In vitro regeneration of cultivated tomato (Solanum lycopersicum L.) has been a constant subject of research because of the commercial value of the crop. Consequently, numerous studies on plant regeneration from a wide range of tissues and organs of wild and cultivated tomato germplasm have been published [1]. These studies demonstrate that organogenesis, the common tomato regeneration pathway, is strongly influenced by genotype as well as by several physical and chemical factors. These reports also document the existence of recalcitrance (partial or total inability to respond to in vitro culture), which greatly limits biotechnological breeding. High regeneration is crucial to the success of techniques such as haploid regeneration, genetic transformation, propagation, somatic hybridization, mutation selection and germplasm storage [2, 3]. For example, the low efficiency of tomato transformation has been associated with the low regeneration potential of the cultivars used [4, 5]. In addition, in some cultivars, buds may be induced but do not develop into shoots [6]. In order to increase regeneration ability in low regenerating tomato cultivars, several introgression programs have been documented [7–10].

The process of in vitro shoot organogenesis usually involves a hormonal response of somatic cells, the dedifferentiation of differentiated cells in order to acquire organogenic competence, cell division of the responding cell(s) and initiation and development of new shoots from the newly dividing cell(s), either directly or indirectly through a callus stage [11, 12]. Thus, many genes may be involved at different steps of this complex process. For instance, the cdc2 gene expression, which encodes p34, a key cell cycle regulator, has been proposed as an indicator of the state of competence to divide [13]. Genes that encode or regulate cytokinins and auxin may clearly influence regeneration. Both types of growth regulators act synergistically to promote cell division and antagonistically to promote shoot and root initiation from callus cultures [14]. In Arabidopsis, a Histidine Kinase (AHK) gene that encodes a cytokinin receptor (CRE1/AHK4) has been identified [15, 16] and linked, like other AHKs, to cell division and regulation [17]. With regard to the initiation of shoot formation, the most characterized gene reported is ESR1, which confers, when overexpressed, cytokinin-independent shoot formation in Arabidopsis root explants [18]. ESR1 encodes a transcription factor belonging to the ethylene-responsive factor (ERF) family and is classified in subgroup VIII-b. The ESR2 gene that encodes a protein that is very similar to ESR1 appears to have redundant functions that regulate shoot regeneration [19]. The expression patterns of other Arabidopsis ERF VIII-b subgroup genes may also be involved in early events of shoot regeneration [20].

Genetic analysis of regeneration in tomato suggests that dominant alleles determine high regeneration capacity [7, 21–24]. However, there is no consensus about the number of genes involved. For instance, Koorneef et al. [25] obtained regeneration segregation ratios in accordance with either a monogenic, digenic or trigenic model depending on the tester tomato line, despite the fact that none of the lines themselves were able to regenerate shoots from root explants. In this study, a dominant allele of S. peruvianum L. (Rg-1), which determines efficient shoot regeneration in tomato root explants, was mapped near the middle of chromosome 3. In addition, a putative allele of Rg-1 from S. chilense (Dunal) Reiche (Rg-2) was reported by Takashina et al. [9] and Satoh et al. [22]. Both alleles may act in combination with other alleles of either tomato or the wild relatives S. peruvianum or S. chilense [22, 25]. On the other hand, Torelli et al. [26] identified a cDNA by mRNA-differential display that corresponded to the LESK1 gene and whose expression is specifically and transiently enhanced by the exposure to the hormonal treatment leading to caulogenesis (shoot induction). This gene encodes a putative serine-threonine kinase and has been reported as an in vitro caulogenesis marker in tomato [27, 28].

Despite ongoing research into the genetic control of in vitro culture traits in tomato and other crops, there is still not enough information regarding which key genes are responsible for low or high regeneration ability, nor even the number of genes involved. The study and characterization of the reported genes and others that might be identified could greatly improve our understanding of the molecular mechanism underlying the different phases of tomato in vitro organogenesis. In the present study, we developed two mapping populations (F₂ and BC₁) from S. lycopersicum (as the recurrent parent) and S. pennellii Correll (as the regenerating parent) and conducted a QTL-based analysis. We hereby report the identification of six QTLs on five chromosomes. These QTLs present high significant LOD scores and together represent a high percentage of phenotypic variance. We also report markers associated with QTL peaks. In addition, we located two candidate genes, Rg-2 and LESK1, and performed a preliminary search for genes situated at QTL peaks. Our findings will complement the current knowledge of the genetics of regeneration and facilitate the development of molecular markers for use in tomato breeding and gene isolation.

The wild tomato species S. peruvianum, S. pimpinellifolium L. and S. chilense were used as sources of regeneration genes in order to study the genetics of the in vitro regeneration in tomato [7, 9, 21]. In this study, we used one accession of S. pennellii (PE-47) as the high regeneration parent [29]. This accession, along with a previously selected low regenerating tomato cultivar (cv. Anl27), was used to develop two mapping populations (F₂ and BC₁). The use of the introgression lines of S. pennellii in the M82 tomato background [30] had been previously ruled out for this analysis because of the high regeneration ability of both parent lines (data not shown). Data in Figure 1 and Table 1 seem to indicate complete dominance for B, partial dominance for R and additive effects for PR. This is in agreement with other reported studies on tomato where dominance, to different degrees, depending on the regeneration trait studied, was also reported [21, 22, 24, 25]. B and R traits show a high correlation in both populations, suggesting that common or linked genes control these traits. The correlation between PR and both B and R was lower. This could imply that other genes may be influencing the PR trait and/or variations between different biological samples are higher in PR (for instance, competition for development due to the presence of different shoots in a similar explant area). Thus, the low sample size may be also a possible explanation for the lower correlation.

Some descendants in the F₂ population showed phenotypes for the PR trait that are more extreme that those shown by the regenerating parent line (Figure 1). Transgressive segregation has already been described in other reports in relation to the genetic control of plant regeneration [31–33], and suggests poligenic inheritance [34]. It also suggests the existence of alleles that promote, and others that inhibit, in vitro regeneration, with only some of the alleles with positive effects occurring in the same parent [34]. In fact, in this study, the SlRg-8 QTL that contributes to regeneration came from the low regenerating parent.

Plant regeneration from cultured tissues is assumed to fall under quantitative genetics [34], although evidence in tomato [22, 25] and other vegetables [35–37] indicates that just a few genes could be responsible for regeneration. We identified 6 QTLs in the BC₁ analysis, which is indicative of the participation of a large number of genes in this character. These QTLs are situated on chromosomes 1, 3, 4, 7 and 8 (Figure 2b). The percentage of variance explained by each QTL ranges from 7.4 to 27%, which is in accordance with the most common range (6-26%) reported in the genetic mapping of QTLs for tissue culture response in plants [34]. We used three traits (B, R and PR) as a measurement of regeneration capability that could be useful for detecting chromosome regions that act at different times.

In the F₂ population, only the QTL of chromosome 7 was identified for all analyzed traits (Additional File 3); the SDMs observed in most chromosome areas where QTLs were detected in the BC₁ population are most likely the cause (Figure 2). The SDMs on chromosomes 1, 3, 6, 10 and 11 were also observed in similar areas in the Tomato-EXPEN 2000 map [38]. SDMs affect the detection power of QTLs when QTLs and SDMs are closely linked [39], as occurred in our case. Deviation from the expected segregation ratio is a common feature of inter-specific tomato crosses [40]. To wit: in a F₂ population from S. lycopersicum x S. pennellii, De Vicente and Tanskley [41] reported a skewness rate of up to 80%.

In the BC1 population, three QTLs were detected for B: SpRg-1, SpRg-3 and SlRg-8. These QTLs may be associated with the first stages of regeneration, that is, hormonal induction response and bud formation. SpRg-1, which explained the highest percentage of variation for B (23.9%), was also identified for the R and PR traits. Given that bud formation is a necessary prerequisite for the production of shoots, it was expected that this major QTL for B would be found for R and PR, which in fact turned out to be the case (Table 2). For R and PR, a common QTL on chromosome 7 (SpRg-7) was also identified. In addition, two QTLs were detected for R (SpRg-4a and SlRg-8) and PR (SpRg-4b and Rg-3). All these QTLs seem to be involved in the development of buds into shoots. As can also be observed in this study, common QTLs for the different regeneration traits, as well as a higher number of QTLs for traits related to plant development compared to those associated with bud induction, have been reported in different studies [42, 43]. For instance, in Arabidopsis, Schianterelli et al. [43] found a common area of chromosome 1 in all analyzed parameters, a peak in chromosome 4 and another in chromosome 5 when they analyzed the total number of regenerated shoots. In wheat, Ben Amer et al. [42] identified three QTLs, two that affect green spot initiation and shoot regeneration and a third that only influences plant formation.

A partial common genetic system controlling the regeneration frequency of diverse types of explants has been reported by Molina and Nuez [36] in melon. This indicates that using different explants for loci detection may lead to the identification of some common QTLs, but also to the possible identification of other new QTLs. Root explants were used by Koornneef et al. [25] and Satoh et al. [22] for phenotyping, at which point two alleles for regeneration ability were located on chromosome 3 of tomato. In the present study, leaves were used for phenotyping and a QTL (Rg-3) in a similar area of chromosome 3 was detected in addition to other QTLs that influence regeneration and were identified on chromosomes 1, 4, 7 and 8. Differences in root and leaf explants for QTL identification were also found in Arabidopsis thaliana [43].

Koornneef et al. [25] located a dominant allele from S. peruvianum (Rg-1) near the middle of chromosome 3 that determines efficient shoot regeneration in tomato root explants. Satoh et al. [22] mapped a putative allele (Rg-2) from S. chilense on this chromosome. The acid invertase gene, reported as a marker linked to Rg-2, was chosen for mapping Rg-2 in our population derived from S. pennellii. The polymorphisms detected in our parents allow us to map this gene in the QTL detected on chromosome 3 that we named Rg-3. We consider Rg-3 to be a putative allele of the Rg-2 gene. Allelism must be confirmed.

The other gene chosen as a candidate was LESK1, which encodes a serine/threonine kinase, and was reported as a marker of competence for in vitro regeneration in tomato [27, 28]. This gene was positioned on chromosome 3, but it is not located in the Rg-3 QTL.

The recent release of the entire genome sequence of tomato provides a powerful tool for interrogating QTL data. In this respect, we have taken a preliminary look at genes located at the peak areas of the detected QTLs, and which could be related to organogenesis. Histidine kinases were reported as cytokinin receptors [15–17]. In our QTL peaks, only one histidine kinase is located in the SpRg-7 QTL. The candidate tomato gene, LESK1, which has been described as a marker for in vitro competence, encodes a serine/threonine kinase. We looked for serine/threonine kinases and found this kind of protein in all identified QTLs. Other putative candidate genes could be ESR1 and its paralogue, ESR2, from Arabidopsis, which are the best-characterized genes related to regeneration [18, 19]. These genes code for ethylene response factors (ERF). We found ERFs, which contain the AP2 domain, in all analysed QTLs with the exception of SpRg-4b. Cyclines related to cell division [13] were found in SpRg-1, Rg-3, SpRg-4a and SpRg-7. MADS-box genes, which have been correlated to adventitious regeneration induction and regulation [44, 45], were found in the SpRg-1, SpRg-4a and SpRg-7 QTL peaks.

The results obtained in this study may very well represent a significant step toward the goal of understanding the processes underlying tomato tissue culture and regeneration responses. We have situated six QTLs on chromosomes 1, 3, 4, 7 and 8, five from S. pennellii and one from S. lycopersicum. The most important QTLs are SpRg-1, which is most likely associated with the morphogenetic response, and SpRg-7, which promotes bud development. A QTL detected on chromosome 3, Rg-3, likely contains a putative allele of the Rg-1 and Rg-2 genes, as is shown by mapping the acid invertase gene linked to Rg-2. QTLs detected on chromosomes 8 and 4 most likely contain genes influencing bud formation and development, respectively.