Deconstructing Flavor in Tomatoes

I’ll start with this caveat: growing conditions can have a huge influence on taste/flavor in tomatoes.  Fruit harvested early or late in the season is usually inferior in flavor compared with those harvested mid-season; high flavor is often favored with plants in light drought stress; and growing area, soil type/fertility and disease incidence/plant health can significantly effect fruit taste.  Perceived “likeability” is also often confounded with non-flavor fruit quality attributes, such as color, texture and skin thickness.  For example, most folks struggle to see anything that is not red and round as a proper tomato.  We have used blind taste tests (literally – I’m talking blindfolds) to get a non-biased flavor assessment on something like GWR bicolors.  Taking off the blindfolds is fun, and informative.  However, there is a genetic component to flavor per se that is important, and subject to modification through plant breeding.

Knife at the ready

In our tomato breeding program, selection for flavor has always been primary.  We taste several thousand tomato plants in our breeding nurseries each year, and it is generally not possible for us to do anything other than apply a simple subjective selection criterion – select what you think tastes good.  We can manage that in our moderately sized breeding program – large companies cannot.  In an effort to perhaps apply more objective criteria and to design more efficient breeding strategies, we’ve been reading more on what is known about the biochemical and genetic basis for tomato taste and flavor

We have conducted tomato tastings at Frogsleap Farm for the last 5-6 years, and are always surprised with the person-to-person variation in ranking tomatoes for flavor.  As with most things that we eat and drink – taste is personal.  That said, there are trends in what people like in tomato taste/flavor – as illustrated in recent scientific studies by Harry Klee and colleagues at the University of Florida.

For decades it has been known that tomato flavor is primarily influenced by three categories of compounds: sugars, acids and volatile aromatics - and that it is a balance between these compounds that is important, not the just level of an individual component per se.  In the next several paragraphs I’ll try and outline the current understanding of the individual flavor components, how they interact, and the potential implications for tomato breeding.

Yilmaz, 2000 (reference) shows the following typical chemical composition of a tomato fruit:

Constituent	% DM
Fructose	25
Glucose	22
Saccharose	1
Citric acid	9
Malic acid	4
Protein	8
Dicarboxylic amino acid	2
Pectic substances	7
Cellulose	6
Hemicellulose	4
Minerals	8
Lipids	2
Ascorbic acid	0.5
Pigments	0.4
Other amino acids, vitamins, and polyphenols	1
Volatiles	0.1

To further set up this discussion; flavor = taste + aroma.  The tongue recognizes five basic tastes; sweet, sour, bitter, salty and savory - all of which come from non-aromatic/volatile constituents.  There are various aromas in tomatoes, favorable and unfavorable, and these are due to volatile compounds, to which we have receptors in the nose.

Sugars

Glucose and fructose are the major sugars in tomato, with glucose predominating early and fructose increasing with advancing fruit maturity.  In a mature ripe tomato the glucose:fructose ratio is generally about 1:1.  The tomato wild relative S. chmielewskii accumulates sucrose, rather than glucose – the result of a recessive loss-of-function allele of the gene controlling production of the enzyme acid invertase.  This allele and its high sucrose phenotype have been introgressed into commercial tomatoes.  Total sweetness index (TSI) is used to indicate sweetness. The contribution each sugar makes to this parameter is described relative to sucrose, and calculated as: [(1.00 × sucrose) + (0.76 × glucose) + (1.50 × fructose)].  When tomato samples are spiked with any of the three sugars, tasters perceive less overall aroma and less ripe tomato flavor.  Flavor intensity was highest with samples spiked with sucrose or fructose (Baldwin andThompson, 2000).

Total soluble solids, measured as Brix, is the sum of sugars (65%), organic acids (13%) and other minor components, and is intrinsically correlated to sugar content in tomatoes.  Although Brix is also highly correlated to sweetness in various tomato tasting experiments, Tieman, et al. found that specific volatile aromatic compounds enhanced perceived sweetness in their tasting experiments (more on this later).

Sweet (beef x cherry) segregate

Smaller fruited types generally have higher sugar content/Brix.  Tomato breeders at the University of Florida crossed a high sugar cherry tomato with a low sugar large fruited type and evaluated segregating progeny (Georgelis, et al. 2004).  They found a negative correlation between sugar content and fruit size or earliness; higher sugars in indeterminate vs determinate types; and no relationship between sugar content and fruit yield.  Others have reported a negative correlation between sugar content and fruit yield, which better matches our personal experience.  We’ve made a lot of crosses between cherry and beefsteak types.  Can you recover cherry level sweetness in a beefsteak segregate? – no.  We have found some pretty tasty larger fruited lines from such crosses, but they are not sugar bombs.

Photosynthesis is the conversion of CO₂ to sugar.  It is the ultimate source of energy in all the food we eat and in the fossil fuels we burn.  Chlorophyll is a green pigment, stored in plastids (chloroplasts), that is critical to photosynthesis.  In most plants chlorophyll is concentrated in leaves.  Tomato plants have chlorophyll/chloroplasts in leaves, stems and in unripe fruit.  In fact we now know that the chloroplasts in unripe fruit are an important source of sugars for the fruit, and that more chloroplasts generally result in improved fruit quality (Nguven, et al 2014).  Dark green unripe fruit is good for fruit quality.  GLK2 is a transcription factor that controls the formation of chloroplasts in tomato fruit.  The GLK2 wild type gene “U” has a phenotype with darker green unripe fruit and lingering green shoulders in ripe fruit.  The loss-of-function mutant “u” has lighter green unripe fruit and no green shoulders – so called uniform ripening.  The uniform ripening trait has been bred into most commercial tomatoes to facilitate commercial harvest and to present a more uniform red fruit (in grocery stores).   There are a lot of reasons why supermarket tomatoes generally taste bad – and this is one of them (Powell, 2012).   Since all heirloom types contain the wild type allele at this locus, any breeding program for improved taste should insure preservation of the wild type.  This is particularly true for progeny from crosses between heirloom and commercial types.

Green shoulders on non-uniform ripening wild type

Numerous studies have shown that generally, perceived sweetness is the most powerful determinate of “likeability” for taste in tomatoes.  That is certainly supported by our less rigorously designed tasting exercises here, although it is not uncommon for someone to say, “that one is too sweet”.  Sweet and sour - the ying and the yang.

Organic acids

Organic acids are what provide tomatoes the sour/tangy counterbalance to sweet.  The two primary organic acids in tomato are citric acid and malic acid.  Malic acid content is high in developing fruit, but decreases as the fruit matures.  Citric acid level increases somewhat during fruit development and generally represents 65-70% of the organic acids in ripe tomato fruit.  Tieman et al. report that tomato fruit citric acid concentration, when corrected for fructose content, was correlated to flavor intensity in their tasting trials.  Tomato fruit also contain low levels of free amino acids.   Glutamic acid, g-aminobutyric acid, glutamine, and aspartic acid comprise about 80% of the total free amino acids in
tomatoes.  Glutamic acid (glutamate) is concentrated in the gel around the seed and although it is a component of the umami (i.e savory) taste in foods, it seems to be negatively correlated with flavor in fresh tomatoes (Buchell et al., 1999).  

There is significant genetic variation for organic acid content and pH in tomatoes, with pH ranging from 4.1 to 4.8.  Low acid tomatoes are generally bland or one-dimensional, and a combination of high acid and high sugar is generally favorable.  In a classic study at UCD (Stevens et al., 1979, see graph below) tasters found that overall flavor intensity was correlated with both acidity (sourness) and sugars (sweetness), but that acidity (ph or titratable acidity) was most closely related to flavor intensity.  Tomato-like flavor, something we all love, was not strongly correlated with either sugar or acids – and is driven instead by volatile aromatics. 

Stevens, et al., 1979

Volatile aromatic compounds
Over 400 volatile aromatic compounds (AKA volatile organic compounds = VOCs), are found in tomato, and are either derived from carotenoid pigments, fatty acids, phenylalanine or free amino acids (Klee, 2010).  VOCs generally accumulate during fruit ripening.  Volatile compounds can be perceived by humans in two ways – sniffed through the nostrils (smell) or forced up behind the palate after chewing/swallowing (flavor).  It is generally agreed that taste is a complex interaction between smell, flavor, sugar and acids. 

Tieman, et al., 2012 (reference) summarizes the results from tasting panels and the relationship between chemical composition and “likeabiliy” in a broad collection of 152 heirloom varieties compared with samples of store-bought tomatoes.  Levels of reducing sugars (fructose and glucose), organic acids (citrate and malate) and 28 VOCs were measured, and multivariate analysis was used establish relationships between concentration of these compounds and various flavor attributes determined by the tasting panel. Flavor intensity was positively correlated with twelve different VOCs, seven of which were significant after accounting for fructose content.  Sweetness was also positively correlated with twelve VOCs, eight of which overlapped with those enhancing flavor, and three of which were independent predictors of sweetness after accounting for fructose content.  Although other studies have reported volatile induced enhancement of sweet, fruity, sour, bitter, smoky, and salty tastes, this study found the most significant enhancement was to sweetness. 

Tomato plants engineered to lack fatty acid-derived VOCs (the predominant family of VOCs in tomato) scored similarly for “likeability” to those with normal levels of these compounds.  In contrast, plants without carotenoid-derived VOC’s were perceived as significantly less sweet, and with a lower “likeability” score.  Other research finds that the concentration and composition of the various carotenoid pigments in tomatoes is directly related to the concentration of specific carotenoid-derived VOCs.  Carotenoid cleavage dioxgenase (CDD) genes code for various CDD enzymes, which in the chromoplast enable the conversion of pro-lycopene (tangerine), lycopene (red), and β-carotene (orange) to specific VOCs, most of which impart a fruity flavor.  The VOC geranial, derived from lycopene, is one of which Tieman et al. reports enhances sweetness.  In yellow and GWR fruited types the cartenoid pathway is truncated  early (see Genetic control of fruit color intomatoes) and virtually no carotenoid-derived VOCs are produced.   Although I have never tasted a true yellow-fleshed tomato that I found exceptional, there are a handful of GWR fleshed heirlooms and several of our GWR breeding lines that have excellent flavor – which in light of these data, is perplexing.

Green flesh - outstanding flavor

Klee and Giovannoni, 2011 (reference) make an interesting point.  Virtually all of the VOCs that contribute to tomato flavor are derived from a chemical compound essential for the human diet – essential fatty acids, essential amino acids and carotenoid pigments.  For example β-carotene is a precursor for Vitamin A, and it’s VOC derivative β-ionone is a key flavor volatile often associated with an intense fruity flavor.  They suggest an effective co-evolution between plants and animals – plants produce essential nutrients that have a characteristic flavor profile and animals evolve the ability to seek out such products, eat them and thus distribute the seed. 

Tieman et al. used standard genomic tools to segregate the tomato lines in their study into genetically related subgroups.  Their conclusion: “Based on these data, we found no obvious genetic subgroups that could explain liking, sweetness, or tomato flavor intensity. There was no obvious genetic clustering of good versus bad taste when varieties were sorted by chemical composition. These latter data also indicate the chemical complexity of liking, as there is no simple pattern of chemical content that separates high from low consumer liking scores.”  They also found extraordinary genetic variation for concentration of critical VOCs among the tomato lines studied, much more so than for sugar or acid concentrations.  This suggests a rich opportunity for breeding once specific VOC linked molecular markers are identified and deployed to enable efficient selection in large populations.

Conclusions

Wet lab characterization of the various flavor components is expensive and labor intensive.  Deployment of molecular markers associated with various genes that together contribute to perceived tastiness will allow more efficient selection and should help deal with the complexity of this trait.  In the mean time I think we (Frogsleap Farm) are stuck with the olfactory tools Mother Nature has provided, which although subject to individual preferences, can certainly send us in the right direction.  We will continue to combine tasting with objective measurements of soluble solids (Brix).  This literature review provided some new guidance – dark green unripe fruit are favorable (avoid uniform ripening types), high cartenoid pigment content is good, and some pigments probably better than others (e.g. β-carotene).  To the extent that the clear skin mutant allele y generally depresses the carotene pathway (see Genetic control of fruit color in tomatoes), I’m steering toward the yellow skinned wild type.  The green flesh phenotype (gf/gf) provides for an incomplete breakdown and conversion of chloroplasts to chromoplasts, which would seem to lead to a decrease in the cartenoid pigment content vs the wild type Gf (e.g red vs purple).  However, we routinely find our best tasting tomatoes carry the gf phenotype – on this point I am perplexed. 

Winner in 2014 "blind taste test"

We’ll be cautious with determinate and dwarf types which just may not have the leaf area to support maximum sugar accumulation, though I know there is a practical place for both types for some growers.  Nailing flavor in cherry and grape types is relatively simple, it gets harder with larger fruited types – which is where we are concentrating now.  Another challenge is to get great flavor with acceptable fruit yield and high fruit quality (non-splitting, firm flesh and good shelf life).  We are making progress on all fronts.  We will also anxiously follow progress in Harry Klee’s lab at the University of Florida, where they continue to provide a better understanding of flavor in tomatoes, and also create positive energy for challenging the bland tasting stereotype we are routinely presented in grocery stores and restaurants.