Cross-reference to Related Applications

This application claims the benefit of United States Provisional Patent Application USSN 62/578,678 filed October 30, 2017, the entire contents of which is herein incorporated by reference.

Field

This application relates to plant biotechnology, in particular to biochemically altering flavor in tomatoes.

Background Tomato flavor is a complex trait comprising a combination of texture, taste (sweet, bitter, sour) and aroma (smoky, hay, vegetal/grassy, earthy) determined by sugars, organic acids, and a cocktail of aroma volatiles that also modulate flavor intensity and sweetness. Several of the aroma compounds are positively associated with consumer preference and others detract from consumer liking (Tieman et al, 2012). Many genes responsible for the production of volatile aroma compounds are known, and can otherwise be identified using traditional map-based cloning approaches or more advanced genomics techniques (Tieman et al, 2017). Cultivated tomato genetics are relatively narrow and volatile profiles of ripe commercial tomatoes are equally as uniform, dominated by "grassy" C6 volatiles like 3-hexenal and hexanal with coincidently low levels of others such as the C5 volatiles, compounds known to be associated with greater consumer liking at higher concentrations (Tieman et al, 2012). Moreover, many cultivated tomatoes accumulate aroma volatiles known to have negative affects on consumer preference, like methyl salicylate and acetate esters such as 2-methylbutyl acetate and butyl acetate (Vogel et al, 2010; Tieman et al, 2010; Goulet et al, 2015). Further, because cultivated tomato genetics are very narrow, there is little variability in aroma related genes, most commercial breeding programs having selected for production traits (i.e. yield, disease resistance, firmness for shipping and shelf life) over objectively superior flavor. As a result, customers are dissatisfied with the flavor of cultivated tomatoes.

Several researchers have used transgenic approaches to improve flavor, for example by inserting an aroma-related gene into the genome and over-expressing the gene using strong promoter. More recently, methods have been developed that focus on associating volatile compounds with consumer liking, then using available varieties/accessions with optimal aroma profiles as parental lines for breeding. Resultant progeny that maintain these ideal volatiles profiles can then be selected for, promoting the development of consumer-preferred fruit. However, all of these approaches rely on transgenic genetic engineering, or alleles of volatile biosynthetic genes that exist in available germplasm. There is very little variation in these genes in cultivated varieties, and heirloom or wild relatives of tomato may be a good source of genetic diversity but can be problematic as parents in a commercial breeding program as many other unwanted traits must then be bred out due to linkage drag.

There remains a need for cultivated tomatoes with differentiated and/or improved flavor, while maintaining good production traits, and a less problematic approach to obtaining cultivated tomatoes with differentiated and/or improved flavor.

Summary

In one aspect, there is provided a polynucleotide comprising: the nucleotide sequence as set forth in SEQ ID NO: 1 , wherein a codon of SEQ ID NO: 1 encoding proline at position 239 of the amino acid sequence as set forth in SEQ ID NO: 2 is replaced with a codon encoding leucine; the nucleotide sequence as set forth in SEQ ID NO: 1 , wherein a codon of SEQ ID NO: 1 encoding tryptophan at position 39 of the amino acid sequence as set forth in SEQ ID NO: 2 is replaced with a stop codon; the nucleotide sequence as set forth in SEQ ID NO: 3, wherein a codon of SEQ ID NO: 3 encoding tryptophan at position 526 of the amino acid sequence as set forth in SEQ ID NO: 4 is replaced with a stop codon; the nucleotide sequence as set forth in SEQ ID NO: 5, wherein a codon of SEQ ID NO: 5 encoding valine at position 68 of the amino acid sequence as set forth in SEQ ID NO: 6 is replaced with a codon encoding methionine; or, the nucleotide sequence as set forth in SEQ ID NO: 5, wherein a codon of SEQ ID NO: 5 encoding glutamine at position 333 of the amino acid sequence as set forth in SEQ ID NO: 6 is replaced with a stop codon.

In another aspect, there is provided a plant cell comprising an aroma-related gene comprising a polynucleotide as defined above.

In another aspect, there is provided a tomato seed comprising a plant cell as defined above. In another aspect, there is provided a tomato plant comprising a plant cell as defined above. In another aspect, there is provided a tomato fruit of the tomato plant as defined above.

In another aspect, there is provided a method of differentiating flavor profile of a tomato fruit, the method comprising: detecting a mutation in an aroma-related gene in a plant of a mutagenized population of tomato plants; cultivating the plant having the mutation in the aroma-related gene to produce a tomato fruit having a differentiated flavor profile in comparison to a tomato fruit of a wildtype tomato plant.

In another aspect, there is provided a use of mutagenesis to identify a tomato plant having fruit with a differentiated flavor profile in comparison to fruit of a wildtype tomato plant.

Further features will be described or will become apparent in the course of the following detailed description. It should be understood that each feature described herein may be utilized in any combination with any one or more of the other described features, and that each feature does not necessarily rely on the presence of another feature except where evident to one of skill in the art.

Brief Description of the Drawings

For clearer understanding, preferred embodiments will now be described in detail by way of example, with reference to the accompanying drawings, in which:

Fig. 1 A to Fig. 1 E depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of C5 volatile compounds emitted from ripe fruit of hpl1 variant W39 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA). The C5 volatile compounds are E-2-pentenal (Fig. 1 A), 3-pentanone (Fig. 1 B), 1 -penten-3-ol (Fig. 1 C), 1 -penten-3-one (Fig. 1 D) and Z-2-penten-1 -ol (Fig. 1 E).

Fig. 2A to Fig. 2C depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of C6 volatile compounds emitted from ripe fruit of hpl1 variant W39 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA). The C6 volatile compounds are Z-3-hexen-1 -ol (Fig. 2A), E-2-hexenal (Fig. 2B) and 1 -hexanol (Fig. 2C).

Fig. 3A to Fig. 3E depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of C5 volatile compounds emitted from ripe fruit of hpl1 variant P239L segregants (aa) of BC0F2 populations compared to wildtype segregants (AA). The C5 volatile compounds are E-2-pentenal (Fig. 3A), 3-pentanone (Fig. 3B), 1 -penten-3-ol (Fig. 3C), 1 -penten-3-one (Fig. 3D) and Z-2-penten-1 -ol (Fig. 3E).

Fig. 4A to Fig. 4C depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of C6 volatile compounds emitted from ripe fruit of hpl1 variant P239L segregants (aa) of BC0F2 populations compared to wildtype segregants (AA). The C6 volatile compounds are Z-3-hexen-1 -ol (Fig. 4A), E-2-hexenal (Fig. 4B) and 1 -hexanol (Fig. 4C).

Fig. 5A to Fig. 5C depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of selected C6 volatile compounds emitted from ripe fruit of loxC variant W526 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA). The C6 volatile compounds are Z-3-hexen-1 -ol (Fig. 5A), E-2-hexenal (Fig. 5B) and 1 -hexanol (Fig. 5C).

Fig. 6A to Fig. 6D depict graphs showing relative abundances of other C6 volatile compounds emitted from ripe fruit of loxC variant W526 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA). The other C6 volatile compounds are 3-hexenal (Fig. 6A), hexanal (Fig. 6B), E-2-hexen-1 -ol (Fig. 6C) and Z-3-hexen-1 -ol acetate (Fig. 6D).

Fig. 7A and Fig. 7B depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of methyl salicylate emitted from ripe fruit of samt variant V68M (Fig. 7A) and samt variant Q333 ^* (Fig. 7B) segregants (aa) of BC0F2 populations compared to wildtype segregants (AA).

Fig. 8A and Fig. 8B depict graphs showing glucose content (g/L, Fig. 8A) and fructose content (g/L, Fig. 8B) in ripe fruit of hpl1 variant W39 ^* and hpl1 variant P239L segregants (aa) of BC0F2 populations compared to wildtype segregants (AA) used in a tetrad taste test. For glucose, ANOVA summary: F = 0.6625; P value = 0.5496; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.1809. For fructose, ANOVA summary: F = 0.5407; P value = 0.6083; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.1527.

Fig. 9A and Fig. 9B depict graphs showing glucose content (g/L, Fig. 9A) and fructose content (g/L, Fig. 9B) in ripe fruit of loxC variant W526 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA) used in a tetrad taste test. For glucose, unpaired t test summary: P value = 0.2096; P value summary = ns; Significant difference among means (P < 0.05) = no; two-tailed P value; t = 1 , df = 4. For fructose, unpaired t test summary: P value = 0.2603; P value summary = ns; Significant difference among means (P < 0.05) = no; two-tailed P value; t = 1 .31 , df = 4.

Fig. 10A and Fig. 10B depict graphs showing glucose content (g/L, Fig. 10A) and fructose content (g/L, Fig. 10B) in ripe fruit of samt variant V68M and samt variant Q333 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA) used in a tetrad taste test. For glucose, ANOVA summary: F = 0.9732; P value = 0.4518; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.2674. For fructose, ANOVA summary: F = 1 .022; P value = 0.4324; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.2771 . Fig. 1 1 A and Fig. 1 1 B depict graphs showing citric acid content (g/L, Fig. 1 1 A) and malic acid content (g/L, Fig. 1 1 B) in ripe fruit of hpl1 variant W39 ^* and hpl1 variant P239L segregants (aa) of BC0F2 populations compared to wildtype segregants (AA) used in a tetrad taste test. For citric, ANOVA summary: F = 4.252; P value = 0.0708; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.5863. For malic acid, ANOVA summary: F = 0.286; P value = 0.7562; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.0455.

Fig. 12A and Fig. 12B depict graphs showing citric acid content (g/L, Fig. 12A) and malic acid content (g/L, Fig. 12B) in ripe fruit of loxC variant W526 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA) used in a tetrad taste test. For citric acid, unpaired t test summary: P value = 0.0212; P value summary = ^* ; Significant difference among means (P < 0.05) = yes; two-tailed P value; t = 3.682, df = 4. For malic acid, unpaired t test summary: P value = 0.7029; P value summary = ns; Significant difference among means (P < 0.05) = no; two-tailed P value; t = 0.4099, df = 4.

Fig. 13A and Fig. 13B depict graphs showing citric acid content (g/L, Fig. 13A) and malic acid content (g/L, Fig. 13B) in ripe fruit of samt variant V68M and samt variant Q333 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA) used in a tetrad taste test. For citric acid, ANOVA summary: F = 1 .04; P value = 0.4255; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.2807. For malic acid, ANOVA summary: F = 0.8922; P value = 0.4857; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.2507.

Fig. 14A and Fig. 14B depict graphs showing texture measurements of ripe fruit with skin (Fig. 14A) and without skin (Fig. 14B) in ripe fruit of hpl1 variant W39 ^* and hpl1 variant P239L segregants (aa) of BC0F2 populations compared to wildtype segregants (AA). Texture measurements were made with a handheld penetrometer and the data shows the force (N) required to penetrate the ripe fruit. With skin, ANOVA summary: F = 1 .581 ; P value = 0.21 10; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.03157. Without skin, ANOVA summary: F = 1 .533; P value = 0.2210; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.03064.

Fig. 15A and Fig. 15B depict graphs showing texture measurements of ripe fruit with skin (Fig. 15A) and without skin (Fig. 15B) in ripe fruit of loxC variant W526 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA). With skin, unpaired t test summary: P value = 0.0088; P value summary = ^** ; Significant difference among means (P < 0.05) = yes; two-tailed P value; t = 2.664, df = 120. Without skin, unpaired t test summary: P value = 0.2241 ; P value summary = ns; Significant difference among means (P < 0.05) = no; two-tailed P value; t = 1 .222, df = 120.

Fig. 16A and Fig. 16B depict graphs showing texture measurements of ripe fruit with skin (Fig. 16A) and without skin (Fig. 16B) in ripe fruit of samt variant V68M and samt variant Q333 ^* segregants (aa) of BC0F2 populations compared to wildtype segregants (AA). With skin, ANOVA summary: F = 1 .61 ; P value = 0.1883; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.02393. Without skin, ANOVA summary: F = 1 .638; P value = 0.1818; P value summary = ns; Significant difference among means (P < 0.05) = no; R square = 0.0241 .

Fig. 17A to Fig. 17D depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of C5 volatile compounds emitted from ripe fruit of hpl1 variant W39 ^* segregants (aa) of a BC1 F2 (Vendor background) population compared to wildtype segregants (AA) of a BC1 F2 (Vendor background) population, a commercial hybrid (Endeavour) and Vineland elite inbred strain (TO14FN0120). The C5 volatile compounds are 1 -penten-3-ol (Fig. 17A), 3-pentanone (Fig. 17B), 1 -penten-3-one (Fig. 17C) and E-2-pentenal (Fig. 17D).

Fig. 18 depicts graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of methyl salicylate emitted from ripe fruit of samt variant Q333 ^* segregants (aa) of a BC1 F2 (Vendor background) population compared to wildtype segregants (AA) of a BC1 F2 (Vendor background) population, a commercial hybrid (Endeavour) and a Vineland elite inbred strain (TO14FN0120). Fig. 19A to Fig. 19E depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of C5 volatile compounds emitted from ripe fruit of frp/1 variant W39 ^* segregants (aa) of BC1 F2 populations compared to wildtype segregants (AA) and the wildtype genetic background TO14FN012 (WT). The C5 volatile compounds are E-2-pentenal (Fig. 19A), 3-pentanone (Fig. 19B), 1 -penten-3-ol (Fig. 19C), 1 -penten-3-one (Fig. 19D) and Z-2-penten-1 -ol (Fig. 19E).

Fig. 20A to Fig. 20C depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of C6 volatile compounds emitted from ripe fruit of frp/1 variant W39 ^* segregants (aa) of BCi F ₂ populations compared to wildtype segregants (AA) and the wildtype genetic background TO14FN012 (WT). The C6 volatile compounds are Z- 3-hexen-1 -ol (Fig. 20A), E-2-hexenal (Fig. 20B) and 1 -hexanol (Fig. 20C).

Fig. 21 A to Fig. 21 E depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of C5 volatile compounds emitted from ripe fruit of frp/1 variant P239L segregants (aa) of BCi F ₂ populations compared to wildtype segregants (AA) and the wildtype genetic background TO14FN012 (WT). The C5 volatile compounds are E-2-pentenal (Fig. 21 A), 3-pentanone (Fig. 21 B), 1 -penten-3-ol (Fig. 21 C), 1 -penten-3-one (Fig. 21 D) and Z-2-penten-1 -ol (Fig. 21 E).

Fig. 22A to Fig. 22C depict graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of C6 volatile compounds emitted from ripe fruit of frp/1 variant P239L segregants (aa) of BCi F ₂ populations compared to wildtype segregants (AA) and the wildtype genetic background T014FN012 (WT). The C6 volatile compounds are Z- 3-hexen-1 -ol (Fig. 22A), E-2-hexenal (Fig. 22B) and 1 -hexanol (Fig. 22C).

Fig. 23 depicts graphs showing original concentrations (ppm/g FW/hr) and normalized concentrations of methyl salicylate emitted from ripe fruit of samt variant Q333 ^* segregants (aa) of BCi F ₂ populations compared to wildtype segregants (AA) and the wildtype genetic background TO14FN012 (WT).

Fig. 24A to 24C depicts °Bx (Brix) percentage by mass for red ripe fruit of variant segregants (aa) of BCi F ₂ populations compared to wildtype segregants (AA) and the wildtype genetic background TO14FN012 (WT). Box and whisker plots showing mean (middle box bar), upper (75%) and lower (25%) quartiles, and maximum and minimum values (upper and lower bars) for n = 6 biological replicates (6 fruit/genotype) of about100 g each. Asterisk ( ^* ) indicates statistically significant differences (T-test, P < 0.05). Detailed Description

Single nucleotide polymorphisms (SNPs) in the nucleotide sequences of certain wildtype genes of cultivated tomato {Solanum lycopersicum) can produce tomato variants having differentiated flavor without unduly impacting production traits. In particular, SNPs in the 13-Hydroperoxide lyase 1 (HPL1 ) gene (Solyc07g049690), Lipoxygenase C (LoxC) gene (Solyc01 g006540) or Salicylic Acid Methyltransferase (SAMT) gene (Solyc09g091550) can produce tomato plants having such a desirable phenotype.

The wildtype cDNA sequence of the HPL1 gene (Solyc07g049690) of S. lycopersicum is set forth in SEQ ID NO: 1 , while the corresponding amino acid sequence of the polypeptide encoded by the wildtype cDNA sequence is set forth in SEQ ID NO: 2. In one embodiment (W39 ^* ), the codon (tgg) coding for tryptophan (W) at position 39 of the HPL1 polypeptide (SEQ ID NO: 2) is replaced with a premature stop codon (e.g. tag) in the corresponding HPL1 gene (SEQ ID NO: 1 ). In another embodiment (P239L), the codon (cct) coding for proline (P) at position 239 of the HPL1 polypeptide (SEQ ID NO: 2) is replaced with a codon (e.g. ctt) for leucine (L) in the corresponding HPL1 gene (SEQ ID NO: 1 ).

The wildtype cDNA sequence of the LoxC gene (Solyc01 g006540) of S. lycopersicum is set forth in SEQ ID NO: 3, while the corresponding amino acid sequence of the polypeptide encoded by the wildtype cDNA sequence is set forth in SEQ ID NO: 4. In one embodiment (W526 ^* ), the codon (tgg) coding for tryptophan (W) at position 526 of the LoxC polypeptide (SEQ ID NO: 4) is replaced with a premature stop codon (e.g. tag) in the corresponding LoxC gene (SEQ ID NO: 3).

The wildtype cDNA sequence of the SAMT gene (Solyc09g091550) of S. lycopersicum is set forth in SEQ ID NO: 5, while the corresponding amino acid sequence of the polypeptide encoded by the wildtype cDNA sequence is set forth in SEQ ID NO: 6. In one embodiment (V68M), the codon (gtg) coding for valine (V) at position 68 of the SAMT polypeptide (SEQ ID NO: 6) is replaced with a codon (e.g. atg) for methionine (M) in the corresponding SAMT gene (SEQ ID NO: 5). In another embodiment (Q333 ^* ), the codon (caa) coding for glutamine (Q) at position 333 of the SAMT polypeptide (SEQ ID NO: 6) is replaced with a premature stop codon (e.g. taa) in the corresponding SAMT gene (SEQ ID NO: 5). It is understood that various different codons can code for the same amino acid. It is understood that the replacement codons in the variant polynucleotides of the present invention may be any suitable alternative codon as listed in Table 1 .

Table 1

In tomato variants where alleles of the mutated genes are in the homozygous state, the variant alleles may confer enhanced levels of five carbon (C5) volatile compounds in ripe fruit. The level of volatile C5 compounds in the ripe fruit may be increased by at least two-fold, for example from 2 to 10 times, in comparison to levels of volatile C5 compounds in the ripe fruit of wildtype tomato plants. The volatile C5 compounds may include, for example, one or more of E-2-pentenal, 3-pentanone, 1 -penten-3-ol, 1 -penten-3-one and Z-

2- penten-1 -ol. Accumulation of volatile C5 compounds in the ripe fruit is accompanied by a decrease in six-carbon (C6) volatile compounds, in comparison to levels of volatile C6 compounds in the ripe fruit of wildtype tomato plants. The level of volatile C6 compounds in the ripe fruit may be decreased by at least two-fold, for example from 2 to 10 times, in comparison to levels of volatile C6 compounds in the ripe fruit of wildtype tomato plants. The volatile C6 compounds may include, for example, one or more of hexanal, E-2-hexenal,

3- hexenal, 1 -hexenol, Z-3-hexen-1 -ol, Z-3-hexen-1 -ol acetate, beta-cyclocitral and Z-4- decenal. Alternatively, or additionally, the tomato variants may have reduced levels of other volatile compounds, for example methyl salicylate, which positively influences flavor perception.

Mutagenesis has been used in plant breeding for decades to develop and introduce desirable traits. A mutagenesis-based method can be used to obtain tomato lines with differentiated and/or improved (i.e. consumer-preferred) flavor. Generally, the method involves creating, isolating and characterizing rare allelic variants of aroma volatile biosynthetic genes in tomato {Solarium lycopersicum). The genes that confer altered volatile profiles may then be introduced by known methods (e.g. marker-assisted backcrossing) into tomato breeding lines and hybrids to engineer novel flavors in a population of cultivated tomato.

The mutagenesis-based method is non-transgenic. Artificial nucleotide sequences and/or nucleotide sequences from other species of plants are not introduced into the tomato plant to alter the flavor profile of ripe tomato fruit. Rather, existing tomato genes implicated in flavor profile are mutated to alter expression of the corresponding proteins, thereby altering biosynthetic production of flavor enhancing and/or flavor de-enhancing volatile compounds in the ripe tomato fruit. Mutations in the genes of tomato plants may be induced by known methods, for example by spontaneous mutation of one or more bases in a gene (e.g. by spontaneous hydrolysis), or by exposing the plant to a mutagen (e.g. physical or chemical mutagens). Some examples of mutagens include ionizing radiation, fast neutron bombardment, reactive oxygen species (ROS), deaminating agents, polycyclic aromatic hydrocarbons (PAH), alkylating agents, alkaloids, bromine, sodium azide, heavy metals and the like. In one embodiment, ethyl methanesulfonate (EMS) may be used to induce mutagenesis. Mutagenesis may occur at one or more bases of the gene. In one embodiment, single nucleotide polymorphisms (SNPs) may be induced. Since mutagenesis-based methods typically yield loss-of-function alleles, this approach is particularly well-suited for inactivating production of aroma volatiles that detract from liking.

Detection of desired mutations in mutagenized populations of tomato plants may be accomplished by any desired method, for example by a polymerase chain reaction (PCR) method. Oligonucleotide primers derived from the gene of interest may be prepared and PCR may be used to amplify regions of the gene of interest from plants in the mutagenized population. In one embodiment, amplified mutant genes may be annealed to wild-type genes to find mismatches between the mutant genes and wild-type genes. Detected differences may be traced back to the plants which had the mutant gene thereby revealing which mutagenized plants will have the desired expression. In another embodiment, Deep Variant Scanning (DVS) bioinformatics pipeline may be used to call likely single nucleotide polymorphisms (SNPs, i.e. mutations) relative to the wild type sequence. DVS may comprise amplifying an aroma-related gene by polymerase chain reaction (PCR) from pooled DNA samples representing mutagenized plants from the mutagenized population of tomato plants, pooling amplicons produced from the PCR and sequencing the amplicons by pair-end sequencing to produce paired-end reads or obtaining a paired-end sequence reads for the amplicons, merging the paired-end reads into composite reads and mapping the composite reads to reference sequences to identify the mutation in the aroma-related gene, and identifying a member of the population comprising the identified mutation in the aroma-related gene. The identified mutation in the aroma-related gene may be identified, for example, by high-resolution DNA melting (HRM) and assigned to an individual M2 family.

Once mutagenized plants having the desired allelic variant are identified, populations of the mutagenized plants may be obtained by planting seeds from the mutagenized plants and cultivating the plants to obtain the population of mutagenized plants. Further, the desired variant genes may be introduced by known methods (e.g. marker-assisted backcrossing) into tomato breeding lines and hybrids to engineer novel flavors in a population of cultivated tomato. For example, variant lines that exhibit altered volatile compound profiles compared to wildtype segregants can be backcrossed to the wildtype genetic background via marker-assisted selection to purify an SNP of interest and to ultimately provide a breeding line harboring an aroma gene variant. Marker-assisted selection can then be used to introduce the variant allele into elite inbred lines that are employed to create hybrid tomato lines for commercial production.

EXAMPLES Materials and Methods:

Mutagenesis of tomato plant lines

Chemical mutagenesis can be performed by methods known in the art, for example the methods described by Shikata and colleagues (Shikata et al. 2015) or Garcia and colleagues (Garcia et al. 2016) to generate M2 seed stocks. Screening for allelic variants in aroma genes

Variants of aroma volatile biosynthetic genes were isolated from an EMS- mutagenized tomato population via a reverse genetics method called Deep Variant Scanning (DVS) (Banks et al., 2016, the entire contents of which is herein incorporated by reference). The DVS method identifies tomato lines harboring single nucleotide polymorphisms (SNPs) in the target gene, in this case aroma-related genes.

Target genes of interest (e.g. hpl1, loxC, samt) in tomato were amplified by PCR from pooled DNA samples representing mutagenized M2 plants from a tomato population. Amplicons representing multiple genes were pooled stoichiometrically and prepared for high-throughput lllumina sequencing with a Nextera™ XT kit. The Deep Variant Scanning (DVS) bioinformatics pipeline was used to call likely single nucleotide polymorphisms (SNPs, i.e. mutations) relative to the wild type sequence. Putative mutations were confirmed and assigned to an individual M2 family by High-Resolution DNA melting.

Breeding

Seeds of the variant tomato lines (M2) harboring the desired genetic variations were sown in 72-cell flats in soil and the resulting seedlings genotyped with respect to the single nucleotide polymorphism of interest to determine if they are homozygous or heterozygous for the variant allele (aa), or if they are wildtype segregants (AA). At least three individuals of each genotype were transplanted to soil in 1 -gallon pots or to a hydroponic system and cultivated under standard greenhouse production conditions to yield red, ripe fruit. Heterozygotes were backcrossed to the recurrent parent, Vendor VFT or to elite inbred lines, and progeny were subsequently selfed to generate homozygous variants segregating in the F2.

In other experiments, variant alleles of hpl and samt in the Vendor genetic background were introgressed via marker-assisted selection into the cultivated tomato genetic background, a Vineland elite inbred "TO14FN0120", to provide BCi F ₂ seed segregating for each mutation of interest. Homozygous variant (aa) and wildtype (AA) segregants of the BCi F ₂ populations, as well as the wildtype background TO14FN0120 were cultivated under standard greenhouse hydroponic cultivation conditions to yield ripe BC1 F2 fruit.

Volatile emissions testing of ripe tomato fruit

At least three red ripe tomatoes from each line were harvested, chopped into small pieces (approx. 1 cm ³ ). Aroma volatiles were collected from the headspace of the freshly chopped fruit by flowing carbon-filtered air over the chopped fruit for 1 hour and collecting the sample in a divinylbenzene resin (HayeSep Q) column built out of a modified disposable syringe (Tieman et al. 2012). Nonyl acetate (5 ppm) was included as a standard. The sample was eluted into a GC vial with 300 μΙ of dichloromethane and injected into the GC- MS via an autosampler. The aroma volatiles were identified and quantified using gas chromatography-mass spectrometry (Bruker 436-GC and Scion MS detector (electron ionization source, 70 eV) and calibration curves generated using authentic standards of the following volatiles: 3-methylbutanal, 2-methylbutanal, 1 -penten-3-ol, 1 -penten-3-one, 3- pentanone, 3-methylbutanenitrile, 3-methyl-1 -butanol, 2-methyl-1 -butanol, E-2-pentenal, 1 - pentanol, Z-2-Penten-1 -ol, isobutyl acetate, butyl acetate, 3-Methyl-1 -pentanol, E-3-hexen-

1 - ol, E-2-hexenal, Z-3-hexen-1 -ol, 1 -hexanol, isopentyl acetate, 2-methylbutyl acetate, heptanal, E-2-heptenal, benzaldehyde, 1 -octen-3-one, 6-methyl-5-hepten-2-one, 6-methyl- 5-hepten-2-ol, 2-isobutylthiazole, phenylacetaldehyde, salicylaldehyde, guaiacol, nonanal,

2- phenylethanol, benzyl cyanide, Z-4-decenal, methyl salicylate, beta-cyclocitral, benzothiazole, Z-citral, E-citral, 2,4-decadienal, Eugenol, Z-6,10-dimethyl-5,9-undecadien- 2-one, E-6,10-dimethyl-5,9-undecadien-2-one, beta-ionone. Volatile profiles were analyzed using multivariate statistics module provided by Metaboanalyst 3.0 (Xia and Wishart, 2016). Sugars and acids testing of ripe tomato fruit

At least three red ripe tomatoes were individually homogenized using a food processor (Magic Bullet model 1001 ), then homogenates were transferred to 50 ml tubes and centrifuged at 4696 x g for 15 minutes (Sorvall ST16, Thermo Scientific). The supernatants were collected in separate tubes for further analysis and the pellet was discarded. Malic acid content of undiluted supernatant was determined using a Malic acid quantification assay kit (K-LMALQR, Megazyme). Supernatants were diluted 20-fold in ddH ₂ 0 for citric acid determination using a citric acid assay kit (K-CITR, Megazyme). Glucose and fructose concentrations were determined using supernatant diluted 50-fold with ddH ₂ 0 and a glucose/fructose assay kit (K-FRGLQR, Megazyme). All sugar and acid assays were performed in 96-well plate format (Grenier BioOne 96-well plates) according to the kit manufacturer's instructions. Spectrophotometric readings were performed using a BioTEK EON plate reader and Gen5 software (version 2.05, BioTEK).

Texture testing of ripe tomato fruit

Texture was assessed with a handheld penetrometer (PCE FM200, PCE Instruments). With the skin still on the tomato, three readings were taken at the maximum force required to push through the skin. A sharp knife was then used to slice a thin section of skin 1 -2 cm, and penetrometer readings on the section with no skin were taken in triplicate. Triplicate readings were averaged to determine the mean values for "with skin" and "without skin", Flavor testing of ripe tomato fruit

Twenty-eight panelists (19 women, 9 men) were recruited and screened for eligibility to participate. Panelists completed 4 unspecified tetrad tests (Delwiche and O'Mahoney, 1996; Ennis and Jesionka, 201 1 ), each on a different set of tomatoes. In each tetrad test, subjects were presented with four samples - two samples from one group (wildtype, AA) and two from another (variant, aa). Panelists were asked to taste all samples, from left to right, and to group the samples into two groups based on similarity. Samples were identified with a 3-digit code and each set of four samples was presented in a randomized, balanced block design. Additionally, each panelist completed the 4 tetrad tests in a balanced block design, to minimize the first position bias and carry-over effects. A one- minute forced break was implemented between each tetrad set. Panelists were asked to record the reason for their choice, and to comment on product characteristics they felt differentiated the products. Panelists were provided with unsalted crackers along with both filtered water and low sodium sparkling water and advised to rinse between each sample. The session was conducted under green light to minimize any visual bias and the entire session lasted approximately 30 minutes.

Data were analyzed for the difference testing by entering the number of participants that made the "correct" grouping (N _c ); grouped two products together from the same product type. The parameters of the test were: alpha(a): 0.05 and delta: 1 .47. Using Thurstonian modeling, the parameters, including N _c were input into V-Power™ (Sensory Software tool based on Discrimination Test Planning and Analysis by Carr, T., Jesionka, V.). The program's output stated p-value and statistical significance.

Brix (°Bx) level testing of ripe fruit

Chopped fruit from aroma volatile analysis were collected in a beaker and 300 μΙ of tomato fruit juice was analyzed with a digital refractometer (Atago™ PAL-1 ) to measure the degrees Brix for at least 6 tomatoes per genotype.

Preference profiling

Between 10-1 1 members of a trained sensory panel (8 females, 3 males) participated over the course of 2 evaluation sessions seven days apart. Assessors rated the intensity of tomatoes using a lexicon of taste, mouthfeel, flavor/aroma, texture and overall flavor intensity attributes. The lexicon of tomato attributes for sensory profiling is summarized as follows: Aroma/Flavor - overall flavor intensity, grassy/vegetal, hay, tomato stem, metallic, earthy, smoky, lemony.

Taste/Mouthfeel - sweet, acid, salt, bitter, umami, astringent.

Texture - crunchy, viscosity, juicy, seedy, meatiness, firmness of flesh, skin toughness.

Tomato samples were presented one by one. For each sample, assessors received two wedges, one at a time and were instructed to bite into the first wedge to assess aroma/flavor and taste/mouthfeel attributes, pass back the sample cup to receive the second wedge which would be used to rate the textural attributes. Panelists rated their perceptions on 15-cm line scales calibrated with a reference standard intensity, and anchored from "low" to "high". A replicate of each sample was profiled during the same session. Samples were presented using a randomized, balanced block design to minimize the first position bias and carry-over effects. Evaluations were conducted under green light to minimize visual cues. Data were collected using the sensory software EyeQuestion™ (Logic 8, the

Netherlands), and were analyzed with XLStat™ (Addinsoft, France).

A 3-way mixed model ANOVA with 2-way interaction was conducted on the sensory profiling data of the 22 attributes. The LSD (Least Significant Difference) pairwise comparison was used as the post-hoc test. An alpha of 0.1 was used to determine statistical significance. Data were analyzed respectively as day 1 , day 2, and "global" (day 1 and day 2 combined) to establish differences for "overall flavor intensity".

A 2-dimension External Preference Map (PREFMAP) was conducted using the quadratic model and best fit option. Preference data were considered significant at p=<0.05. The PREFMAP is based on the 18 tomatoes consumers tasted in 2013; all other tomatoes are projected onto the sensory space as predictions.

Results and Discussion:

Homozygous variant tomato fruit ("aa" segregants) tasted by 28 panelists (19 staff/community, 9 trained panelists) in the tetrad test described above were described as having "more flavor", and "sharper flavor" compared to wildtype tomato fruit ("AA" segregants). Evaluation of the fruit using the tetrad test determined a significant difference in flavor between variant (aa) and wildtype (AA) fruit, as shown in Table 2. Table 2

Improved flavor perception of the homozygous variant fruit may be attributed, at least in part, to increased levels of C5 volatile compounds and decreased levels of C6 volatile compounds and methyl salicylate.

As shown in Fig. 1 A to Fig. 1 E, Fig. 3A to Fig. 3E, Fig. 19A to Fig. 19E and Fig. 21 A to Fig. 21 E, the levels of five C5 volatile compounds associated with flavor enhancement are increased in the ripe fruit of both of the HPL1 variants (W39 ^* and P239L) in comparison to the wildtype, generally by at least about 1 .5 times and in some cases by about 4 times. As shown in Fig. 2A to Fig. 2C, Fig. 4A to Fig. 4C, Fig. 5A to Fig. 5C, Fig. 6A to Fig.

6D, Fig. 20A to Fig. 20C and Fig. 22A to Fig. 22C, the levels of various C6 volatile compounds associated with flavor de-enhancement are decreased in the ripe fruit of both of the HPL1 variants (W39 ^* and P239L) and the LoxC W526 ^* variant in comparison to the wildtypes, generally by at least about 2 times. Of particular note, the W526 ^* LoxC variant allele in the homozygous state confers up to a about a 10-fold reduction in 'green leaf six- carbon volatiles (e.g. hexanal, hexenal, etc.) in ripe fruit. A substantial reduction in C6 volatiles leads to differentiated flavor, where the flavor is reported as "nice and sweet" and having "more tomato flavor" as compared to wildtype fruit.

As shown in Fig. 7A to Fig. 7B and Fig. 23, the levels of methyl salicylate in the ripe fruit of the SAMT V68M variant allele in the homozygous state are significantly reduced compared to the wildtype, while the SAMT Q333 ^* variant allele in the homozygous state significantly reduces or even substantially abolishes methyl salicylate content in ripe fruit. Methyl salicylate has a wintergreen aroma, sometimes perceived as a "pharmaceutical", "plastic" or "pesticide" aroma. Reduction of methyl salicylate can positively influence flavor perception. The homozygous Q333 ^* fruit has "much better flavor" according to panelists in the tetrad tests, which detected a significant difference between the homozygous variant and the wildtype fruit (Table 2). As shown in Fig. 8A, Fig. 8B, Fig. 9A, Fig. 9B, Fig. 10A and Fig. 10B, sugar content (glucose and fructose) of all of the variants compared to wildtype segregants were not significantly different, suggesting that aroma profile, and not sugar content, was driving product differentiation in the tetrad test. As shown in Fig. 1 1 A, Fig. 1 1 B, Fig. 12A, Fig. 12B, Fig. 13A and Fig. 13B, acid content (citric acid and malic acid) of all of the variants compared to wildtype segregants were not significantly different (except for citric acid in the LoxC variant), suggesting that aroma profile, and not acid content, was driving product differentiation in the tetrad test.

As shown in Fig. 14A, Fig. 14B, Fig. 15A, Fig. 15B, Fig. 16A and Fig. 16B, texture/firmness of all of the variants compared to wildtype segregants was also not significantly different (except for the LoxC variant with skin), suggesting that aroma profile, and not texture/firmness, was driving product differentiation in the tetrad test.

Many commercial hybrid tomatoes accumulate very low levels of C5 volatiles such as 3-pentanone and E-2-pentenal. As shown in Fig. 17A, Fig. 17B, Fig. 17C and Fig. 17D, the ripe fruit of the HPL1 variant W39 ^* has much higher content of various C5 volatile compounds than a commercial hybrid (Endeavour) and an elite inbred strain (TO14FN0120). In the homozygous state, variant alleles of HPL1 such as W39 ^* confer as much as 10-fold increases in the C5 volatile content of ripe fruit leading to differentiated flavor. Many commercial tomato cultivars accumulate methyl salicylate. The SAMT Variant allele Q333 ^* substantially inactivates methyl salicylate production when in the homozygous state as shown in Fig. 18, which can contribute to the differentiated aroma found in the tetrad test. As seen in Fig. 18, levels of methyl salicylate in wildtype segregants as well as a commercial hybrid (Endeavour) and an elite inbred strain (TO14FN0120). As shown in Fig. 24C, the samt Q333 ^* variant has a Brix level that is not significantly different than that of the wildtype (AA) segregants and the wildtype (WT) genetic background. Although the hpl variants (Fig. 24A and Fig. 24B) exhibited Brix levels significantly lower than the wildtype (AA) segregants and the wildtype (WT) genetic background as shown in Fig. 24A and Fig. 24B, these variants are still predicted to be preferred by 80-100% of consumers based on preference mapping from sensory profiling.

Sensory profiling was conducted to evaluate the following cultivars: TO14FN0120 (elite inbred line), HPL W39 aa, LoxC W526 aa, HPL P239L aa, and SAMT Q333 ^* aa. Analysis of Variance (ANOVA) identified statically significant differences between the products for 8 sensory attributes (a=0.10): salty and umami taste; smoky flavor/aroma; as well as, crunchy, viscosity, meatiness, firmness of flesh and skin toughness. Intensities of the 8 sensory attributes for each product were determined and multiple comparisons indicating statistically significant differences between the products for each sensory attribute are provided in Tables 3-10.

Table 3: Multiple comparison of smoky

Table 4: Multiple comparison of salty

Table 5: Multiple comparison of umami

Table 6: Multiple comparison of crunchy

Table 7: Multiple comparison of viscosity

Table 8: Multiple comparison of meatiness

Table 9: Multiple comparison of firmness of flesh Table 10: Multiple comparison of skin toughness

The 5 cultivars assessed were projected onto a tomato preference map (created in 2014) to predict consumer preference. LoxC W526 aa, located in the 60-80% consumer satisfaction zone, was predicted to have substantially lower consumer preference than the other 4 cultivars which fell in the ideal red zone with 80-100% satisfaction. HPL P239L aa was predicted as the most preferred variety for all three consumer groups, followed in descending order of preference by TO14FN0120, HPL W39 aa and SAMT Q333 ^* aa, however these four cultivars all exhibited very similar predicted consumer preference.

A revised preference map (prefmap) was created in 2018, which excluded the attribute salty. A review of all tomato sensory profiling conducted from 2014 to 2018 concluded that while there may be statistically significant differences in perceivable salty flavor in tomatoes, these differences are minor when predicting consumer preference and are not primary drivers of consumer liking. The predicted placement of the 5 cultivars onto this revised prefmap are similar to the original. LoxC W526 aa had the lowest predicted preference at only 40-60% consumer satisfaction. T014FN0120, HPL W39 aa, HPL P239L aa and SAMT Q333 ^* aa all remained in the ideal consumer preference zone of 80-100% predicted liking. HPL P239L aa consistently has the highest predicted preference when compared to TO14FN0120, HPL W39 aa, HPL P239L aa and SAMT Q333 ^* aa on the prefmap.

There are statistically significant differences between TO14FN0120, HPL W39 aa, LoxC W526 aa, HPL P239L aa, and SAMT Q333 ^* aa for salty and umami taste, smoky flavor/aroma, as well as, crunchy, viscosity, meatiness, firmness of flesh and skin toughness. HPL P239L aa was predicted to be the most preferred, and LoxC W526 aa the least preferred tomato when projected on the preference map. Only minor differences were found in predicted consumer preference between HPL P239L aa, TO14FN0120, HPL W39 aa and SAMT Q333 ^* aa.

References: The contents of the entirety of each of which are incorporated by this reference. Banks, Travis W., et al. (2016) High Throughput Method of Screening a Population for Members Comprising Mutation(s) in a Target Sequence. United States Published Patent Application US 2016/0047003, published February 18, 2016.

Delwiche, J., O'Mahony, M. (1996) Flavour discrimination: An extension of thurstonian 'Paradoxes' to the tetrad method. Food Quality and Preference. 7(1 ): 1 -5. Ennis, J. M, Jesionka, V. (201 1 ). The power of sensory discrimination methods revisited. Journal of Sensory Studies. 26(5) : 371 -382.

Garcia, V., Bres, C, Just, D., Fernandez, L., Tai, F. W. J., Mauxion, J. P., Alseekh, S. (2016). Rapid identification of causal mutations in tomato EMS populations via mapping- by-sequencing. Nature Protocols. 1 1 : 2401 -2418. Lewinsohn, Efraim, et al. (2001 ). Enhanced levels of the aroma and flavor compound S- linalool by metabolic engineering of the terpenoid pathway in tomato fruits. Plant Physiology. 127: 1256-1265.

Shikata, M., Hoshikawa, K., Ariizumi, T., Fukuda, N., Yamazaki, Y., Ezura, H. (2015). TOMATOMA update: phenotypic and metabolite information in the micro-tom mutant resource. Plant and Cell Physiology. 57: e1 1 -e1 1 .

Tandon, Kawaljit S., et al. (2001 ). Characterization of fresh tomato aroma volatiles using GC-olfactometry. Proceedings of the Florida State Horticultural Society. 1 14: 142-144.

Tieman, Denise, et al. (2010). Functional analysis of a tomato salicylic acid methyl transferase and its role in synthesis of the flavor volatile methyl salicylate. Plant Journal. 62: 1 13-123.

Tieman, Denise, et al. (2012). The chemical interactions underlying tomato flavor preferences. Current Biology. 22: 1035-1039.

Tieman, Denise, et al. (2017). A chemical genetic roadmap to improved tomato flavor. Science. 355: 391 -394. Vogel, Jonathan T., et al. (2010). Carotenoid content impacts flavor acceptability in tomato {Solarium lycopersicum). Journal of the Science of Food and Agriculture. 90: 2233-2240.

Xia, J. and Wishart, D.S. (2016) Using MetaboAnalyst 3.0 for Comprehensive Metabolomics Data Analysis. Current Protocols in Bioinformatics. 55:14.10.1 -14.10.9 Yauk, Yar-Khing, et al. (2014). Manipulation of flavour and aroma compound sequestration and release using a glycosyltransferase with specificity for terpene alcohols. Plant Journal. 80: 317-330.

The novel features will become apparent to those of skill in the art upon examination of the description. It should be understood, however, that the scope of the claims should not be limited by the embodiments, but should be given the broadest interpretation consistent with the wording of the claims and the specification as a whole.