tomato fruits and result in bicolor and yellow
phenotypes
Bulot, B., Isabelle, S., and Goulet, C.
Département de Phytologie, Faculté des Sciences de l’Agriculture et de l’Alimentation, Université Laval. Pavillon Envirotron, 2480 boulevard Hochelaga, Québec QC, G1V0A6, Canada
3. 1. Résumé
Les caroténoïdes sont des pigments qui peuvent s’accumuler dans les plastes des fleurs et des fruits et ainsi contribuer à leur couleur caractéristique. Le lycopène est par exemple le caroténoïde responsable de la couleur rouge des tomates. Cependant, certains cultivars, appelés « bicolores », accumulent le lycopène de façon localisée et ne deviennent jamais totalement rouges. Nous démontrons dans cette étude que le gène de la phytoène synthase 1 (PSY1), premier gène du sentier de production des caroténoïdes, est responsable de ce patron de coloration. Solanum habrochaites est une espèce sauvage apparentée à la tomate qui produit des fruits verts. Nous avons déterminé, pour la lignée LA3923, que l’introgression du génome de S. habrochaites au niveau du locus contenant PSY1 est responsable d’une baisse de l’expression de PSY1 et donc du phénotype bicolore des fruits. Dans le cas des cultivars bicolores de S. lycopersicum, le phénotype des fruits est causé une délétion de 3787 pb en amont du gène PSY1. Cette délétion élimine une partie de la région 5’UTR du gène, ce qui affecte vraisemblablement l’efficacité de sa traduction et empêche la production de caroténoïdes de façon normale. Nous avons pu également déterminer que la mutation ry, située à l’extrémité 3’ du gène
PSY1, est le résultat d’un réarrangement chromosomique complexe entre le dernier exon du gène PSY1 et le début du gène retrouvé en aval. Les fruits jaunes portant cette mutation, au contraire des cultivars jaunes caractérisés par l’insertion d’un LTR du rétrotransposon Rider au niveau de l’exon 1 de PSY1 (mutation r), peuvent dans certaines conditions accumuler du lycopène, bien qu’en très faible quantité par rapport aux cultivars bicolores. Ces informations témoignent de l’importance du gène PSY1 dans le sentier de
production des caroténoïdes, et de la façon dont différentes variations structurelles dans ses séquences promotrices et codantes peuvent affecter la couleur des fruits.
3.2 Abstract
Carotenoids are pigments that can accumulate in flowers and fruits plastids to contribute to their characteristic colors. For example, lycopene is the carotenoid that gives tomatoes their red color. Some tomato cultivars however present a particular phenotype called « bicolor ». They produce yellow tomatoes with red sections, especially at the blossom-end of the fruit. In this study, we show that phytoene synthase 1 gene (PSY1), the first gene of carotenoid production pathway is responsible for bicolor phenotype.
Solanum habrochaites is a tomato wild related species that produces green fruits. In the
line LA3923, the introgression of S. habrochaites genome at PSY1 locus is responsible for the down-regulation of the gene, and thus the bicolor phenotype of the fruits. In S.
lycopersicum bicolor cultivars, a 3787 bp deletion in PSY1 promoter causes the same
coloration pattern. Since the deletion contains a part of the 5’UTR region of PSY1, translation efficiency is likely decreased. Furthermore, we identified that PSY1 ry mutation
is caused by a genomic rearrangement between PSY1 last exon and the first exons of the downstream gene. Since in yellow ry tomatoes the mutation happens at the end of the
gene, fruits are still able in certain conditions to accumulate lycopene, though to a lesser extent than in bicolor cultivars. In contrast, yellow fruits with the r mutation, which consists of the insertion in PSY1 exon 1 of one LTR of the Rider transposon, have a totally non- functional protein. These results attest of PSY1 importance in carotenoid production pathway, and the way different structural variations in the promoter or coding sequences of the gene can affect fruits colors.
3.3 Introduction
Carotenoids, a very diversified and important group of pigments in life, are naturally synthesized by plants, algae, fungi and bacteria (Bartley et al. 1994). Carotenoids can be found in chloroplasts, where they play a role in photoprotection and light harvesting for photosynthesis (Young 1991, Horton and Ruban 2004). Carotenoids also accumulate in
chromoplasts, in specialized structures like plastoglobules, crystals, tubules, and inner membrane, where they are responsible of the color, from yellow to red that can be displayed by flowers and fruits (Klee and Giovannoni 2011, Ljubesić et al. 1991). These bright colors help plants to attract pollinators or frugivorous animals for seed dissemination
(Dicke and Baldwin 2010). Tomatoes are known to accumulate important quantities of carotenoids in their chromoplasts and more specifically lycopene which confers to the fruits their red color (Li and Yuan 2013).
Carotenoids can also be the precursors of important molecules like phytohormones ABA and strigolactones, as well as several volatile compounds involved in many fruits aroma (North et al. 2007, López-Ráez et al. 2008, Ohmiya 2009). These molecules are called apocarotenoids because they are formed by the oxidative cleavage of carotenoids. In plants, CCDs enzymes (Carotenoid Cleavage Dioxygenase) are known to catalyze this oxidative cleavage. For example, CCDs can use lycopene as a substrate to produce the volatile 6-methyl-5-hepten-2-one (MHO) with a citrus-like aroma. The same enzymes can cleave ζ-carotene to form geranylacetone, a volatile with a fruity and fresh aroma. Apocarotenoid volatiles are known to have a positive impact on tomato flavor and increase the sweetness perception (Vogel et al. 2010, Tieman et al. 2012, Bartoshuk and Klee 2013). In addition to the apocarotenoids, about 30 volatile compounds derived from fatty acids and amino acids are thought to be essential to the distinct flavour of tomatoes
(Ohmiya 2009, Ilg et al. 2014).
The significant color change of tomatoes during ripening, as well as the availability of a large population of colored mutants make the species particularly valuable for researches on the carotenoid production pathway, which has been extensively studied. Carotenoid synthesis starts with the condensation of two molecules of geranyl geranyl diphosphate by the enzyme phytoene synthase (PSY). This reaction generates the first carotenoid, 15-cis-phytoene which is also colorless (Bartley et al. 1992). In tomatoes, three isoforms of phytoene synthase have been identified with different spatio-temporal expression patterns. The isoform PSY1 is expressed in the ripening fruits, while PSY2 and PSY3 are respectively present in leaves and roots (Giorio et al. 2008). This phytoene synthase reaction is considered as one of most important regulatory step of the carotenoid pathway (Fraser et al. 2007). The carotenoid 15-cis-phytoene is then modified by a phytoene desaturase (PDS), yielding successively 15-9’-di-cis-phytofluene and 9,15,9’-tri-
cis-ζ-carotene (Pecker et al. 1992). Subsequently, the ζ-carotene isomerase enzyme
(ZISO) converts 9,15,9’-tri-cis-ζ-carotene in 9,9’-di-cis-ζ-carotene, which goes through two other desaturation reactions, managed by a ζ-carotene desaturase (ZDS) (Hirschberg 2001, Bouvier et al. 2005). The resulting molecule, 7,9,7,9’-tetra-cis-lycopene
(prolycopene), has its configuration transformed from cis to trans by a carotenoid isomerase (CRTISO) in order to generate the carotenoid all-trans-lycopene, also called lycopene (Isaacson et al. 2002). At this point, the pathway can take two directions depending of the type of rings formed. The lycopene-β-cyclase (LYCB) catalyses one or two cyclisation to yield γ-carotene and β-carotene respectively. Alternatively, the lycopene can modified by the lycopene-ε-cyclase (LYCE) to produce δ-carotene, who can then become α-carotene by the subsequent addition of one β-ring by the lycopene-β-cyclase
(Fantini et al. 2013).
A few tomato cultivars produce bicolored fruits that are mostly yellow at maturity but with contrasted red flesh sections. In some cultivars, these red zones are limited to the blossom end of the fruit and only appear occasionally, while other cultivars show a more consistent coloration with red streaks that can be observed in every fruits. MacArthur and Young described the first group as yellow tomatoes showing pink colors in hot summer weather (Young and MacArthur 1947). This type of coloration was suggested to be linked to the same locus as the yellow fruits (locus r) and the locus was named ry(Young 1956). Phytoene synthase 1 (PSY1) was later on identified as the gene responsible for the yellow phenotype, and the color originates from the absence of carotenoid synthesis during the ripening of the fruit (Fray and Grierson 1993). The PSY1 loss of function causes a reduction of more than 99% of lycopene content in fruit, that passes from 107,5 mg.g FW-1 in red fruits to 0,7 mg.g FW-1 in yellow fruits (Rêgo et al. 1999). The r mutant was described has having a shorter transcript and the ry mutant was lacking the last 25 amino acids that were replaced by an unrelated sequence of 23 amino acids (Fray and Grierson 1993). In contrast to the ry locus that was well studied, little is known about the second group of bicolor tomatoes with a stronger phenotype. In this study, we investigate further the genetic basis for bicolored fruits in tomato using heirlooms cultivars and introgression lines derived the wild related species S. habrochaites.
3.4 Materials and Methods
Growth of plant materials
All tomato introgression lines and cultivars were grown in University Laval, Québec. Tomato seeds were first planted in greenhouse for germination. Plants selected were then grown in greenhouse or in the field until they produced ripe fruits. The bicolor introgression line 3923 is derived from a cross between S. lycopersicum (cultivar E6203, accession LA4024) and S. habrochaites (LA1777) (Monforte and Tanksley 2000). A new introgression population was then developed between S. lycopersicum (cultivar E6203, accession 4024) and LA3923. The two parents and the F1 generation were grown in the field (summer 2014). The F2 and F3 generations of this new introgression population were respectively grown in greenhouse (autumn 2015) and in the field (summer 2016). F1 and F2 generations of LA3923 previously selected for possessing a fragment of S.
habrochaites (LA1777) genome at the top of chromosome 3 (3923-locus 3 and 3923-locus
3a lines) or at the bottom of chromosome 2 (3923-locus 2 line) were respectively grown in greenhouse (winter 2016) and in the field (summer 2017). Plants in greenhouse were grown in an entirely random design. In the field, plants were placed following randomized replicated plots with three plots of three to five plants. The S. lycopersicum parent (4024) was present in all designs as a control. Bicolor cultivars (Granada, Grapefruit, Isis Candy, Mom’s, Mortgage Lifter bicolor, Mr. Stripey and Striped German), red cultivars (Ailsa Craig, Ballada, Gigantesque, Gregori’s Altai, Homer’s German Oxheart, Mexico Midget, Pruden’s Purple, Vendor and LA4024) and yellow cultivars (Barry Crazy Cherry, Butter Apple, Gallina’s Yellow, Gold Currant, Lemon Drop, Mirabelle, Poma Amoris Minora Lutea and Téton de Vénus jaune) were grown in the field over 3 summers (2015-2017) in randomized replicated plots of two plots of five plants. The F2 population derived from the cross between Granada (bicolor) and Azoychka (yellow ry) was grown in greenhouse (winter
2017).
Volatile compounds analysis
Ripe fruits of LA4024, 3923-locus 3a and 3923-locus 2 lines were harvested, pooled and chopped before 100g were placed in glass tubes as described by Schmelz et
al. 2003. During 1 hour, hydrocarbon filtered air (Agilent, www.agilent.com) arrived at one extremity of the tube, carried volatiles released by tomatoes and led them to be trapped by
a divinylbenzene resin column (HayeSep Q) placed at the other extremity of the tube. Volatiles trapped on the HayeSep Q column were eluted by a solution composed of methylene chloride and the internal standard nonyl acetate. Volatile samples were then analyzed by a gas chromatograph (Agilent 6890) with a DB-5MS UI column (30m, 0.250nm diam., Agilent, www.agilent.com). Each volatile compound (Agilent,
www.agilent.com) was identified by known standards and using a gas chromatograph coupled to a mass spectrometer (GC-MS, Agilent 5977B) (Agilent, www.agilent.com). For each volatile compound, one-way ANOVA followed by Tukey tests were performed in order to assess which lines or cultivars were significantly different from the others. When necessary, statistical analyses were carried out on log10 transformed data to correct for variance homogeneity.
Genetic markers design
InDel and HRM markers were designed by comparing the genome sequences of S.
lycopersicum (version SL3.0 of the Heinz genome assembly (Fernandez-Pozo et al. 2015)), and S. habrochaites (cv. LYC4) (SAMEA3138934, Tomato Genome Consortium 2012). Homologous sequences were aligned with MAFFT program (Katoh and Standley 2013) to find an insertion or a deletion (InDel marker), or a SNP (HRM marker) . The mutation had to be flanked by conserved sequences between the two species to insure proper amplification by PCR. The primers were designed with Primer3 (Rozen and Skaletsky 2000). The sequences of the primers are available on Supplementary Table
3.10. For the F2 population derived from Granada x Azoychka crosses, the A to G
substitution mutation presents in PSY1 intron 4 of r yellow cultivars was used to predict the F2 segregation ratio with genetic markers (Supplementary Figure 3.11, Supplementary
Table 3.10).
Quantitative PCR
Ripe fruits of LA4024, 3923-locus 3a and 3923-locus 2 lines were frozen in liquid nitrogen and store at -80°C. Total RNA was extracted following the EZ-10 Spin Column Plant RNA Miniprep Kit protocol (Biobasic, www.biobasic.com). Possible genomic DNA contamination was removed by DNAse treatment step (Qiagen, www.qiagen.com) before purifying total RNA with a second run of the EZ-10 Spin Column Plant RNA Miniprep Kit protocol. Quantification of total RNA was evaluated on NanoDrop (ThermoFisher,
Step qRT-PCR Kit (QuantaBio, www.quantabio.com) from 100ng of total RNA. The primers were designed for the amplification of PSY1 mRNA. In order to be sure to only amplify mRNA and not genomic DNA, the forward primer was designed to begin at the end of exon 4 and finish at the beginning of exon 5, while the reverse primer was located at the beginning of exon 6. The sequences and positions of the primers are available in
Supplementary Figure 3.10. The quantification of PSY1 transcripts was determined with
a standard curve. Expression levels have been compared with Tukey test. Significant differences are indicated with letters (p<0.05).
Transcriptome sequencing and analysis
Total RNA was extracted from ripe fruits of on ripe fruits of LA4024, 3923-locus 3a, red, bicolor and yellow cultivars with the same protocol as for qPCR. The RNA-Seq libraries were prepared following the Zhong et al. 2011 protocol that has been modified and adapted for paired-end 125 pb RNA-Seq (Supplementary Figure 3.12). The libraries were sent to McGill University sequencing platform (www.gqinnovationcenter.com) to be analyzed on a Bioanalyzer (Agilent, www.agilent.com) and sequenced paired-end 125 bp with HiSeq 4000 Illumina technology (Illumina, www.illumina.com). The quality of the sequencing data was then evaluated with FastQC program. The bad qualities reads and Illumina adapters were removed with Trimmomatic (Bolger et al. 2014). The reads were then aligned to the Solanum lycopersicum reference genome (version SL3.0) available at
Sol Genomics Network (Fernandez-Pozo et al. 2015) with Hisat2 (Kim et al. 2015) and gene expression (FPKM) was assessed with Stringtie (Pertea et al. 2015). The packages edgeR (Robinson et al. 2010), DESeq2 (Love et al. 2014) and Limma-voom (Ritchie et al. 2015) from the Bioconductor repertory of R (Bionconductor, www.bioconductor.org) were then used to perform the differential gene expression analysis. PSY1 expression levels have been compared with Tukey test. SNPs calling in 4024 and 3923-locus 3a lines was performed with SAMtools and bcftools (Li et al. 2009), and SNP density was evaluated with Sniplay(Dereeper et al. 2011). The program regtools was used in order to extract the splicing variants that can be present in each samples, and visualization was performed thanks to the IGV software (Robinson et al. 2011). Yellow cultivars PSY1 sequences were compared with r (X67143.1) and ry (X67144.1) mutant sequences available on NCBI. SRA
data of yellow cultivars from project PRJNA353161(Tieman et al. 2017) were analyzed to study the two versions of PSY1 reads in the ry mutation.
3.5 Results
Mapping of a QTL for bicolored fruits
Fruits of the tomato related wild species Solanum habrochaites do not accumulate lycopene and stay green even at maturity (Grumet et al. 1981). This particularity makes introgression lines derived from S. habrochaites very useful to identify QTL involved in carotenoid synthesis. All the lines of S. habrochaites introgression populations possess small regions of S. habrochaites genome in a Solanum lycopersicum background, and a phenotypic difference between an introgression line and the tomato parent can be attributed to genes that are located in the wild species introgressed region. In the introgression population between S. lycopersicum (cultivar E-6203, accession LA4024) and the related wild species S. habrochaites (accession LA1777) developed by Monforte and Tanksley (2000), the lines 3922 and 3923 have been characterized to produce bicolor fruits that accumulate 88% less lycopene than the 4024 red parent. They also are the only lines that possess one introgression of S. habrochaites genome at the end of the chromosome 2, and it’s the only introgressed locus that they possess (Monforte and Tanksley 2000, Mathieu et al. 2009). Since in S. habrochaites fruits, the lycopene production pathway isn’t active, it has been hypothesized that this specific chromosome 2 locus contains a potential QTL involved in lycopene synthesis in fruits (Mathieu et al. 2008). The line LA3923 possesses a shorter S. habrochaites introgression segment than the line LA3922 and was selected to reduce furthers the region responsible for the bicolored phenotype. The heterozygous F1 plants resulting from the cross between LA4024 and LA3923 produced red tomatoes confirming that the red allele is dominant over the bicolor allele. New smaller introgression lines between LA4024 and LA3923 bicolor line were selected using InDel markers. One line showed a phenotype that was incongruous with the recessive nature of the bicolor alleles and we quickly conclude that the phenotype was not linked to the introgression on chromosome 2 (Supplementary Table 3.1). The initial map of the introgression lines was done with a relatively small number of markers for each chromosome and small-undetected introgressions likely persisted in some of the lines. Since alleles of PSY1 are responsible for the yellow phenotypes, a high resolution- melting (HRM) marker was designed in the first exon of PSY1 to detect if LA3923 is having one of those undetected introgressions on chromosome 3. The genotyping results at the PSY1 markers showed that plants producing red tomatoes were either homozygous for S.
lycopersicum genome or heterozygous, while plants with bicolor fruits were all
homozygous for S. habrochaites genome (Supplementary Table 3.2). These results suggested that the QTL is linked to PSY1 or located near the gene on chromosome 3. New introgression lines were selected from the cross between LA4024 and LA3923 to isolate the introgression on chromosome 2 (LA3923-locus 2) and chromosome 3 (LA3923- locus 3). The introgression line on chromosome 3 has bicolored fruits while the introgression line on chromosome 2 has red fruits. The line 3923-locus 3 has a small S.
habrochaites introgression on top of chromosome 3 between 4,006,743 and 8,455,278 bp.
The locus was further reduced by using a smaller introgression (LA3923-locus 3a)
(Supplementary Table 3.3). The RNA sequencing results of the parental line LA4024 and
the line LA3923-locus3a were used to perform a SNP density analysis. Thanks to the high level of SNPs existing between S. lycopersicum and S. habrochaites, the size of the lycopene related QTL was reduced to about 690 kbp between position 4,074,672 bp and 4,764,329 bp. Less than 75 genes are present on this locus (Supplementary Table 3.4) and only PSY1 seems related to the carotenoid pathway. SNP density analysis was also used to verify if other small-undetected introgression were present on other chromosomes. No introgression other than the one on chromosome 3 was identified by the analysis.
Apocarotenoid volatiles emission in bicolor fruits
In order to determine the impact of carotenoid accumulation on apocarotenoid volatiles emission, volatile analysis was performed on ripe fruits from LA4024, 3923-locus 2 and 3923-locus 3a plants. The bicolor line 3923-locus 3a showed a 93% and 98% decrease in 6-methyl-5-hepten-2-one and geranylacetone respectively compared to the parental line LA4024 (Fig. 3.1). The red-fruited line 3923-locus 2 shows a smaller but significant reduction of both volatiles (30% and 39% respectively). It is likely that the locus on chromosome 2 contains a gene influencing the production of apocarotenoid volatiles.
Figure 3.1. Apocarotenoid volatiles emission from fruits of bicolor introgression line.
Emission of the apocarotenoids geranylacetone (A) and 6-methyl-5-hepten-2-one (B) (±SE, n=6). The parental line LA4024 has red fruits while the introgression line LA3923 derived from S. habrochaites has bicolor fruits. A subset of lines was developed to isolate each of the wild species segments from LA3923; 3923 locus 2 have a segment at the bottom of chromosome 2 while 3923 locus 3a have a small fragment at the top of chromosome 3. The segment on chromosome 3 is responsible for the bicolor phenotype. Yellow and red sections of fruits from 3923-locus 3a plants have been isolated to be analyzed separately (±SE, n=4). Significant differences between the different lines are indicated with letters (p<0.05).
To determine if there is a difference in the emission of apocarotenoids in the red and