As part of my recent fascination with plant biology, I read a paper called “A chemical genetic roadmap to improved tomato flavor“, by Tieman et al. This led me down a rabbit hole of a surprisingly large literature on the topic of tomato flavor. Of course it’s obvious why crop breeders would want to make better tomatoes (some background on this from the extensive media coverage). What I find really interesting, though, are the broader lessons about genetics and bioengineering that come out of tomatoes as a model system.1
I’m particularly struck by the theme of tradeoffs. In some sense, this is the key challenge for all of modern crop genetics. Why do intensively bred crop varieties, which have undergone selection for yield and market appeal, often taste worse than their ancestor plants? Is it possible to make a convenient and attractive tomato that is also delicious? Or, is the tradeoff due to fundamental biological constraints and therefore can’t be overcome by breeding or genetic engineering?
Historically, it was common for crop breeding to actually overcome (apparent) tradeoffs. When two varieties of a crop are crossed (either by artificial breeding or by natural cross-pollination), some of the resulting offspring plants might have favorable traits from both parents. This is possible because many traits are encoded by different sets of genes that can be “stacked” together into the same plant. This has proceeded unconsciously for thousands of years, and deliberately for more than a hundred, to produce the amazing and useful variety of grains, fruits, and vegetables we have today.2
However, once in a while, selecting for one trait makes another one worse. A case in point is the Uniform ripening gene in tomatoes.3 Heirloom tomato plants usually have the U (“uppercase u”) version of this gene, which gives them a patchy green and red appearance, even when fully ripe. Besides being unattractive to some, this also makes it harder to judge ripeness. Modern tomato varieties have the u (lowercase) version of this gene and therefore have an even red color. Unfortunately, there is a good reason that U plants have green patches–they contain more chlorophyll in the tomato fruits, which leads to more photosynthesis, more sugar, and ultimately better taste. On the other hand, the u plants look better but taste bland. So there is a tradeoff between appearance and taste because the U/u gene both promotes uneven appearance (undesirable) and increased sugar content (desirable), but you can’t have one without the other. Since breeders were only focused on selecting for more uniform-looking tomatoes, they inadvertently also selected for less tasty ones as a result.
The researchers were able to figure out the U/u tradeoff because they already suspected a connection between the gene and the traits of interest. But what if they didn’t know what gene might be involved? How do you find the gene(s) underlying a trait of interest?
This is where the latest paper from Tieman et al. comes in. The researchers mapped the genes responsible for many, if not all, of the different ways that a tomato can taste good. They did this by sequencing the genomes of and measuring the flavor compounds contained in each of 398 tomato varieties, modern, heirloom, and all. Because every variety had a slightly different mixture of compounds, as well as diverse sequences in their genomes, the researchers could do a Genome-Wide Association Study (GWAS) to see if any of the changes in chemical profile were correlated with particular genetic variants. They found that a bunch of flavor compounds have been lost in modern bred tomatoes through mutations that turned off the gene producing them. This can now serve as a sort of blueprint for breeders and genetic engineers to restore flavors to tomatoes simply by putting the “working” version of the associated gene back into the plant.
One question you might have is how Tieman et al. know if a certain chemical is a “flavor” compound. Surely lots of chemicals can be detected in tomatoes (and whose production can even be mapped to genetic variation) but most probably have no relevance at all to flavor?
To figure out which chemicals actually make tomatoes taste good to people, Tieman et al. made an ingenious innovation–they simply asked them. They had a panel of 170 people taste-test hundreds of tomato varieties and correlated these results to the chemical profiles of each variety. The chemicals flagged by this analysis as being important to flavor were the ones used in the genetic mapping. Unsurprisingly, they found that sugar content was the most important contributor to “consumer liking” (a.k.a. tastiness), and was reduced in many modern tomato varieties. They also identified a number of compounds previously thought to be unimportant to flavor because of their low levels. Often, as a proxy for “importance to flavor”, researchers focus on chemicals present at high concentrations relative to their threshold for human perception. Tieman et al.’s result shows that this might miss some important compounds because it doesn’t take human flavor preferences, as well as potentially non-additive flavor combinations, into account.
The really cool thing about the taste test was that consumers weren’t just asked whether they liked a certain tomato, but also how it rated on multiple flavor “axes”. This allowed the authors, for example, to find a number of volatile compounds that increased the perception of sweetness without increasing sugar content. This happens via “retronasal olfaction”, or the smelling of molecules that waft into the back of your nose from foods in mid-swallow. This is different from “orthonasal” olfaction, which is the more typical smelling of odors that go into your nostrils. It turns out that retronasal olfaction can actually change your perception of flavors that you normally think of as being taste rather than smell — sweetness, sourness, etc.4
The discovery of sweetness-enhancing volatile molecules (and their associated genes) is a big deal, because it might help to address the most important tradeoff with tomatoes–the more sugar a tomato has, the smaller (and lower-yield) it is. There is no easy way around this because the tradeoff comes from fundamental cellular economics–carbon atoms that go toward sugar molecules cannot go toward cellulose and other molecules that make up the bulk of the fruit. The other kind of economics also plays into this. Tomato growers are paid by the pound (and by the attractiveness of the tomatoes), and rarely sell straight to consumers, the only people who care about taste. So one potential way to engineer a high-yield, yet still tasty, tomato is to compromise slightly on sugar, but make up the loss in perceived sweetness with the right cocktail of flavor compounds.
This brings me back to a general theme when trying to engineer around an apparent tradeoff in biology. The difficulty of coming up with a solution depends on how “deep” the root cause is in terms of biological mechanism. Tradeoffs that are simply due to the wrong combinations of genes existing in nature are easy to solve simply through crossing and reassortment (for this reason, most wouldn’t even call these examples tradeoffs). Relatedly, a tradeoff due to a certain gene regulatory program working a certain way can potentially be circumvented by decoupling some of the key genes from the natural regulation. This is a little harder, in that some work must be done to elucidate the regulatory mechanism and identify the engineering targets. “Re-wiring gene regulation” is such a common (proposed) solution to problems in synthetic biology that you could probably use it as a loose definition for the field.
If the approaches above can’t circumvent an apparent tradeoff, it might mean that we don’t understand the system well enough — maybe there are some knobs that can be tuned that we haven’t discovered. On the other hand, it might mean that the tradeoff arises from fundamental constraints of biology, or worse, chemistry or physics. These may be impossible or prohibitively difficult to overcome. The tomato sugar-yield tradeoff is probably an example of this. No matter how much carbon a plant has available, it will always have to choose between allocating it to sugar or to bulk.
Of course, if we could make plants grow faster and produce biomass more quickly, it makes the tradeoff between different types of biomass less constraining in practice. Amazingly, this is not a crazy idea. In my next post, I will talk about how researchers are thinking about achieving this, which could set the stage for a new agricultural revolution in the next few decades. The heart of the story lies in another tradeoff, of a very different kind, which involves the biochemistry of the world’s most abundant enzyme.
For a scholarly review on tomato flavor genetics, see: Klee, H.J., and Tieman, D.M. (2013). Genetic challenges of flavor improvement in tomato. Trends Genet. 29, 257–262.
I’m currently reading the excellent Hybrid by Noel Kingsbury, a history of plant breeding which covers these topics in depth.
The Uniform ripening story is from the paper: Powell, A.L.T., Nguyen, C. V., et al. (2012). Uniform ripening Encodes a Golden 2-like Transcription Factor Regulating Tomato Fruit and Chloroplast Development. Science. 336, 1711–1715.
The concept of retronasal olfaction is discussed further (and tomato-tasting panels are introduced) in: Tieman, D., Bliss, P., et al. (2012). The chemical interactions underlying tomato flavor preferences. Curr. Biol. 22, 1035–1039.