by coloured dots.

In comparison to syntenic sp. B: two proximal BACs display an inverted order on chromosome 12S (S, short arm) of S. chilense (Fig. 2g; 12b in Fig. 3); all BACs on 6S are inverted in S. ochranthum (Fig. 2b, 6a in Fig. 3); two distal BACs on 6S (Fig. 2b; 6b in Fig. 3); and 7S (Fig. 2c; 7b in Fig. 3) is inverted in S. lycopersicoides and S. pennellii. Between S. lycopersicoides and S. pennellii, three proximal BACs on 6S are inverted (Fig. 2b; 6c in Fig. 3). BAC orders in S. etuberosum are very similar to that in syntenic sp. A, except for inverted orders of BACs on 7S, 9S and 10L (L, long arm) (Fig. 2c–e; 7c, 9b and 10b in Fig. 3). On chromosome 10L, the BAC order in S. etuberosum is similar to the order in syntenic sp. B.

We could not obtain interpretable FISH signals for chromosome arms 5S and 12S of S. melongena, which is presumably caused by translocations involving these chromosome arms (Doganlar et al., 2002; Wu et al., 2009a,b). In comparison to all other species, BACs on 7S, 9S and 10L are inverted in S. melongena (Fig. 2c–e; 7a, 9a and 10a in Fig. 3). The order of BACs on 7S is similar to both S. lycopersicoides and S. pennellii (Fig. 3). Interestingly, BAC H251G05 showed double signals on 6S of S. melongena (Figs 2b, 3), which can be interpreted as breakpoints in the chromosomal target of this BAC.

Phylogenetic analysis

The trees we made differ somewhat depending on the methods used (datasets and resulting trees are presented as Supporting Information Figs S1–S3; Tables S1–S4), but a general pattern is clear: S. etuberosum and the clade (polytomy) including S. pinnatisectum, S. bulbocastanum, S. megistracrolobum, S. tarijense and S. tuberosum (syntenic sp. A) are placed basal in all trees. The remaining species are joined in a clade in all trees, in which all identify a clade joining S. pennellii and S. lycopersicoides, a branch with S. ochranthrum, and a branch including S. peruvianum, S. habrochaites, S. pimpinellifolium and S. lycopersicum (syntenic sp. B).

Distance-based approaches (SoRT2 by Huang et al. (2010) in our study: Fig. S1; Table S1) rely on overall similarity instead of (syn-)apomorphic character states, which make us favour the trees based on MP. The most conservative (and least resolved) tree of our two parsimony approaches results from a simple coding method in which specific BAC-orders on chromosomes are considered as (unordered) character states (Fig. S2; Table S2). In another approach (proposed by Müller et al., 2003), the presence and absence of adjacent loci are used as characters, leading to decidedly higher resolution, as well as high jackknife values (82–98%) for all clades except for the combination of the S. lycopersicoides/S. pennellii clade with the tomato clade, which received only 74% jackknife support (Fig. S3; Tables S3, S4). This method, however, considers the adjacent segments as independent characters which they formally are not (i.e. one inversion typically leads to two new characters). Because of this, we will use the simplest (and most conservative) MP method (based on BAC-orders) as base for our phylogenetic reconstruction. Under this method a character state can change to any other state with equal chance.

Phylogenetic reconstruction

After deriving a phylogenetic tree using the conservative MP method based on BAC-patterns (Fig. S2), we used this tree as a starting point to resolve the remaining polytomies by the analyses described below. The most likely way in which the karyotypes of the studied Solanum species evolved are presented in Fig. 3. By inferring ancestral and derived states with S. melongena as outgroup, we could place inversion events on the branches of our phylogenetic reconstruction (Fig. 4, left panel). This reconstruction is identical to the tree produced by MP tree based on segment coding (after Müller et al., 2003).

Figure 4.

The topology of our phylogenetic reconstruction (left panel) in comparison with a generalized tree (right panel) derived from published phylogenies of a comparable set of material (see the Materials and Methods section; Spooner et al., 1993; Peralta & Spooner, 2001; Spooner et al., 2005; Rodriguez et al., 2009). The polytomy within the potato group in the generalized tree only indicates the lack of information about the relationships among these species in the used references. Red lines indicate the different positions of Solanum lycopersicoides, S. ochranthum and S. etuberosum. Inversion events that were numbered in Fig. 3 are indicated on the tree (left panel).

The ancestral state for chromosome 5 could not be established because the order of BACs in the outgroup could not be determined. Syntenic sp. A and S. etuberosum show a similar order of BACs, which differs from all other species that have the four distal BACs inverted. Recent literature (e.g. Peralta et al., 2008) considers the potatoes (section Petota) as sisters to the tomatoes s.l. (sections Lycopersicoides, Juglandifolia and Lycopersicon) and section Etuberosum as sisters to the combined group of potatoes and tomatoes s.l. Both S. etuberosum and section Petota can thus be considered basal compared to the tomatoes s. str. (section Lycopersicon). This leads us to consider the distal inversion (5a in Fig. 3) as a synapomorphy for the clade including the remaining species from S. ochranthum onward.

The ancestral BAC order for chromosome 6 is similar to our outgroup, as potatoes (syntenic sp. A) and S. melongena share the same order. However, there is a difference between the groups because the middle BAC shows two foci in S. melongena. This may point to a duplication/deletion event (a duplication in S. melongena or deletion in our ingroup) or a small interstitial inversion in which the chromosome breakpoint lies within the used BAC. Because S. ochranthum shows a similar BAC order to that of the potato clade and S. etuberosum, we consider S. ochranthum as the most basal of the wild tomatoes (and as such, we place S. ochranthum as a sister to the clade including S. pennellii, S. lycopersicoides and S. lycopersicum,Fig. 4). Looking at the remainder of species, the most parsimonious solution is to assume a whole arm euchromatin inversion giving rise to the order found in syntenic sp. B and S. chilense. A distal inversion then joins S. pennellii and S. lycopersicoides in a clade, and a proximal inversion separates S. lycopersicoides from S. pennellii. This hypothesis (placing S. pennellii and S. lycopersicoides as sister taxa) contrasts with all existing phylogenetic reconstructions as proposed by Spooner and collaborators (Spooner et al., 1993, 2005; Peralta & Spooner, 2001; Rodriguez et al., 2009) in which S. lycopersicoides is usually considered (together with S. sitiens) to be sister to a clade consisting of the species of section Juglandifolia and section Lycopersicon (including S. pennellii). If we follow previously proposed phylogenies, we must assume a complex rearrangement (in which the S. lycopersicoides BAC order is generated directly from the ancestral S. melongena/potato type), followed by two consecutive inversions giving rise to S. pennellii and later to syntenic sp. B. We here favour the simple inversion scenario for two reasons: first, inversions are more common than complex rearrangements that involve multiple breaks, as all inversions we found can be interpreted as simple inversions; and second, the top inversion on 7S supports the sister group relationship between S. pennellii and S. lycopersicoides.

The ancestral karyotype for chromosome 7 is easily established by the fact that S. melongena and Capsicum annuum (data not shown) share the same order of the BACs involved. Supporting evidence for this comes from comparative genomics based on COS-markers which also indicates that S. melongena is collinear with C. annuum (Doganlar et al., 2002). A top inversion of the most distal BACs led to the BAC order found in most species. Three species show another order of BACs: S. etuberosum has the whole arm inverted, and S. pennellii and S. lycopersicoides have the ancestral BAC order. The most parsimonious explanation is based on the assumption that a reversal took place in these two species, joining these two species together in a clade separate from all other species. As mentioned in the Results section on chromosome 6, this concurs well with the most parsimonious solution for events that happened for 6S. If we assume that S. lycopersicoides and S. pennellii have the ancestral karyotype, we would introduce various inconsistencies to our tree: either various incidences of parallel (convergent) evolution in three independent lines, or the placement of S. pennellii and S. lycopersicoides in a position basal to the potato clade, which is very unlikely given all other information (and introducing many conflicts to the tree).

The evolutionary scenario on chromosome 9 is straightforward. As Capsicum annuum and eggplant are collinear (Wu et al., 2009a,b; Wu & Tanksley, 2010), we may assume that S. melongena represents the ancestral type. A top inversion gave rise to the BAC order found in most species (9a in Fig. 3). A subsequent interstitial inversion of three BACs gave rise to the S. etuberosum type, whereas a proximal inversion is unique to S. chilense.

Chromosome 10L provides a unique insight into the position of S. etuberosum with respect to the potato clade. Solanum etuberosum is usually placed basal to a combined potato–tomato clade (Fig. 4, right panel). All investigated species (except for our outgroup S. melongena) have the same (green) BAC placed proximally, suggesting that this is ancestral for the whole group. The question then is to determine what type (potato type or the type of the other species is ancestral). The use of S. melongena as outgroup for chromosome 10 must be done with caution, as previous research showed that S. melongena experienced extensive rearrangements on chromosome 10. Eggplant 10L is a mosaic of tomato chromosomes 5, 10 and 12 (Wu et al., 2009a). The S. melongena type is nevertheless very close to the potato type, because it can be obtained by a single whole arm inversion from the potato type, as was also shown by the analysis by Wu et al. (2009a). From Wu et al. (2009b) and Wu & Tanksley (2010) it can be inferred that Capsicum annuum chromosome 10 has a similar order to the potato. This substantiates the hypothesis of the potato/Capsicum-type representing the ancestral state, and the tomato/S. etuberosum type being derived. We therefore place syntenic sp. A basal to S. etuberosum, which in turn becomes the sister of a group including all other species, including tomato.

The ancestral state on chromosome 11 is represented by S. melongena, because it shares its BAC order with syntenic sp. A and S. etuberosum. The inversion of the four most distal BACs joins S. ochranthum with the tomato and other species. Rearrangements on chromosome 12S are straightforwardly placed on the tree. Although due to translocations we have no data on eggplant, the proximal position of the green BAC is ancestral (i.e. present in all species, except S. chilense). If we are right in assuming that syntenic sp. A and S. etuberosum should be placed basally to the other investigated species, inversion 12a (Fig. 3) is a shared derived character for S. ochranthum and its sister clade (including S. chilense).

Discussion

The aim of our study is to use chromosomal rearrangements as phylogenetic markers for the study of chromosome evolution in the Solanaceae family. We show that the hybridization of BACs on the chromosomes of crops and their related species enables us to confirm directly the genomic collinearity, to show inversions between their homeologues, and to reconstruct the most likely way in which the karyotypes of the studied species evolved.

Comparison to published phylogenies

The reconstruction of Solanum phylogeny has a long history, in which the proposed interrelationships between potato, tomato and their close relatives have changed many times when new data became available. To compare the topology of our reconstructed tree (Fig. 4, left panel) with existing data, we constructed a generalized tree derived from published phylogenies of a comparable set of material (Spooner et al., 1993, 2005; Peralta & Spooner, 2001; Rodriguez et al., 2009). In this generalized tree (Fig. 4, right panel), the following sistergroup relationships are present: after the outgroup consecutive branches lead to the representatives of sections Etuberosum, Petota, Lycopersicoides, Juglandifolia and Lycopersicon. We note three remarkable differences between our tree and the generalized tree: (1) the position of S. etuberosum, (2) the position of S. ochranthum, and (3) the presence of a clade containing S. pennellii and S. lycopersicoides. Based on the inversion in 10L that S. etuberosum shares with all species higher up in the tree, the clade containing potato and its wild relatives (section Petota) is placed basally to S. etuberosum in our tree, whereas the reverse is suggested by all earlier studies. The basal placement of S. ochranthum with respect to S. lycopersicoides is a consequence of the order of BACs on chromosome 6. The grouping of S. pennellii in a clade with S. lycopersicoides is supported by S. pennellii and S. lycopersicoides sharing an inversion on chromosome 7 and a presumed additional inversion on chromosome 6. This poses a strong contrast with almost all previously published trees where S. pennellii is firmly nested in a ‘tomato group’ (Peralta & Spooner, 2001; Spooner et al., 2005; Rodriguez et al., 2009), with one exception where S. pennellii was suggested to be closest to S. lycopersicoides (Zuriaga et al., 2009).

Our evolutionary hypothesis is exactly similar to the MP-segments tree that we generated following the method of Müller et al. (2003), and shows significant differences with the phylogenetic trees as previously published. Notably, our data are not the first chromosome data that challenge evolutionary relationships proposed earlier: a recent study on pairing configurations in spreads of synaptonemal complexes of tomato × wild species hybrids (Anderson et al., 2010) casts doubt on the proposed phylogeny based on chromosome synteny. In that study S. pennellii and S. habrochaites (two species that are sister according to Rodriguez et al. (2009)) showed remarkably high levels of pairing irregularities, suggesting that both were equally far apart from one another as they are from tomato, which was thought to be more distantly related.

We also note that the data we obtained for chromosome 10 are not entirely in concordance with previously published data. Compared with the results reported by Doganlar et al. (2002) and Wu et al. (2009a), our data showed a reversed orientation of BACs on S. melongena for the whole arm. We identified a large inversion comprising most of the euchromatic part between S. lycopersicoides and S. tuberosum; while Pertuzéet al. (2002) described marker synteny between these two species. In addition, we found that S. lycopersicum and S. lycopersicoides are collinear, while Chetelat et al. (2000) suggested chromosomal rearrangements between these two species. A possible explanation for all of these observed differences is that different accessions were used in the different studies and that chromosomal rearrangements exist among accessions within one species. Another possibility is that the genetic studies misinterpreted the inversions due to low marker density or suppression of recombination.

Previous trees were based on extensive datasets using a multitude of markers, sequences and morphological traits. How can we best explain these apparent inconsistencies? Chromosomal rearrangements behave no differently from other markers, and their correct interpretation might be obscured by similar ‘noise’ to which all characters are subjected such as homoplasy (parallel evolution or reversals) or complex modes of speciation. The possibility of interspecific hybridization or incomplete lineage sorting should surely be considered when studying the evolutionary history of Solanum, especially in the light of the documented crossability between various investigated species. Interspecific hybridization followed by backcrosses to either one of the parents could lead to introgression of new alleles into a recipient species. In case of inversions, there is the possibility of introgression of whole inversions into a new background, becuase recombination will unlikely happen in small inverted segments during meiosis (Verlaan et al., 2011). The possibility of hybridization followed by introgression of inverted segments could account for, for example, the small distal inversions on 6S and 7S that join S. lycopersicoides and S. pennellii in a clade separate from other species.

Notes on rearrangements

It is remarkable that 75% (12 out of 15) of the inversions we studied involved the most distally placed BAC on chromosome arms. It is unknown whether these inversions are the result of complete arm inversions (i.e. including the telomere) or whether there was a breakpoint in the sub-telomere heterochromatin.

One inversion on chromosome 7S is remarkable, as it represents the only character reversal. Given our observation that 75% of inversions involve the distal-most BAC, the occurrence of a distal reversal might not be entirely unpredicted: any breakpoint in a similar interval as the inversion preceding it, will in the far majority of inversions result in a reversal. We also note that the reuse of chromosomal breakpoints has frequently been documented in other species (Wu & Tanksley, 2010).

When the placement of rearrangements on the phylogenetic tree is considered, inversion events appear to cluster on the branch leading to S. ochranthum and its sister group. Such a clustering could point to an evolutionary ‘bottleneck’ in which a small population fixed a number of inversions, radiating into many species at a later point. Alternatively, the presence of many rearrangements could be taken to indicate the passage of a long time period (assuming something like a constant ‘inversion rate’). The absence of rearrangements within the potato clade among the studied wild and cultivated potato species suggests that these represent relatively recent splits. This is not surprising in the light of recent publications (Jacobs et al., 2008, 2011). The split between the close relatives of tomato (i.e. commonly referred to as the section Lycopersicon) shows considerably more variation, suggesting that these species might have started diverging earlier than the potatoes.

Conclusions

Our study features advances in molecular cytogenetic tools that support plant genetics, genomics and breeding programmes. In general, our data support the current grouping of Solanum species into different sections: most previously defined sections are identified herein as having an unique order of chromosomal segments. There are nevertheless a few remarkable differences in our phylogenetic reconstruction compared to earlier studies. The apparent conflict between our hypothesis and previously proposed hypotheses points to the possibility that the evolutionary history of Solanum may have seen its share of reticulation: interspecific hybridization followed by introgression of inverted segments. Alternatively, incomplete lineage sorting may have played a role in the apparent complexity of Solanum phylogeny.

The progress of more and better sequencing practices will reveal novel and detailed information on chromosome rearrangements among species. The occurrence of chromosomal rearrangements stresses the important of a correct physical map in ordering scaffolds of related species that are being re-sequenced. The cross-species BAC-FISH method as presented here will remain an indispensible tool in comparative genomics as it reveals chromosomal rearrangements without a priori de novo sequencing of the species involved. In addition, cross-species BAC-FISH can detect chromosomal rearrangements involving heterochromatin areas and vice versa, and so may shed light on possible epigenetic changes in the chromosome regions under study. Finally, cross-species BAC-FISH is a powerful instrument for the detection of pairing failure between introgressed homeologous regions thus explaining the absence of crossovers, a phenomenon known as linkage drag (e.g. Verlaan et al., 2011).

Acknowledgements

The authors would like to thank Bert Essenstam, Michel Hoogendoorn, Fien Meijer-Dekens, Dirk Jan Huigen, Jack Vossen and Jan de Boer for their help in arranging plant materials and potato BACs. We also acknowledge our colleagues of the Centre for Genetic Resources, the Netherlands (CGN), Dr. Gerard M. van der Weerden of the Solanum collection of the Botanical and Experimental Garden of Nijmegen University, and Dr. Dörthe Dräger, RijkZwaan B.V. de Lier, the Netherlands for providing various Solanum species and accessions.

Ancillary