Translational Genomics for Crop Breeding (Abiotic Stress, Yield and Quality) || Genomics-Assisted Breeding for Tomato Fruit Quality in the Next-Generation Omics Age

Chapter 12

Genomics-Assisted Breeding for Tomato FruitQuality in the Next-Generation Omics AgeMatthew P. Kinkade and Majid R. Foolad

Abstract

The cultivated tomato, Solanum lycopersicum L., is the most consumed and most popular vegetablecrop in the world. It is the primary source of the carotenoid lycopene, a highly beneficial dietaryantioxidant whose consumption may reduce the incidence of certain cancer and heart diseases inhumans. Tomato is also an important model system for genetics and genomics studies that haveled to many discoveries, including identification and development of some of the earliest molecularmarkers and genetic maps, fine mapping and cloning of the first plant disease resistance gene, andfine mapping and cloning of the first QTL. Marker-assisted selection (MAS) has been employedextensively in tomato for improving many simple traits. However, MAS has not been frequently usedto advance complex traits in tomato, although many QTLs have been identified for various quantitativetraits. In this chapter we look beyond the heavily reviewed QTL analysis approaches and focus mainlyon the use of new “omics” technology and its potential use for tomato breeding, in particular forimproving fruit quality. With the dawn of the genomics and next-generation sequencing ages, the roleof genomic tools in applied tomato breeding is changing. The tomato genome has been sequenced andthe information is freely available. Genomic and transcriptomic resources and bioinformatic methodshave become available, metabolomic methods have been established, segregating populations havebeen analyzed for alterations in key metabolic traits on an “omic” scale, and large, multi-facetedomics databases have been constructed. Reverse genetics approaches, such as TILLING, have recentlybeen employed to produce novel disease resistance and fruit quality traits in tomato. Mutagenizedpopulations have been developed for use in TILLING approaches, and the bioinformatic workflowsto handle high-throughput identification of mutations in candidate genes have been published. Thus,the pillars now exist upon which a genomics-assisted breeding scheme could be devised for tomato.The challenge is how to seamlessly incorporate these types of analyses, selection methods, and toolsinto a practical breeding program, and determine whether or not the time and expense required forsuch studies can be justified for use in contemporary tomato variety development programs.

Translational Genomics for Crop Breeding, Volume II: Abiotic Stress, Yield and Quality.Edited by Rajeev K. Varshney and Roberto Tuberosa.C© 2013 John Wiley & Sons, Inc. Published 2013 by John Wiley & Sons, Inc.

193

194 TRANSLATIONAL GENOMICS FOR CROP BREEDING

Introduction

Tomato (Solanum lycopersicum L.) is an eco-nomically and nutritionally important cropgrown throughout the world, and has been usedas a model organism for genetics, genomics andphysiological studies for the past 70 years. Manyfactors have contributed to its suitability forboth basic and applied research; this includesease of culture, short life cycle, photoperiodinsensitivity, high self fertility and homozygos-ity, great reproductive potential, ease of con-trolled pollination and hybridization, amenabil-ity to asexual propagation and whole plant regen-eration, and availability of a wide array of mutantand genetic stocks (http://tgrc.ucdavis.edu/;http://www.sgn.cornell.edu/). In addition, S.lycopersicum (hereafter, “tomato”) is a diploidspecies with a rather small genome (∼0.95 pg/1C, 950 Mbp) that lacks extensive gene dupli-cation. Throughout the past 70 years, modu-lating tomato fruit quality to suit grower andend-user specifications has been a focal point inmany tomato breeding programs (Foolad 2007a).As with nearly all crops in the United States,practical tomato breeding was originally ledprimarily by researchers at public institutions,and breeding efforts and various germplasmcollection expeditions by C.M. Rick and oth-ers have formed the basis of the present-daytomato cultigen. Because of the justified focuson tomato by early genetics and molecular biol-ogy researchers, there have also been manywidely-heralded discoveries using the tomatosystem, including but not limited to identifi-cation and development of some of the earli-est plant molecular markers and genetic maps(Tanksley et al. 1982; Tanksley and Orton1983), and fine mapping and cloning of thefirst plant disease resistance gene, Pto (Mar-tin et al. 1993a; Martin et al. 1993b), and thefirst QTL, fw2.2 (Frary et al. 2000). Tomatohas also been an excellent system for breedingpurposes, evidenced by the continual introgres-sion of desirable genes and phenotypes fromwild species and improvement of the crop dis-

ease resistance, fruit quality, and yield (Foolad2007a).

Brief History of Tomato

The Solanum genus is thought to have evolvedapproximately 12 million years ago (Mya), withthe tomato clade (section Lycopersicon) radi-ating from the potato clade (section Solanum)at about 7 Mya (Wikstrom et al. 2001). Sincethen, tomato species have evolved to inhabita vast array of elevations, soil types, and cli-mates (Taylor 1986). In fact, although a tropicalplant, tomatoes are now grown in some formin every region of the world, from the tropicsto within a few degrees of the Arctic Circle(Foolad 2007a). Solanum sect. Lycopersicon ismonophyletic, consists of 13 different species,and has been thoroughly reviewed (Labate et al.2007). The present-day cultivated tomato is theproduct of more than six centuries of domestica-tion and selective breeding, presumably initiatedby Mayan agriculturalists (Kalloo 1991; Peraltaand Spooner 2005). The cultivated tomato isthought to have originated either in Mexico orPeru, and an ensuing controversy over the exactgeographic origin has persisted since the late19th century (Peralta and Spooner 2005). How-ever, despite the wide distribution of the genus inthe Andean region, Mexico has been consideredthe most likely center of domestication (Rick1976b). Regardless of origin and place of domes-tication, natives were already cultivating tomatowhen European explorers arrived in the Ameri-cas in the late 15th century, and subsequently theSpaniards introduced the crop to the Old Worldearly in the 16th century (Rick 1976b; 1978;Peralta and Spooner 2005). Europeans were ini-tially suspicious of tomato due to its morpholog-ical resemblance to poisonous nightshade, thuspreventing widespread use of the tomato as afood until the late 19th century (Rick 1978).Commercial production of tomato in the U.S.began in 1847 at Lafayette College, in Easton,Pennsylvania, leading to a major vegetable pro-duction industry in the mid 20th century (Foolad

GENOMICS-ASSISTED BREEDING FOR TOMATO FRUIT QUALITY 195

2007a). Since then, the tomato has become anintegral part of the human diet as well as anincredibly useful organism for the study of plantgenetics and breeding.

Economic and Nutritional Valueof Tomato

The tomato is a major vegetable crop producedand consumed around the world, and has enjoyeda vast expansion in production over the last50 years. In 1961, total world production oftomato was about 26.2 million tons; in 2009,total world production was about 131 milliontons (FAOSTAT 2012), as a result of increasedacreage and advances in mechanized agricul-ture, industrial processing, and plant breeding.China is the world leader in tons of toma-toes produced (∼41.9 million tons), followedby the United States [14.2 million tons; (FAO-STAT 2012)]. However, it is estimated that theagricultural value of tomatoes produced in theU.S. is more than twice that of tomatoes pro-duced in China [gross production value per tonin constant 2004–2006 US$; (FAOSTAT 2012)].In addition to economic value, fresh tomatoesand tomato-based products constitute an impor-tant supplier of nutrients to humans by virtue oftotal volume consumed. According to the UnitedStates Department of Agriculture, a 100 g serv-ing of tomato paste contains 36 mg calcium,1.014 g potassium, 21.9 mg vitamin C, 38.5 mgcholine, 900 μg β-carotene, 28.8 mg lycopene,and smaller amounts of other vitamins, miner-als, and bioactive components (USDA 2012).Tomato is the premier dietary source of thecarotenoid lycopene; approximately 90% of theaverage human’s yearly intake of lycopene isfrom tomato or tomato-based products (Rao andRao 2007). Recently, this compound has beenidentified as a highly beneficial dietary antioxi-dant via studies that have shown that the pres-ence of this compound is inversely related to theoccurrence of certain types of cancer, as wellas chronic heart diseases (Rao and Rao 2007).The mode of action of lycopene is thought to

be via its system of conjugated double bonds,enabling the compound to quench reactive oxy-gen species (ROS), triplet chlorophylls (in plantsand phototrophic bacteria), and UV radiation(Demmig-Adams and Adams 2002). It has beenshown in human epidemiological studies thatintake of lycopene results in a heightened totalantioxidant capacity, and lower levels of oxi-dized protein and serum lipids (Rao 2004). Thisfinding is significant because increased levelsof oxidation are correlated with increased inci-dence of various diseases, including heart dis-ease, cancer, inflammation, and macular degen-eration (Demmig-Adams and Adams 2002). Anested case-control study concluded that therewas a significant inverse relationship betweenplasma lycopene concentration, dietary intake,and prostate cancer risk in humans older than 65without a family history of incidence (Wu et al.2004). These findings imply that dietary antiox-idants may have a prohibitive effect on sponta-neous prostate cancer occurrence, but this effectmay be secondary to the influence of a stronghereditary component. Despite the conclusionsof several epidemiological studies, which indi-cate that increased plasma lycopene may havea prohibitive effect against the incidence ofprostate cancer, a recent critical review of allstudies concluded that a defensible link betweenlycopene and reduced cancer incidence couldnot be made due to various confounding fac-tors (Von Low et al. 2007). The U.S. Food andDrug Administration recently issued a reportexamining two requests by several commer-cial entities wishing to present qualified healthclaims on food labels regarding the ability oflycopene to prevent certain human cancers, andthe agency concluded that there was little evi-dence supporting the association of tomato con-sumption with reduced cancer risk (Kavanaughet al. 2007). It is expected that more definitiveclinical or in vitro trials will be conducted inthe near future in order to support or refute thislink. Although much has been accomplished interms of identifying the effect of lycopene ondisease prevalence, its actual mode of action


at a molecular level has yet to be definitivelyelucidated.

Practical Breeding Considerations

One of the goals of pre-breeding research intomato is the discovery of novel genetic ele-ments that could be used by tomato breed-ers to modulate important traits in a practicalmanner. Since transgenic approaches for tomatogenetic improvement have remained generallyunacceptable to consumers, new genetic varia-tion has either been introgressed into the culti-gen from closely related wild species or gen-erated via mutagenesis. Due to the fact thatmost tomato fruit quality traits, such as solublesolids content (SSC), viscosity, and carotenoidaccumulation, exhibit complex inheritance andinteract with the environment, the quantitativetrait locus (QTL) approach to identifying usefulgenetic variation has been utilized extensivelyfor pre-breeding purposes. This technique alsohas allowed researchers to develop mapping pop-ulations and identify QTLs within the context ofgermplasm improvement. With the developmentof publicly available molecular maps and coresets of markers, numerous quantitative geneticsstudies have been conducted using various typesof intra- and interspecific populations of tomato,and hundreds of putative QTLs affecting fruitquality traits such as SSC, carotenoid content,fruit size, acidity, firmness, and taste have beenreported and extensively reviewed elsewhere(Foolad 2007a, b; Labate et al. 2007). Concur-rently, much of the commercial-scale tomatobreeding capacity shifted from public to pri-vate institutions. Thus, published results fromsubsequent verification of identified QTLs inrelevant genetic backgrounds and environmentshas been limited, with a few notable exceptions(e.g., Lecomte et al. 2004; Yates et al. 2004;Chaıb et al. 2006; Kinkade 2010). Further, mostdetected QTLs have not been used for practi-cal breeding purposes, with a few exceptions,such as QTLs LIN5 and LS1, which purport-edly increase fruit SSC and are found in some

of today’s processing tomato cultivars (Fridmanet al. 2000; Schaffer et al. 2000; Causse et al.2004; Foolad 2007a). The use of reverse genet-ics approaches, such as “targeting induced locallesions in genomes” (TILLING), for appliedbreeding purposes has, until recently, been hin-dered by the considerable time, labor, and costrequired to produce and screen mutant popula-tions. Recently this approach has been used intomato to produce novel disease resistance (Pironet al. 2010) and fruit quality traits (Gady et al.2012). With the advent of high-throughput pointmutation detection schemes utilizing cheap next-generation sequencing technologies and novelbioinformatic pipelines, mutation breeding hasre-emerged as an attractive approach to gen-erating new genetic variation in tomato breed-ing material (Gady et al. 2009; Rigola et al.2009).

In this chapter we look beyond the heav-ily reviewed QTL analyses that have dominatedtomato literature until recently. Although manyQTLs have been identified for various fruit qual-ity traits in tomato, extremely few are activelyused in today’s commercially grown tomatocultivars (Foolad 2007a; Kinkade 2010). Thereare several legitimate reasons for this discrep-ancy. Many detected QTLs and their associ-ated molecular markers are based on interspe-cific mapping populations of tomato, and thusmay be specific to those populations. Subse-quent verification of such QTLs, their individ-ual effects, and associated markers in breedingpopulations is a time-consuming, expensive, andrisky task. This is especially true when the QTLsare originally described in genetic populationsdistantly related from or irrelevant to a givenbreeder’s germplasm and/or in different agricul-tural environments from a given breeder’s tar-get climatic region. Also, the lack of a univer-sal set of high-throughput marker assays thatcan be utilized to discriminate between geno-types within the cultivated tomato germplasmhas hindered the widespread use of genomicsinformation to assist the breeding process. Intomato, these concerns and roadblocks are


slowly being dealt with, as discussed later in thischapter.

The most popular inbred development tech-nique currently used by practical tomato breed-ing programs is pedigree-based selection, withthe goal of producing desirable inbred lines andsubsequently F1 hybrid cultivars utilizing theinbred lines as parents. By nature, this processfavors traits that are simply inherited and can bevisually evaluated. Further, traits such as yield,vine cover and size, disease resistance, and matu-rity are generally considered higher breedingpriorities than modulating complex fruit qualitytraits such as viscosity and carotenoid content.With a suitably large number of diverse materi-als entering the inbred development pipeline, onecan obtain desired fruit quality parameters usingphenotypic selection; but obtaining inbred lineswith the combination of these parameters along-side improved disease resistance and/or accept-able yield remains difficult, since selection inearly generations focuses primarily on horticul-tural traits, not fruit quality. For example, ina processing tomato breeding program, a plantwith an unacceptably small vine but desirableviscosity is not useful and often would not beselected. Thus, in a hypothetical breeding pro-gram where marker-assisted selection (MAS) isnot applied for fruit quality traits, the number ofinbred lines with desirable horticultural charac-teristics, disease resistance, and thick viscosityat the end of an inbred development pipelineis a function of the number of F2 populationsused, the frequency and number of alleles con-ferring the desirable fruit quality characteristics,the presence of genetic linkages, the breeder’sskill, and random chance. In most cases, this sit-uation results in a low probability of success, aswell as a highly inefficient germplasm develop-ment program. In tomato, MAS is widely usedfor single-gene traits, such as disease resistanceor determinacy, but traits with more complicatedinheritance, such as sugar profile, viscosity, andcarotenoid profile, are still largely subject to tra-ditional phenotypic selection (Foolad 2007a).However, if one desires “enhanced β-carotene

content” as a breeding objective, while also tak-ing into account the grower requirements forsuccessful tomato varieties, utilizing solely tra-ditional breeding techniques is exceedingly inef-ficient (Barone et al. 2009). This area of diffi-culty is precisely where “omics” tools appliedto large, relevant, diverse populations, usinglarge numbers of informative molecular mark-ers and high-throughput phenotyping protocols,can have the largest impact on practical tomatobreeding.

With the dawn of the age of genomicsand next-generation sequencing, the role ofgenomics in applied tomato breeding hascome into focus. The overall goal of apply-ing genomics strategies to aid breeding, thatis, association of phenotypic variation withsequence variation on a genome-wide scale,will not change. Successful application of next-generation “omics” techniques in combinationwith conventional breeding techniques mayresult in the identification and commercial appli-cation of novel traits, as seen with other cropspecies (Tuberosa and Salvi 2006; Tuberosa et al.2007; Dahmani-Mardas et al. 2010). Such traitsmay be difficult to deal with using conven-tional breeding techniques alone. For example,various tomato fruit quality traits are expen-sive to measure, and modulating these traits atthe phenotypic level often results in negativeconsequences to fruit yield (see Schauer et al.2006). The tomato genome has been sequencedand the information is freely available (TomatoGenome Consortium 2012), genomic and tran-scriptomic resources and bioinformatic meth-ods are publicly available, metabolomic meth-ods have been established and reported in theliterature, and research groups are beginningto analyze segregating populations for alter-ations in key metabolic traits on an “omic”scale, while concurrently collecting data fromrelevant agronomic traits. Further, mutagenizedtomato populations have been developed foruse in TILLING approaches (Menda et al.2004; Minoia et al. 2010; Okabe et al. 2011)and the bioinformatic workflows to handle


high-throughput identification of mutations incandidate genes have been published (Gady et al.2009; Rigola et al. 2009). All of this is leadingto the capability of using contemporary com-binatorial “omic” methods to achieve breed-ing outcomes in tomato; whether or not prac-tical tomato breeding outcomes will be obtainedand/or explicitly reported in the literature usingthese methods remains to be seen.

Key Advances EnablingGenomics-Assisted Breedingin Tomato

The tools required for genomics-assisted breed-ing in tomato are available: the tomato genomicsequence has become available for publicuse, early proof-of-concept “omics” studieshave been conducted on genetic populationsof tomato, large-scale marker identificationprojects have produced multitudes of markers foruse within the tomato cultigen, new TILLINGpopulations have been developed and character-ized, and large, multi-faceted omics databaseshave been constructed, which allow users to com-pare multi-omics datasets in order to constructcorrelational networks of genes. Yet, as in manyother crop species, single-gene MAS (mainlyfor disease resistance) remains the most widely-used molecular breeding technique in practicefor tomato, and the status of “genomics-assistedbreeding” of tomato in the literature remainsnascent at best. In fact, a literature search of“genomics assisted breeding of tomato” cur-rently returns a litany of review articles but scantprimary research reports. Nevertheless, the pil-lars upon which a genomics-assisted breedingscheme could be devised now exist for tomato.In this section, we describe these resourcesand some early examples from the literaturethat have begun to bridge the gap betweenbasic genomics and practical breeding outcomesrelated to tomato fruit quality. The challenge ishow to seamlessly incorporate these types ofanalyses and selection methods into a practi-cal breeding program, and determine whether or

not the time and expense required for such stud-ies can be justified to the contemporary tomatobreeder.

Early Breeding Research and the S.pennellii LA716 Introgression Lines

The cultivated tomato, S. lycopersicum L., hasbecome a model system for quantitative geneticstudies for a variety of reasons. Tomato is adiploid organism and a self-pollinator, seed-to-seed generation time is as little as threemonths, controlled hybridizations are easy toconduct, wide phenotypic variation exists withinthe available germplasm, and diverse geneticpopulations are relatively simple to construct(Foolad 2007a, b; Labate et al. 2007). Thebody of work led by C. M. Rick and col-leagues paved the way for tremendous geneticimprovement of tomato varieties using acces-sions from the related wild tomato species (i.e.,Solanum sect. Lycopersicon species other than S.lycopersicum). Based on collections and botan-ical studies of wild tomato species as wellas isozyme surveys of genetic variation withinSolanum sect. Lycopersicon, which indicatedextremely low genetic variability within the cul-tivated species, it was determined that wildgermplasm held the key to the improvement ofmany important agricultural traits in tomato (seeRick and Fobes 1974, 1975; Rick 1976a, 1978,1979). As a result, some breeders and geneti-cists delved back into the wild tomato germplasm(accessions of botanically related, but undomes-ticated tomato Solanum species) in order toidentify and transfer beneficial traits into breed-ing lines. As early as 1982, molecular markers(specifically isozymes) were being used to maptomato genes involved with phenotypic char-acters (Tanksley et al. 1982). These develop-ments resulted in the creation of segregatinginterspecific populations of tomato, and, com-bined with collaboration between public andprivate researchers, enabled the genetic anal-ysis of various phenotypic traits (reviewed inFoolad 2007a; b). Since then, a wealth of QTL


mapping studies has been conducted using thetomato system (reviewed in Foolad 2007a andLabate 2007), which has subsequently sparkedin-depth transcriptomic and metabolomic stud-ies, in addition to sequencing the tomato genome(Mueller et al. 2009). Combined with the tomatogenome sequence, such information can nowbe leveraged in the search for candidate genesunderlying detected QTLs, or in the search fornew QTLs. As a result, researchers have usedsegregating interspecific populations to iden-tify new alleles affecting various traits andto subsequently clone and characterize them(reviewed in Foolad 2007a). A majority of stud-ies has been conducted using most widely knowntomato population, the S. pennellii-based intro-gression lines [ILs; (Eshed and Zamir 1995)].Many other tomato “immortalized” mappingpopulations with useful traits have been con-structed (see Foolad 2007b; Ashrafi et al. 2009)and are expected to be heavily utilized infuture for both strategic and applied breedingresearch.

Although more time-consuming to developthan many other mapping populations, IL pop-ulations can be more useful for QTL mappingand phenotyping, since any significant differ-ence(s) between an IL and the recurrent par-ent is due to a single introgressed segment ofthe donor S. pennellii genome. However, onceQTLs are detected, the physical size of thegenomic bin in which each QTL resides maybe quite large; therefore time saved by remov-ing background wild genomic intervals prior toQTL analysis may be moot, as further backcross-ing, marker mapping, recombinant selection, andvalidation experiments are still required priorto practical application. Thus we posit that theoverall “work” required to identify and trans-fer QTLs to elite germplasm using ILs is simi-lar to other population types (see Barone et al.2009). Nevertheless, in tomato, the S. pennellii-based IL population has been the foundation ofmany genetic discoveries and continues to bea useful tool for contemporary omics experi-mentation in the public sector (Liu et al. 2003;

Rousseaux et al. 2005; Schauer et al. 2005;Schauer et al. 2006; Fraser et al. 2007a; Lipp-man et al. 2007; Bermudez et al. 2008; Schaueret al. 2008). QTLs for altered soluble solidscontent, yield, fruit size, metabolic traits, andmany others, have been identified using theILs, and some of these QTLs are actively usedby commercial breeding programs (Rousseauxet al. 2005; Lippman et al. 2007). This popula-tion has also been used for metabolomic stud-ies (Schauer et al. 2006; Fraser et al. 2007a;Lippman et al. 2007; Bermudez et al. 2008;Schauer et al. 2008), which will be discussedlater in this chapter. In addition, markers devel-oped using the interspecific populations con-structed for quantitative genetics studies helpedform the basis of the tomato genome sequencingproject.

Sequencing of the Tomato Genome

The tomato genome sequencing project offi-cially began in 2004 as an international col-laboration among ten countries (Mueller et al.2009). A BAC-by-BAC approach, beginningwith physical mapping and BAC (bacterial arti-ficial chromosome) tiling followed by “goldenpath” sequencing, was initially proposed, andas sequencing technology advanced rapidlyduring the 2000s, this approach was utilizedto buttress next-generation sequencing datasets(Mueller et al. 2009). A formal report onthe genome sequence and insights into theevolution of tomato has recently been pub-lished (Tomato Genome Consortium 2012). AnFTP site within the Solanaceae Genomics Net-work (SGN; http://solgenomics.net) for massdata download has also been developed forresearchers wishing to conduct their own anno-tation and/or sequence investigation (such asrepeat masking, transposon detection, gene iden-tification, etc.; Mueller et al. 2009). Gene identi-fication using the raw sequence and the MAKERannotation tool (Cantarel et al. 2007), fol-lowed by track visualization using Apollo (Lewiset al. 2002), is now possible and has been


conducted for QTL candidate gene identificationand marker development in targeted genomicregions (Kinkade 2010; Kinkade and Foolad,unpubl. results).

New Tomato Genomics Resources

A publicly available genome sequence has sup-ported the direct application of genomics totomato breeding on a number of levels. For thefirst time, tomato researchers are now able toidentify potential markers directly within regionsof interest, design primers using the genomesequence, and screen for polymorphism withintheir population of interest (Kinkade 2010;Kinkade and Foolad, unpubl. data). In addition,as genotyping-by-sequencing (GBS) becomesmore cost-effective, a reference genome willassist researchers in mapping the tens of thou-sands of putative single nucleotide polymor-phisms (SNPs) identified by this process withincontemporary breeding populations. An exten-sion of this approach has been taken up bythe Solanaceae Coordinated Agricultural Project(SolCAP; http://solcap.msu.edu). Focused onimproving genomics resources and training forpotato, tomato, and pepper, and translating theseresources into applied outcomes, SolCAP hasintroduced a highly successful series of train-ing articles and videos on its eXtension web-site (http://www.extension.org/plant_breeding_genomics). SolCAP has identified thousandsof SNPs within the cultivated tomato (Simet al. 2012) and potato (Hamilton et al.2011) germplasm and made them available onan Infinium array for the breeding commu-nity to utilize. At the very least, this projecthas provided successful training tools and asubstantial, universal set of informative SNPmarkers to the broader tomato community.Applied breeding outcomes in tomato by theSolCAP consortium have yet to be formallypublished; future research as part of Sol-CAP is anticipated to elucidate alleles con-trolling variation in carbohydrate and vitaminmetabolism.

In parallel to the SolCAP and tomato sequenc-ing projects, the Tomato Functional GenomicsDatabase [TFGD; (http://ted.bti.cornell.edu/; Feiet al. 2011)] has made a variety of expressedsequence tag (EST) datasets, metabolite anal-yses, microarray experiments, and small RNAexperiments available for public query andanalysis. Rather than simply act as a repos-itory for experimental results, the TFGD hasexpanded its capabilities for analysis of microar-ray data, sRNA (small RNA) experiments, andmetabolomic datasets (Fei et al. 2011). This hasenabled researchers to analyze data in a consis-tent, reliable manner – one of the key develop-ments that must occur if we are to build upon pre-vious knowledge and utilize it in a practical way.Each of these resources has been integrated (orat least linked) with SGN, which has positionedSGN as the current nexus for tomato omics infor-mation (maps, markers, analysis tools, pheno-types, and genotypes). A “breeder’s toolbox” hasbeen made available for researchers searching formarkers, phenotypes, and/or QTLs, although thepractical utility of information contained withinthis portal is currently limited. Nonetheless, mas-sive amounts of tomato genomics informationand resources are freely available through SGN,and these resources continue to support appliedresearchers.

Lastly, reverse genetics approaches to tomatogermplasm improvement have emerged as acost-effective way to generate novel genetic vari-ation. With the development of efficient bioin-formatics pipelines to detect point mutations inlarge populations, coupled with massively par-allel sequencing strategies utilizing cheap next-generation sequencing technologies, TILLINGcan now compete with forward genetics strate-gies for pre-breeders’ attention. Combined withthe fact that a large body of basic plant biologyresearch has been conducted using the tomatosystem, which can be leveraged to target spe-cific genes or gene families involved in impor-tant phenotypes, this approach is well-suited foruse in tomato genetic improvement. In addi-tion, researchers can obtain and evaluate many


lines with mutations in candidate genes muchmore quickly than developing materials usingthe QTL method. Another advantage to TILL-ING compared to the QTL approach is that themutations are already incorporated into a com-pletely isogenic background; once useful traitsare identified, they can be easily transferredto breeding populations. TILLING populationshave been developed using the processing tomatolines M82 (Menda et al. 2004) and Red Setter(Minoia et al. 2010), as well as the laboratory-friendly Micro-Tom (Okabe et al. 2011). At thevery least, these populations are important proof-of-concept examples for other researchers, andwill most likely serve as integral resources forpre-breeding efforts in the future. Gady et al.(2009) and Rigola et al. (2009) report threedifferent high-throughput methods of detectingpoint mutations in tomato EMS (ethyl methane-sulfonate) mutant populations, and each methodresults in an acceptable mutation detection rate inthe populations analyzed. Both reports describethe use of next-generation sequencing technolo-gies coupled with elegantly tailored poolingstrategies.

Fruit Quality Traits Targeted forGenomics-Assisted Breeding inTomato

As discussed earlier, many agronomically impor-tant traits in tomato can be easily phenotypedand as a result, conventional phenotypic selec-tion and field-based breeding practices remainmost effective for developing improved tomatobreeding lines and commercial cultivars. How-ever, other traits, such as sugar metabolite pro-file, carotenoid content and profile, and fla-vor profile are more difficult to phenotype inbreeding-scale systems and therefore tend tobe neglected by practical breeding programs.These complex traits have been targeted bytomato researchers in an attempt to provide toolsand analyses that utilize contemporary proto-cols and equipment, complement conventionalbreeding practices, and produce real outcomes.

Such studies are described in the followingsections.

Primary Metabolites

The observed phenotypes associated with fruitquality traits are inherently the result of metabo-lite production and composition within the fruit,and the genetic components of the phenotypicvariances for such traits have been shown tobe substantial. Early studies reporting metabolicprofiling of segregating populations and attempt-ing to construct correlational networks of genesinvolved with metabolic QTLs have made signif-icant progress (Schauer et al. 2005; Schauer et al.2006; Fraser et al. 2007a; Bermudez et al. 2008;Schauer et al. 2008). Recently tomato genomics-assisted breeding research has benefited fromseveral studies aiming to identify and describevariation in primary metabolites. These metabo-lites play important roles in the determination ofkey tomato fruit quality attributes, such as solu-ble solids and acidity. What is widely accepted isthat primary and secondary metabolite levels canbe substantially different between tomato geno-types, even within breeding germplasm, and alsodiffer as a result of environmental conditions.Deeper understanding of the inheritance of pri-mary metabolite QTLs and thorough investiga-tion of the ultimate effect (or lack thereof) ofmodulating the accumulation of these metabo-lites on fruit quality is still required in orderto establish legitimacy of these QTLs. Recentstudies have sought to explore these questionsusing the tomato system (Schauer et al. 2006;Bovy et al. 2007; Bermudez et al. 2008; Schaueret al. 2008).

Current high-throughput metabolic profilingprotocols are amenable to quantifying aqueouscompounds; therefore, such species of com-pounds were the focus of the first metaboliteprofiling studies in tomato. Overy and colleagues(2005) examined the metabolic profiles of one S.lycopersicum accession, one S. pennellii acces-sion, and 5 S. pennellii IL accessions, anddemonstrated that metabolic profiling combined


with principal component analysis could indeedbe used to distinguish tomato genotypes. A sim-ilar, yet more robust experiment was carried outby Schauer and colleagues (2005), who reportedGC/MS (gas chromatography/mass spectrome-try) profiling of accessions of 5 Solanum speciesand described substantial differences in sugars,organic acids, amino acids, and other metabo-lites between the different species. Neither ofthese studies examined large populations in anagricultural setting, as opposed to studies con-ducted by Schauer and colleagues (2006 and2008), who reported extensive evaluations of theS. pennellii IL population for metabolic profileas well as agronomic traits. Schauer et al. (2006)were the first to evaluate an entire genetic pop-ulation of tomato for metabolite profile, in anagriculturally relevant setting, along with agro-nomic traits such as soluble solids, yield, fruitsize, and so forth. This study elucidated spe-cific, genotype-dependent differences in specificmetabolites, such as higher fructose in introgres-sion line 6-3, and also established positive andnegative correlational networks between specificmetabolites as well as between various classes ofmetabolites and agronomic traits. While manymetabolic QTLs were identified, many of thesewere also coupled with harvest index-associatedQTLs, casting some doubt on the practical util-ity of the detected QTLs (Schauer et al. 2006).It remains unclear to what extent negative link-age drag influenced the findings from this study.However, it is clear from this study that metabo-lite QTLs do exist in wild germplasm, metabolicnetworks can be constructed in tomato using wildgermplasm sources, metabolic data can be col-lected in parallel to agronomic and horticulturaldata, key nodes of regulation can be identified onan omics scale, and these nodes may be exploitedto alter metabolic traits in tomato.

A later study examined the inheritance ofthese primary metabolic traits in the same pop-ulation (Schauer et al. 2008). Building on theresults of the previous study, three years’ worthof metabolite data were combined to identifymetabolite QTLs conserved across years and to

examine the heritability of the traits. In addi-tion to the homozygous S. pennellii IL (intro-gression lines) population, the researchers alsoanalyzed heterozygous ILs [ILHs; (Semel et al.2006)]. From the 43 QTLs detected over the threeyears of experiment, a striking number of themwere fairly heritable. Although some heritabilityestimates fluctuated widely over the three yearsof this study, this is to be expected when deal-ing with traits that are subject to environmentalconditions (Schauer et al. 2008). In addition, itwas reported that most metabolite QTLs weredominant, which is a promising result from abreeding standpoint, since most tomato breed-ers are attempting to develop hybrid cultivars.The researchers also compared metabolic net-works between the IL and ILH populations, andfound that a significant amount of the previouslydetected associations between metabolite QTLsand harvest index were abolished in the het-erozygotes (Schauer et al. 2008). This findingis an important discovery for breeders attempt-ing to modulate these traits. Bermudez and col-leagues (2008) took a candidate gene approachin order to further study QTLs identified bySchauer and colleagues (2006) and to asso-ciate sequence variation within candidate generegions with the previously detected QTLs. Theapproach utilized available genomic resourcesto (1) identify potential candidate genes withingenomic regions of interest, (2) compare allelicvariation between the two original parentallines, and (3) correlate transcription of somecandidate genes with metabolite abundance.Although Bermudez and colleagues (2008) werelimited by the lack of a complete genomesequence, this did not prevent the identifica-tion of many promising candidate alleles under-lying previously detected QTLs; in addition,the approach simultaneously produced gene-specific markers that could be used in practicalapplication.

There is still a long way to go in theeffort to combine metabolite profiling, genomics,and practical breeding for tomato fruit quality.However, the use of immortalized populations


(ILs), common phenotyping platforms, transpar-ent bioinformatic methods, and relevant environ-mental conditions strengthened and legitimizedthe findings of Schauer and colleagues (2006 and2008). It is hoped that future studies adhere tothese examples. The resource limitations of theBermudez and colleagues (2008) study are nolonger relevant, now that a genome sequence anda large set of genome-wide markers amenableto high-throughput assays are available. Thesestudies set the stage for the next generation ofgenomics-assisted breeding strategies in tomato.

Carotenoids

Carotenoids are C40 terpenoid compounds foundabundantly in tomato fruit. Ripe tomato fruitscontain significant amounts of lycopene and β-carotene (the precursor of vitamin A), and smallamounts of phytoene, phytofluene, lutein, andzeaxanthin. The concentration of each of thesecompounds depends on the genotype, the matu-rity stage of the fruit, and the environmental con-ditions in which the plants are grown. The insolu-bility of carotenoids in aqueous solutions createschallenges for high-throughput quantification ofthese compounds, and therefore carotenoid datawas not reported in the Schauer and colleagues(2006 and 2008) studies. This class of secondarymetabolites is ubiquitous throughout nature; thepresence of conjugated double bonds in a longpolyene chain renders carotenoids indispens-able for several reasons, including quenchingROS generated by electron transport mecha-nisms and photosynthetic reactions, harvestingphotons, and, due to their distinct spectral qual-ities, attracting pollinators and seed disper-sal agents (Hirschberg 2001; Bramley 2002).Throughout evolutionary time, carotenoids havebeen recruited for a variety of extremely impor-tant cellular purposes, including light harvesting,dissipation of excess solar energy, scavenging ofROS, strengthening sight (in animals), and thebiosynthesis of hormones (Vershinin 1999).

Since animals cannot synthesize these com-pounds de novo, carotenoid levels in animals are

directly related to the amount ingested throughthe diet (Demmig-Adams and Adams 2002). Onesuch compound, lycopene, has been co-optedby some plants to attract seed dispersal agentsthrough its bright red appearance. Lycopene isresponsible for the deep red hue of ripe tomatofruit and is the most abundant carotenoid in thisorgan, and since deep red fruit color is a con-ventional tomato breeding objective, elevatinglycopene content is a parallel objective to fruitcolor. The biosynthesis, and to some extent reg-ulation, of carotenoid accumulation is fairly wellunderstood, although some uncertainty remainsregarding the exact mechanism that funnelsmetabolites into the carotenoid pathway. Thesetopics have been extensively reviewed and dis-cussed (Hirschberg 2001; Fraser and Bramley2004; Fraser et al. 2008).

As follows from current knowledge (andspeculation) surrounding carotenoids and humanhealth, increasing levels or altering types ofcarotenoids in tomato is of importance froma nutritional standpoint. β-carotene is the onlycarotenoid defined as a “nutrient” by the USDA,yet a large proportion of the relevant literaturehas focused on modulating lycopene content intomato, as lycopene is the most predominantcarotenoid in the red tomato fruit (Foolad 2007a).Also, elevated fruit carotenoid content in tomatocan be considered a value-added trait, potentiallyresulting in increased consumer desire for suchvarieties or food products, although the effectof increased carotenoids on tomato product saleshas yet to be definitively quantified. A recent sen-sory study concluded that consumers associateincreased tomato fruit quality and better overalltaste with red color; however, under conditions inwhich the fruit color was masked, the associationwas abolished (Stommel et al. 2005). Regardlessof whether or not lycopene actually provides ameasurable health benefit, when faced with thechoice between a deeply red tomato and a lightlyred or yellow tomato, consumers seem to pre-fer the former. Stommel and colleagues (2005)claim this preference results from a visual asso-ciation between intense red fruit color and better


overall quality and taste. In addition, for process-ing tomatoes, the raw tomatoes used for red pastemust meet a certain grade of redness in orderto be accepted by processing plants. Followingthis line of thought, a better understanding ofthe genetic factors affecting lycopene accumu-lation in tomato will allow for more successfultraditional breeding and metabolic engineeringof this important crop, with the eventual goalsof increased economic benefit for breeders andgrowers (by elevating demand, not necessarilyprice) as well as potentially increased nutritionalbenefit to consumers.

Due to the nascent status of omics-basedresearch to identify novel carotenoid regulatorytargets in tomato and the relative ease of selectingtomatoes of the desired color, commercial breed-ers continue to employ a series of mutations iden-tified in the mid-20th century in order to mod-ulate carotenoid type and accumulation level.Several naturally occurring tomato mutations instructural enzymes of the carotenoid biosyntheticpathway have been observed, reported in the lit-erature, and are actively employed for breedingpurposes. Of these, r (null mutation in PSY),β (LCY-B not down-regulated in response toripening), Delta (lycopene-δ-cyclase not down-regulated in response to ripening), tangerine (t;mutation in CRTISO) and old-gold (og) or crim-son (ogc; mutation in LCY-B), are most signif-icant, owing to their drastic fruit phenotypes(Ronen et al. 1999; Ronen et al. 2000; Liu et al.2003). Among these, og/ogc is the only mutantthat results in increased lycopene accumulation,albeit at the expense of β-carotene. The ogc

mutation is widely used in commercial vari-eties to increase fruit color intensity (Faria et al.2003). Commercial varieties harboring some ofthe other mutations have been developed andused as ingredients in an expanding array ofprocessed food products or sold for fresh mar-ket consumption. Reports of QTLs that modu-late carotenoid content have been published andreviewed by Foolad (2007a).

While much is known about the synthesisof carotenoids in plants, the “next frontier” is

the elucidation of how carotenoid accumulationis regulated in tomato fruit and the harness-ing of such knowledge for breeding purposes(Fraser and Bramley 2004). Unlike most plants,tomato hyper-accumulates lycopene during fruitripening by converting chloroplasts into chro-moplasts, and has evolved a specialized sys-tem to specifically produce this compound incomparatively large amounts. Intuitively, thiswould require a constant supply of metabolicprecursors, a consistently high amount of activebiosynthetic machinery, and a tightly controlledmetabolic flux specifically favoring carotenoidbiosynthesis over other competing pathways.Thus, there are two main questions that have yetto be answered in this regard: What is the spe-cific regulatory mechanism utilized by tomato toconvert fruit from a photosynthetic tissue into ahighly specialized carotenoid factory? How doesthis mechanism differ between low-lycopene S.lycopersicum varieties and high-lycopene wildtomato accessions, such as some S. pimpinelli-folium accessions, assuming carotenoid biosyn-thetic genes are functioning normally in bothcases?

It is generally hypothesized that lycopeneaccumulation in ripe tomato fruit is influencedby the action of many regulatory genes (the spe-cific natures of most are unknown), and in mostsegregating populations, variation in lycopenecontent is continuous, which indicates quanti-tative inheritance of the trait. Also of note, ithas been observed that environmental condi-tions, as well as fruit size (which has been nega-tively correlated with lycopene content), highlyinfluence the final concentration of lycopene inthe fruit. So, intuitively, one would concludethat lycopene accumulation in tomato fruit mustdepend not only on the action of biosyntheticgenes, which have been the major focus of basicresearchers until recently, but also on as-of-yetunidentified regulatory interactions (which maybe involved in metabolic flux, transcriptional andpost-transcriptional regulation, response to light,or a combination of these), and indirect mecha-nisms present in the plant and the environment


that influence ripening, nutrient acquisition, andphotosynthetic productivity.

Studies focusing on the application of con-temporary genomics information to tomatobreeding for the purpose of modulatingcarotenoid levels have been conducted. A TILL-ING approach was employed by Gady and col-leagues (2012) in order to identify and describetwo different mutations in the PSY1 gene thatexhibited significant phenotypic effects. Using amutagenized population described in Gady andcolleagues (2009), the researchers identified M2individuals with point mutations in the PSY1sequence and produced tomato lines with nulland weak alleles of PSY1 that resulted in alteredcarotenoid accumulation (Gady et al. 2012). Thisstudy is an example of the utility of TILLING forapplied breeding objectives. Now that rapid andcost-effective identification of mutants is pos-sible, it is expected that TILLING approacheswill be employed by pre-breeders to generatenovel genetic variation in more genes that areinvolved in fruit color, carotenoid accumulation,sugar accumulation and content, ripening, andmore.

Concurrently, forward genetics approachesare still routinely employed to mine existinggenetic variation in closely related wild speciesof tomato. One recent multi-year QTL studyidentified two QTLs, stably inherited from aS. pimpinellifolium accession, that significantlyincreased fruit lycopene content in a recom-binant inbred line (RIL) population (Ashrafiet al. 2009, 2012). This research was followedup with a marker-assisted backcross (MABC)experiment to simultaneously verify the QTLeffects, fine map the QTL regions, and developnear-isogenic lines useful for breeding purposes(Kinkade 2010; Kinkade and Foolad, unpubl.results). In this case, a large number of molecu-lar markers were required to execute the MABCbreeding and fine mapping experiments effec-tively; therefore, new simple sequence repeat(SSR) markers were developed utilizing avail-able tomato genomic BAC end sequences physi-cally located near the QTL regions. Field exper-

iments were conducted using BC2 and BC2F2

populations segregating for different QTL inter-val sizes, and fruit lycopene content was mea-sured by high performance liquid chromatogra-phy (HPLC). As a result, the phenotypic effectof one QTL (lyc12.1) was verified as stable inthe genetic background analyzed (Kinkade 2010;Kinkade and Foolad, unpubl. data). In addition,lyc12.1 was delimited to a ∼0.3 cM region ofchromosome 12, near-isogenic lines (NILs) withlyc12.1 in a relevant cultivated tomato geneticbackground were developed, and tightly-linkedco-dominant markers for tracking the QTL wereproduced (Kinkade 2010; Kinkade and Foolad,unpubl. data). The results of this research weremade possible in part by the availability oftomato genomics resources, which were success-fully employed to translate findings from a QTLstudy into a validated breeding tool for increasinglycopene content. Further, this series of studiesindicates that current tomato genomics tools areuseful enough to develop tiny introgressions con-taining useful alleles from wild accessions andtransfer them to cultivated genetic backgroundswith minimal linkage drag. However, more high-throughput techniques are required to developcorrelational networks and identify MAS targetsfor carotenoid regulation from large populationsin an efficient manner.

Omic-scale identification of carotenoid reg-ulatory mechanisms in segregating populationshas just begun to be conducted, and some ofthe necessary tools have been developed for thetomato system. Fraser and colleagues (2007b)demonstrated that high-throughput metabolicprofiling techniques based on MALDI/TOF-MS (matrix-assisted laser desorption/ionizationtime-of-flight mass spectrometry) could beemployed to evaluate tomato populations seg-regating for carotenoid accumulation level andcarotenoid profile using crude extracts. Becausequantifying the various species of carotenoidsin large populations is laborious and expen-sive, this has tremendous utility for genomics-assisted breeding for increased carotenoid con-tent. Although it has not been definitively


demonstrated that MALDI/TOF-MS can com-pletely replace conventional HPLC (largely dueto the fact that lycopene and β-carotene haveidentical m/z values), this technique offersrapid identification of high- or low-accumulatinggenotypes and also has the ability to detect othercarotenoids in tandem.

Future Directions

It is hoped that subsequent studies adopt thesystems biology approach of Schauer and col-leagues (Schauer et al. 2005, 2006, 2007) in con-junction with the carotenoid profiling techniquespioneered by Fraser and colleagues (Fraser andBramley 2004; Fraser et al. 2007a,b, 2008),and utilize relevant genetic backgrounds if thereis truly a desire to translate forward geneticsfindings into breeding practices. The systemsapproach takes advantage of genetic variation toproduce correlational networks and infers keyregulators of important traits from these net-works. Research teams can then employ contem-porary high-throughput techniques to identify (orgenerate) genetic variation within these regula-tory genes, in order to produce novel metabolictraits that can be tested under relevant agricul-tural conditions. It is not out of the realm of con-sideration that continued tomato omics studiescan, by exploiting the well-developed genomicsresources for tomato (Mueller et al. 2005) andthe continuously-maturing techniques for inte-grating various types of omics data, fuel thedevelopment of tomato omics resources similarto those developed for yeast (Myers et al. 2005)or other biological systems.

The exploration, characterization and incor-poration of abundantly available natural geneticvariation, or generation of variation using muta-genesis in agriculturally relevant genetic back-grounds, still represents an opportunity for prac-tical breeders to identify and incorporate novelalleles for fruit quality traits into the cultigen.As described above, genomic and metabolomicexperimental tools have been developed usingtomato as a model system to mine this variation.

With the advent of genotyping by sequencingand SNP arrays, the inability to obtain extremelylarge sets of informative polymorphic markerswithin the tomato cultigen is no longer an issue.There is no reason why a similar overall approachto that of Davuluri and colleagues (2005) can-not be executed for carotenoid regulatory tar-gets identified by correlational networks whileusing publicly acceptable sources of geneticvariation, rather than genetic engineering. Earlyapplied TILLING studies have produced tomatomaterials with altered carotenoid content (Gabyet al. 2012) and potyvirus resistance (Piron et al.2010). Other examples include the developmentof melon materials with improved shelf life char-acteristics (Dahmani-Mardas et al. 2010). Thiseffort will require a well-coordinated collabo-rative effort among bioinformaticians, chemists,molecular biologists, and breeders, and align-ment on a practical, concrete breeding objectiveto pursue with such an approach.

In general, tomato fruit quality is consideredby practical tomato breeders to be a lower pri-ority than yield, disease resistance, and otherproduction-related traits. There are extremelyvalid and unavoidable reasons for this. How-ever, consumers are increasingly dissatisfiedwith the quality of fresh tomatoes available tothem (Kolata 2012). With the technologies avail-able to applied research and development enti-ties (public and private), it is now feasible togenerate high-yielding, durable tomato varietieswith superior fruit quality if the desire to do soexists. In the case of altering tomato carotenoidaccumulation, we speculate that the main reasonthis has not been widely pursued is that con-sumers are unwilling to pay more for increasedantioxidant content, and thus this trait has notbeen a priority for breeders, growers, or pro-cessors. As a result, currently, private researchand development firms are not willing to investheavily in this area, nor are seed companieswilling to license such traits for hefty sumsfrom the public sector, though this may not bethe case in the future. On the other hand, pri-mary metabolic traits can be related to yield and


compose the final fruit quality attributes thatare prioritized by the customers of seed com-panies. Association of individual metabolite lev-els or metabolite profiles with critical fruit qual-ity metrics, such as soluble solids and viscosity,concurrent with field testing in relevant geneticbackgrounds, will hasten the incorporation ofgenomics-assisted findings into tomato breedingpractices.

References

Ashrafi, H., Kinkade, M.P., and Foolad, M.R. 2009. Anew genetic linkage map of tomato based on a Solanumlycopersicum × S. pimpinellifolium RIL population dis-playing locations of candidate resistance ESTs. Genome52:935–956.

Ashrafi, H., Kinkade, M., Merk, H., and Foolad, M.R.2012. Identification of novel QTLs for lycopene con-tent and other fruit quality characteristics in a Solanumlycopersicum x S. pimpinellifolium RIL population. MolBreeding published online; DOI 10.1007/s11032-011-9643-1.

Barone, A., Di Matteo, A., Carputo, D., and Frusciante,L. 2009. High-throughput genomics enhances tomatobreeding efficiency. Current Genomics 10:1–9.

Bermudez, L., Urias, U., Milstein, D., Kamenetzky, L.,Asis, R., Fernie, A.R., Van Sluys, M.A., Carrari, F., andRossi, M. 2008. A candidate gene survey of quantitativetrait loci affecting chemical composition in tomato fruit.Journal of Experimental Botany 59:2875–2890.

Bovy, A., Schijlen, E., and Hall, R. 2007. Metabolicengineering of flavonoids in tomato (Solanum lycoper-sicum): The potential for metabolomics. Metabolomics 3:399–412.

Bramley, P.M. 2002. Regulation of carotenoid formationduring tomato fruit ripening and development. Journalof Experimental Botany 53:2107–2113.

Cantarel, B.L., Korf, I., Robb, S.M.C., Parra, G., Ross,E., Moore, B., Holt, C., Sanchez Alvarado, A., andYandell, M. 2007. MAKER: An easy-to-use annota-tion pipeline designed for emerging model organismgenomes. Genome Research 18:188–196.

Causse, M., Duffe, P., Gomez, M.C., Buret, M., Damidaux,R., Zamir, D., Gur, A., Chevalier, C., Lemaire-Chamley,M., and Rothan, C. 2004. A genetic map of candi-date genes and QTLs involved in tomato fruit size andcomposition. Journal of Experimental Botany 55:1671–1685.

Chaıb, J., Lecomte, L., Buret, M., and Causse, M. 2006. Sta-bility over genetic backgrounds, generations, and years ofquantitative trait locus (QTLs) for organoleptic quality intomato. Theoretical and Applied Genetics 112:934–944.

Dahmani-Mardas, F., Troadec, C., Boualem, A., Leveque,S., Alsadon, A.A., Aldoss, A.A., Dogimont, C., andBendahmane, A. 2010. Engineering melon plants withimproved shelf life using the TILLING approach. PLoSONE 5:e15776.

Davuluri, G.R., van Tuinen, A., Fraser, P.D., Manfredonia,A., Newman, R., Burgess, D., Brummell, D.A., King,S.R., Palys, J., Uhlig, J., et al. 2005. Fruit-specific RNAi-mediated suppression of DET1 enhances carotenoid andflavonoid content in tomatoes. Nat Biotech 23:890–895.

Demmig-Adams, B., and Adams, W.W. 2002. Antioxi-dants in Photosynthesis and Human Nutrition. Science298:2149–2153.

Eshed, Y., and Zamir, D. 1995. An introgression linepopulation of Lycopersicon pennellii in the cultivatedtomato enables the identification and fine mapping ofyield-associated QTL. Genetics 141:1147–1162.

FAOSTAT. 2012. Food and Agriculture Organization of theUnited Nations. http://faostat.fao.org.

Faria, M.V., Maluf, W.R., Azevedo, S.M.d., Andrade Junior,V.C.d., Gomes, L.A.A., Moretto, P., and Licursi, V.2003. Yield and post-harvest quality of tomato hybridsheterozygous at the locialcobaca, old gold-crimson orhigh pigment. Genetics and Molecular Research 2:317–327.

Fei, Z., Joung, J.-G., Tang, X., Zheng, Y., Huang, M., Lee,J.M., McQuinn, R., Tieman, D.M., Alba, R., Klee, H.J.,et al. 2011. Tomato Functional Genomics Database: Acomprehensive resource and analysis package for tomatofunctional genomics. Nucleic acids research 39:D1156–D1163.

Foolad, M.R. 2007a. Genome mapping and molecular breed-ing of tomato. Int’l J Plant Genomics Vol. 2007, ArticleID 64358, 52 pages. doi:10.1155/2007/64358:1-52.

Foolad, M.R. 2007b. Molecular mapping, marker-assistedselection and map-based cloning in tomato. In: R.K.Varshney and R. Tuberosa (eds.) Genome Assisted CropImprovement. Vol. 2, Genome Applications in Crops. Dor-drecht, The Netherlands: Springer, pp. 307–356.

Frary, A., Nesbitt, T.C., Frary, A., Grandillo, S., vander Knaap, E., Cong, B., Liu, J.-P., Meller, J., Elber,R., Alpert, K.B., et al. 2000. fw2.2: A quantitative traitlocus key to the evolution of tomato fruit size. Science289:85–88.

Fraser, P.D., and Bramley, P.M. 2004. The biosynthesisand nutritional uses of carotenoids. Progress in LipidResearch 43:228–265.

Fraser, P.D., Enfissi, E., and Bramley, P.M. 2008. Geneticengineering of carotenoid formation in tomato fruitand the potential application of systems and syntheticbiology approaches. Arch Biochem and Biophys 483:196-204.

Fraser, P.D., Enfissi, E.M., Goodfellow, M., Eguchi,T., and Bramley, P.M. 2007a. Metabolite profiling ofplant carotenoids using the matrixassisted laser desorp-tion ionization time-of-flight mass spectrometry. Plant J.49:552–564.


Fraser, P.D., Enfissi, E.M.A., Halket, J.M., Truesdale,M.R., Yu, D., Gerrish, C., and Bramley, P.M. 2007b.Manipulation of phytoene levels in tomato fruit: Effectson isoprenoids, plastids, and intermediary metabolism.Plant Cell 19:3194–3211.

Fridman, E., Pleban, T., and Zamir, D. 2000. A recombi-nation hotspot delimits a wild-species quantitative tratilocus for tomato sugar content to 484 bp within an inver-tase gene. Proc. Natl. Acad. Sci. U.S.A. 97:4718–4723.

Gady, A.L.F., Hermans, F.W.K., Van de Wal, M.H.B.J, vanLoo, E.N., Visser, R.G.F., and Bachem, C.W.B. 2009.Implementation of two high through-put techniques ina movel application: Detecting point mutations in largeEMS mutated plant populations. Plant Methods 5:13.

Gady, A.L.F., Vriezen, W.H., Van de Wal, M.H.B.J., Huang,P., Bovy, A.G., Visser, R.G.F., and Bachem, C.W.B.2012. Induced point mutations in the phytoene synthase1 gene cause difference in carotenoid contant duringtomato fruit ripening. Mol Breeding 29:801–812.

Hamilton, J.P., Hansey, C.N., Whitty, B.R., Stoffel,K., Massa, A.N., Van Deynze, A., De Jong, W.S.,Douches, D.S., and Buell, C.R. 2011. Single NucleotidePolymorphism discovery in elite North American potatogermplasm. BMC Genomics 12:302.

Hirschberg, J. 2001. Carotenoid biosynthesis in floweringplants. Current Opinion in Plant Biology 4:210–218.

Kalloo, G. 1991. Genetic improvement of tomato. Berlin andHeidelberg: Springer-Verlag.

Kavanaugh, C.J., Trumbo, P.R., and Ellwood, K.C. 2007.The U.S. Food and Drug Administration’s evidence-based review for qualified health claims: Tomatoes,lycopene, and cancer. J. National Cancer Institute99:1074–1085.

Kinkade, M.P. 2010. Identification and isolation of geneticelements conferring increased tomato fruit lycopene con-tent. PhD diss., Department of Horticulture, The Penn-sylvania State University, University Park, 136 pages.

Kolata, G. 2012. Flavor is price of scarlet hue of tomatoes,study finds. The New York Times, published June 28,2012.

Labate, J.A., Grandillo, S., Fulton, T., Munos, S., Caicedo,A.L., Iris Peralta6, Ji, Y., Chetelat, R.T., Scott, J.W.,Gonzalo, M.J., et al. 2007. Tomato. In: C. Kole (ed.)Genome mapping and molecular breeding in plants.Berlin: Springer-Verlag, pp. 1–125.

Lecomte, L., Duffe, P., Buret, M., Servin, B., Hospital, F.,and Causse, M. 2004. Marker-assisted introgression offive QTLs controlling fruit quality traits into three tomatolines revealed interactions between QTLs and geneticbackgrounds. Theor Appl Genet 109:658–668.

Lewis, S., Searle, S., Harris, N., Gibson, M., Iyer, V.,Richter, J., Wiel, C., Bayraktaroglu, L., Birney, E.,Crosby, M., et al. 2002. Apollo: A sequence annota-tion editor. Genome Biol. 33 (12):research0082.0081-0082.0014.

Lippman, Z., Semel, Y., and Zamir, D. 2007. An integratedview of quantitative trait variation using tomato interspe-

cific introgression lines. Current Opinion in Genetics andDevelopment 17:545–552.

Liu, Y.-S., Gur, A., Ronen, G., Causse, M., Damidaux,R., Duffe, P., Buret, M., Hirschberg, J., and Zamir, D.2003. There is more to tomato fruit colour than candidatecarotenoid genes. Plant Biotech. J. 1:195–207.

Martin, G.B., Brommonschenkel, S.H., Chunwongse, J.,Frary, A., Ganal, M.W., Spivey, R., Wu, T., Earle,E.D., and Tanksley, S.D. 1993a. Map-based cloning ofa protein kinase gene conferring disease resistance intomato. Science 262:1432–1436.

Martin, G.B., Vicente, M.C.d., and Tanksley, S.D. 1993b.High-resolution linkage analysis and physical chrarac-terization of the Pto bacterial resistance locus in tomato.Mol. Plant-Micr. Interact. 6:26–34.

Menda, N., Semel, Y., Peled, D., Eshed, Y., and Zamir, D.2004. In silico screening of a saturated mtuation libraryof tomato. Plant J. 38:861–872.

Minoia, S., Petrozza, A., D’Onofrio, O., Piron, F., Mosca,G., Sozio, G., Cellini, F., Bendahmane, A., and CarrieroF. 2010. A new mutant genetic resource for tomato cropimprovement by TILLING technology. BMC ResearchNotes 3:69.

Mueller, L.A., Lankhorst, R.K., Tanksley, S.D., Giovan-noni, J.J., White, R., Vrebalov, J., Fei, Z., van Eck, J.,Buels, R., Mills, A.A., et al. 2009. A snapshot of theemerging tomato genome sequence. The Plant Genome2:78–92.

Mueller, L.A., Solow, T.H., Taylor, N., Skwarecki, B.,Buels, R., Binns, J., Lin, C.-W., Wright, M.H., Ahrens,R., Wang, Y., et al. 2005. The SOL genomics net-work: A comparative resource for Solanaceae biologyand beyond. Plant Physiol. 138:1310–1317.

Myers, C., Robson, D., Wible, A., Hibbs, M., Chiriac, C.,Theesfeld, C., Dolinski, K., and Troyanskaya, O. 2005.Discovery of biological networks from diverse functionalgenomic data. Genome Biology 6:R114.

Okabe, Y., Asamizu, E., Saito, T., Matsukura, C. Ariizumi,T., Bres, C., Rothan, C., Mizoguchi, T., and Ezura, H.2011. Tomato TILLING technology: Development of areverse genetics tool for the efficient isolation of mutantsfrom Micro-Tom mutant libraries. Plant Cell Physiol.52:1994–2005.

Overy, S.A., Walker, H.J., Malone, S., Howard, T.P., Baxter,C.J., Sweetlove, L.J., Hill, S.A., and Quick, W.P. 2005.Application of metabolite profiling to the identificationof traits in a population of tomato introgression lines.Journal of Experimental Botany 56:287–296.

Peralta, I.E., and Spooner, D.M. 2005. History, origin andearly cultivation of tomato (Solanaceae). In: M.K. Raz-dan and A.K. Mattoo (eds.) Genetic improvement ofsolanaceous crops: Tomato. Enfield, NH: Science Pub-lishers, pp. 1–24.

Piron, F., Nicolai, M., Minoia, S., Piednoir, E., Moretti,A., Salgues, A., Zamir, D, Caranta, C. Bendahmane,A. 2010. An induced mutation in tomato eIF4E leads toimmunity to two potyviruses. PLoS ONE 5:e11313.


Rao, A.V. 2004. Processed tomato products as a source ofdietary lycopene: Bioavailability and antioxidant proper-ties. Canadian Journal of Dietetic Practice and Research65:161–165.

Rao, A.V., and Rao, L.G. 2007. Carotenoids and humanhealth. Pharmacological Research 55:207–216.

Rick, C.M. 1976a. Natural variability in wild species ofLycopersicon and its bearing on tomato breeding. Genet.Agraria 30:249–259.

Rick, C.M. 1976b. Tomato, Lycopersicon esculentum(Solanaceae). In: N.W. Simmonds (ed.), Evolu-tion of crop plants. London: Longman, pp. 268–273.

Rick, C.M. 1978. The tomato. Sci. Amer. 23:76–87.Rick, C.M. 1979. Potential improvement of tomatoes by

controlled introgression of genes from wild speies. In:Proc. Conf. Broadening Genetic Base of Crops. Pudoc:Wageningen, pp. 167–173.

Rick, C.M., and Fobes, J.F. 1974. Association of an allozymewith nematode resistance. Rpt. Tomato Genetic Coop.24:25.

Rick, C.M., and Fobes, J.F. 1975. Allozyme variation in thecultivated tomato and closely related species. Bul. TorreyBot. Club 102:376–384.

Rigola, D., van Oeveren, J., Janssen, A., Bonne, A.,Schneiders, H., van der Poel, H.J.A., van Oursouw, N.J.,Hogers, R.C.J., de Both, M.T.J., and van Eijk, M.J.T.2009. High-throughput detection of induced mutationsand natural variation using KeyPoint technology. PLoSONE 4:e4761.

Ronen, G., Carmel-Goren, L., Zamir, D., and Hirschberg, J.2000. An alternative pathway to beta-carotene formationin plant chloroplast discovered by map-based cloning ofBeta and old-gold color muation in tomato. Proc NatlAcad Sci USA 97:11102–11107.

Ronen, G., Cohen, M., Zamir, D., and Hirschberg,J. 1999. Regulation of carotenoid biosyntheis duringtomato fruit development: Expression of the gene forlycopene epsilon-cyclase is down regulated during ripen-ing and is elevated in the mutant Delta. Plant J. 17:341–351.

Rousseaux, M.C., Jones, C.M., Adams, D., Chetelat, R.,Bennett, A., and Powell, A. 2005. QTL analysis offruit antioxidants in tomato using Lycopersicon pennel-lii introgression lines. Theoretical and Applied Genetics111:1396–1408.

Schaffer, A.A., Levin, I., Oguz, I., Petreikov, M., Cin-carevsky, F., Yeselson, Y., Shen, S., Gilboa, N., andBar, M. 2000. ADPglucose pyrophosphorylase activ-ity and starch accumulation in immature tomato fruit:The effect of a Lycopersicon hirsutum-derived introgres-sion encoding for the large subunit. Plant SciencE 152:135–144.

Schauer, N., Semel, Y., Roessner, U., Gur, A., Balbo, I.,Carrari, F., Pleban, T., Perez-Melis, A., Bruedigam,C., Kopka, J., et al. 2006. Comprehensive metabolicprofiling and phenotyping of interspecific introgression

lines for tomato improvement. Nat Biotech 24:447-454.

Schauer, N., Semelb, Y., Balboa, I., Steinfatha, M., Repsil-bera, D., Selbiga, J., Plebanb, T., Zamir, D., and Fernie,A.R. 2008. Mode of inheritance of primary metabolictraits in tomato. The Plant Cell 20:509–523

Schauer, N., Zamir, D., and Fernie, A.R. 2005. Metabolicprofiling of leaves and fruit of wild species tomato: Asurvey of the Solanum lycopersicum complex. J. Exp.Bot. 56:297–307.

Semel, Y., Nissenbaum, J., Menda, N., Zinder, M., Krieger,U., Issman, N., Pleban, T., Lippman, Z., Gur, A., andZamir, D. 2006. Overdominant quantitative trait locifor yield and fitness in tomato. Proc. Natl. Acad. of Sci.U.S.A. 103:12981–12986.

Sim, S., Durstewitz, G., Plieske, J., Wieseke, R., Ganal,M., Van Deynze, A., Hamilton, J., Buell, C.R., Causse,M., Wijeratne, S., and Francis, D.M. 2012. Develop-ment of a large SNP genotyping array and generationof high-density genetic maps in tomato. PLoS ONE 7:e40563.

Stommel, J.R., Abbott, J., Saftner, R.A., and Camp, M.2005. Sensory and objective quality attributes of beta-carotene- and lycopene-rich tomato fruit. J. Amer. Soc.Hort. Sci. 130:244–251.

Tanksley, S.D., Medina-Filho, H., and Rick, C.M. 1982. Useof naturally-occuring enzyme variation to detect and mapgenes controlling quantitative traits in an interspecificcross of tomato. Heredity 49:11–25.

Tanksley, S.D., and Orton, T.J. 1983. Isozymes in plantgenetics and breeding. Amsterdam: Elsevier.

Taylor, J.B. 1986. Biosystematics of the tomato. In: J.G.Atherton, and J. Rudich (eds.) The tomato crop: A sci-entific basis for crop improvement. New York: Chapman& Hall, pp. 1–34.

Tomato Genome Consortium. 2012. The tomato genomesequence provides insights into fleshy fruit evolution.Nature 485(7400):635–641.

Tuberosa, R., and Salvi, S. 2006. Genomics approachesto improve drought tolerance in crops. Trends in PlantScience 11:415–412.

Tuberosa, R., Salvi, S., Giuliani, S., Sanguineti, M.C.,Bellotti, M., Conti, S., and Landi, P. 2007. Genome-wideapproaches to investigate and improve maize response todrought. Crop Science 47:1–23.

USDA. 2012. National Nutrient Database for Standard Ref-erence. http://ndb.nal.usda.gov.

Vershinin, A. 1999. Biological functions of carotenoids:Diversity and evolution. Biofactors 10:99–104.

Von Low, E.C., Perabo, F.G., Siener, R., and Muller,S.C. 2007. Facts and fiction of phytotherapy for prostatecancer: A critical assessment of preclinical and clinicaldata. In Vivo 21:189–204.

Wikstrom, N., Savolainen, V., and Chase, M.W. 2001.Evolution of the angiosperms: Calibrating the familytree. Proc. R. Soc. Lond. Ser. B. Biol. Sci. 268:2211-2220.


Wu, K., Erdman, J.W., Schwartz, S.J., Platz, E.A., Leitz-mann, M., Clinton, S.K., DeGroff, V., Willett, W.C., andGiovannucci, E. 2004. Plasma and dietary carotenoids,and the risk of prostate cancer. Cancer EpidemiologyBiomarkers & Prevention 13:260–269.

Yates, H.E., Frary, A., Doganlar, S., Frampton, A.J., Ean-netta, N.T., Uhlig, J., and Tanksley, S.D. 2004. Compar-ative fine mapping of fruit quality QTLs on chromosome4 introgression derived from two wild tomato species.Euphytica 135:283–296.

Documents

Translational Genomics for Crop Breeding (Abiotic Stress, Yield and Quality) || Genomics-Assisted Breeding for Tomato Fruit Quality in the Next-Generation Omics Age