These favorable wild QTL alleles are a valuable resource for breeding programs after they have been fixed in NILs or ILs and after the superior performance of the line has been validated in comparison to the cultivated recurrent parent in replicated field experiments.
From the tomato AB-QTL populations, MAS was applied to develop NILs that contain specific QTL al-leles derived from the wild donors S. habrochaites, S. pimpinellifolium, and S. peruvianum, and that are able to significantly improve the performance of the elite variety (Tanksley etal. 1996; Bernacchi etal.
1998a, Monforte and Tanksley 2000a, b; Monforte et al. 2001; Yates et al. 2004). Bernacchi et al. (1998a) evaluated the agronomic performance of 23 NILs containing either S. habrochaites or S. pimpinellifolium in-trogressions, in five locations worldwide. This study revealed that 22 out of 25 (88%) of the quantitative factors exhibited, in at least one location, the phenotypic improvement that had been predicted in the previous QTL analysis of the BC3 populations (Bernacchi etal. 1998a). This indicated that most of the QTLs detected in the BC3 were not spurious and can be manipulated via MAS. However, the significance at which QTLs were detected in the BC3 families as well as the degree of conservation of QTL across locations, as detected in the initial QTL analysis, seemed to be only modest predictors for QTL significance and QTL x environment interactions observed in the derived NILs. This study also highlighted the high frequency with which unexpected phenotypes (both beneficial and detrimental), not predicted by the original QTL analysis, can be observed in the derived NILs. Possible explanations for these findings include linkage drag, pleiotropic effects or epistatic interactions that could not be detected in the BC3 mapping population perhaps because of the higher overall level of variance still present in these AB populations as compared to NILs.
What is particularly interesting is the magnitude of the improvements observed in the NILs compared to the control elite variety for many of the target traits (Bernacchi et al. 1998a). For example, per-location gains over the elite control (cv. E6203) ranged from 6 to 22% for soluble solids content; from 14 to 33% for fruit color; from 20 to 28% for total yield, from 15 to 48% for red yield, and from 9 to 59% for Brix° x red yield. The results obtained for soluble solids and yield can be compared to the historical genetic gains achieved for the same traits in processing tomatoes using conventional breeding strategies. For example, the per-location gains achieved for the trait Brix° x yield, after 4 years required for the first cycle of AB-QTL breeding using S. habrochaites as donor parent, ranged from 9 to 59%, resulting in yearly gains between 2.3 to 14.8%. For the same trait, the estimates of annual genetic gains obtained with conventional breeding were only of 0.9% and 1.5% for the Israeli and California data, respectively (Grandillo et al. 1999).
Given the amenable properties of introgression lines and the potential of exotic germplasm as a source of valuable genetic variation for the improvement of complex traits, Zamir (2001) proposed investing in the development of a genetic infrastructure of "exotic libraries" in order to enhance the rate of progress of in-trogression breeding based on wild species resources. An exotic library consists of a set of ILs, each of which carries a single, possibly homozygous, marker-defined chromosomal segment that originates from a donor exotic parent, in an otherwise homogeneous elite genetic background. The set of ILs would together represent the entire donor genome. Initially, one disadvantage of this type of population was the long time required for their development; however, the availability of numerous marker-screening technologies has now made the construction of such libraries a more efficient process that can be completed after ten generations of crossing and marker analysis (Young 1999).
In tomato, the first exotic library with genome wide coverage was developed by Eshed and Zamir (1995) from a cross between the wild green-fruited species S. pennellii (LA716) and the cultivated tomato S. lycopersicum (cv. M82). The library consisted of 50 RFLP-defined ILs and allowed the identification of yield-associated QTLs (Eshed and Zamir 1995). The results of that study highlighted the higher QTL mapping power of IL populations compared with conventional segregating populations such as F2, BCi, or RILs (Zamir and Eshed 1998). For example, while Eshed and Zamir (1995) detected a minimum of 18 and 23 QTLs for fruit size and Brix°, respectively, only a maximum of seven and four QTLs, respectively, were detected for the same traits when using standard mapping populations.
The efficiency of these libraries of ILs in detecting and mapping QTLs underlying traits of agronomic importance is due to the near-isogenic nature of the lines, such that any phenotypic difference between the cultivated recurrent parent and an IL, or the hybrid of the recurrent parent with an IL (ILH) is attributable only to the exotic parent genomic segment. This increases the ability to statistically identify smaller effect QTLs, as the "overshadowing effect" of independent major QTL is eliminated. Since the lines are identical to the recurrent parent, except for the single introgressed segment, their phenotypes generally resemble that of the cultivated parent. This reduces sterility and other problems that occur in breeding-population structures characterized by a high frequency of the exotic parent genome, and allows the lines to be evaluated for yield-associated traits. Another advantage of using IL libraries for QTL analysis is the simplicity of the statistical procedure used to detect QTL, which is based on the comparison of each IL with the recurrent parent for the trait of interest, and is therefore less affected by the issue of experiment-wide error. In addition, the epistatic effects that are mediated by other regions of the donor genome, with the exception of the loci contained in the same introgression line, are eliminated. Finally, the permanent nature of the IL libraries allows testing the phenotypic value of each introgression on multiple replicates. Replicated trials of the same line can be analyzed in different years and/or environments, which enable the estimation of the extent of QTL by environment interactions (Monforte et al. 2001; Liu et al. 2003; Gur and Zamir 2004). This provides more accurate estimates of the mean phenotypic values and increases the power of QTL detection. The permanent nature of these lines also allows several laboratories to collect data for different traits on the same lines, thereby facilitating the integration of data from independent studies and the creation of a comprehensive phenotypic database for general access (Zamir 2001).
The map resolution of a library of ILs is defined by the overlap between contiguous segments (bins) to which genes or QTL can be assigned by juxtaposing the lines (Pan et al. 2000; Liu et al. 2003). Bin lengths vary across the genome, depending on the number, length, and overlap of adjacent segments. Generally the map resolution of the initial ILs contributing to an IL library is relatively low; however, the ILs represent the starting point to develop appropriate sub-ILs that will allow the phenotypic effects of QTL to be fine-mapped to smaller intervals (Paterson et al. 1990).
In order to increase the mapping resolution of the S. pennellii exotic library, an additional 26 sub-ILs have been added, resulting in 76 lines which partition the entire genetic map into 107 bins defined by singular or overlapping segments (Pan et al. 2000; Liu et al. 2003). Over the past decade the 76 ILs and their hybrids have been evaluated for 20 different yield-associated, fruit morphology and biochemical traits. The resulting data are presented, in silico, in a search engine called "Real Time QTL" that displays a range of statistical and graphical outputs that describe in a user-friendly way the components of the genetic variation (http://zamir.sgn.cornell.edu/; Gur et al. 2004).
The S. pennellii IL library was successfully used to identify QTLs for several other traits including disease resistance (Astua-Monge et al. 2000), leaf dissection (Holtan and Hake 2003), fruit nutritional and antiox-idant contents (Rousseaux et al. 2005), and tomato aroma (Tadmor et al. 2002). In the latter study, the analysis of the ripe fruits' volatiles of chromosome 8 ILs allowed the identification of malodorous, a wild species allele negatively affecting tomato aroma that was selected against during domestication.
Compared to populations segregating for an entire genome, ILs are a more powerful genetic tool to study the phenotypic effects of QTL interactions to better understand the nature of epistasis (Tanksley 1993). To address this issue, Eshed and Zamir (1996) crossed ten different homozygous S. pennelli ILs in a half-diallel scheme. The phenotypic values of the 45 double heterozygotes were compared to their respective single heterozygotes and the cultivated control for yield-associated traits (Eshed and Zamir 1996). A high proportion (28%) of the total tested interactions were epistatic (P <0.05), and the detected epistasis was predominantly less-than-additive, i.e., the effect of the double heterozygotes was smaller than the sum of the effects of the corresponding single heterozygotes. The authors suggested the less-than-additive mode of interaction as an underlying genetic model to explain canalized characters, where the phenotype is kept within narrow boundaries despite genetic and environmental disturbances (Eshed and Zamir 1996). This implies that for quantitative traits affected by a large number of QTLs, the less-than-additive interaction ensures that a "loss" of an allele influencing a fitness trait will have a minimal effect on the phe-notype.
ILs can also be used to obtain more precise estimates of the magnitude of QTL x genetic background interaction (G) (Eshed and Zamir 1995, 1996; Monforte et al. 2001; Gur and Zamir 2004; Lecomte et al. 2004b; Chaib et al. 2006). The magnitude of QTL x G interaction is an important issue whenever QTL alleles identified in one genetic background need to be transferred via MAS into other genetic backgrounds. Monforte et al. (2001) found that QTL alleles on the S. peru-vianum chromosome 4 IL (TA1160) maintained their effects across the four processing-type lines used as testers. Similarly, Gur and Zamir (2004) found that the effects on Brix° x yield of three S. pennelli intro-gressions carried in the line IL789 were consistent in all hybrid combinations obtained with four inbred tester lines. On the other hand, Lecomte et al. (2004a) found significant differences among three genetic backgrounds in the improvement of quality traits.
More recently, Semel et al. (2006) used the 76 S. pennellii ILs to assess the contribution of overdom-
inant (ODO) effects on heterosis in the absence of epistasis. For this purpose, the S. pennellii ILs have been evaluated in the field, as homozygous and heterozygous lines, for 35 different yield and fitness traits. QTL analysis identified a total of 841 QTLs, and ODO QTLs were detected only for the reproductive traits. These results suggested that the true ODO model involving a single functional Mendelian locus is a more likely explanation for the heterosis observed in the ILs than the pseudoODO model.
To date, the S. pennellii exotic library is the most studied whole-genome IL population in tomato. Nevertheless, efforts are being invested in the public as well as in the private sectors to generate new library resources starting from interspecific crosses with different wild species of tomato.
For example, from the S. habrochaites (LA1777) AB-QTL population (Bernacchi et al. 1998a, b), Mon-forte and Tanksley (2000a) developed a set of 99 ILs and backcross recombinant inbred lines (BCRILs). Some of the lines contain multiple introgressions, and together they cover approximately 85% of the wild genome.
Similarly to S. pennellii, S. habrochaites is a green-fruited wild species; however, while S. pen-nelli is a drought tolerant species, S. habrochaites (which originates from high altitudes) is a cold tolerant species (Stevens and Rick 1986) (Table 1). S. habrochaites possesses several other characteristics that make it an attractive target as a source of useful QTLs for breeding including dense trichomes, insect resistance, and the synthesis of novel complex secondary metabolites (Stevens and Rick 1986; Van der Hoevenet al. 2000). The AB-QTL studies conducted by Bernacchi et al. (1998a, b) have provided evidence that this wild species also represents a good source of QTL alleles to improve yield and quality. Preliminary QTL studies conducted on some of the S. habrochaites ILs have shown that the lines differ in many traits including yield, leaf morphology, trichome density, as well as fruit characteristics such as shape, size and color. Favorable wild QTL alleles were detected for several of these traits (Monforte and Tanksley 2000b; Monforte etal. 2001; Yates etal. 2004). In addition, the lines differ in biochemical composition including antho-cyanin content (Oyanedel 1999), soluble solids content of the fruit (Monforte and Tanksley 2000b; Monforte et al. 2001) and sesquiterpenes (Van der Hoeven et al. 2000). In order to improve the efficiency of intro-gression breeding based on this wild species, a new set of S. habrochaites ILs, each containing single homozy-
gous wild introgressions, is currently being developed. This set of ILs should ensure whole genome coverage and increase the mapping resolution of the population (S Grandillo and SD Tanksley, pers. comm.).
Starting from the S. pimpinellifolium (LA1589) BC2 populationdevelopedaspartofthe AB-QTLstrat-egy, Doganlar et al. (2002c) derived a set of 196 inbred backcross lines (IBLs). On average the frequency of the cultivated parent allele in the IBL population was 88%, a value consistent with the expected value of 87%. These results confirmed that the recurrent parent's genome was progressively recovered through two backcrosses and four generations of selfing. The 196 lines were evaluated for 22 quantitative traits and a total of 71 significant QTLs were identified, for 48% of which the wild allele improved agronomic performance. Although this population cannot be defined as an "exotic library" yet, as the lines still contain multiple wild introgressions, they represent good starting material for the development of a whole genome population of ILs.
Another exotic library has been developed using the wild tomato-like nightshade Solanum lycop-ersicoides (LA2951) as donor parent, in the genetic background of cultivated tomato (S. lycopersicum cv. VF36) (Chetelat and Meglic 2000; Canady et al. 2005). S. lycopersicoides is native to high elevations (up to 3,600 m) in the Andes, where it survives exposure to freezing temperatures, and is resistant to several insect pests and pathogens that negatively impact tomato production (Chetelat and Meglic 2000). The population of S. lycopersicoides ILs consists of a primary subset of 56 lines which ensure coverage of approximately 96% of the S. lycopersicoides genome. Lines are homozygous whenever possible with a minimum number of introgressed segments per line. A secondary subset of 34 lines provides increased map resolution for specific regions. For this secondary subset, homozygotes were not recovered for certain intro-gressed segments, and several lines are maintained as heterozygous.
Recently, Finkers et al. (2007) reported the development of an IL population (BC5S2) of S. habrochaites LYC4 in the genetic background of S. lycopersicum cv. Moneymaker, an indeterminate growing tomato, by using AFLP as the platform for MAS. The population consists of 30 ILs (15 of which contain a single intro-gression), and together the lines provide an estimated coverage of 95% of the wild species genome. The population was used to identify QTL for resistance to Botrytis cinerea.
In addition to the aforementioned IL populations developed by the public sector, there are several other resources of exotic libraries deriving from crosses with different wild tomato species that have been developed by private companies (Peleman and van der Voort 2003; JD Peleman, pers. comm.). Moreover, several sets of ILs have been developed in Arabidopsis thaliana (Koumproglou et al. 2002), as well as in many other crop plant species including rice (Li et al. 2005b; Tian et al. 2006), wheat (Pestsova et al. 2006), barley (von Korff et al. 2004), lettuce (Jeunken and Lindhout 2004) and melon (Eduardo et al. 2005), and similar population structures have been developed and used in mouse genetics (Singer et al. 2004). These studies reflect an increasing interest within the scientific community to develop permanent congenic populations as efficient tools to understand the genetic and molecular bases of complex traits.
Was this article helpful?