Does genetic manipulation cause unintended changes in food quality and composition? Are genetically modified foods less nutritious than their non-modified counterparts, for example, or different in unknown ways?
Despite extensive cultivation and testing of genetically modified foods, those questions still linger in the minds of many consumers. Now a new study in the March 2014 issue of The Plant Genome demonstrates a potentially more powerful approach to answering them.
In the research led by Owen Hoekenga, a Cornell University adjunct assistant professor, scientists used a water-alcohol solvent to extract roughly 1,000 biochemicals, or metabolites, from the fruit of tomatoes they’d genetically engineered to delay fruit ripening. They then compared this metabolic profile from the modified fruit to the profile of its non-modified parent variety.
Many metabolites, including pigments, amino acids, sugars, and various health-promoting compounds, are known to contribute to fruit quality and nutrition. And extracting and analyzing hundreds of them at once gives researchers a snapshot of the fruit’s physiology, known as the metabolome, which can be compared against others. In this way, metabolomic analysis is very similar to genomics, where geneticists compare DNA sequence data to understand how genetically divergent different organisms are.
When Hoekenga and his colleagues performed their analysis, they did in fact uncover metabolic differences in the modified fruit relative to its parent, although these changes were mostly seen in biochemicals related to fruit ripening, Hoekenga says. “So that’s part of an intended effect.”
But when the scientists compared the metabolome of the genetically modified tomato with those of a wide assortment of garden, heirloom, and other non-modified tomatoes, they found no significant differences overall. In other words, although the modified tomato was distinct from its parent, its metabolic profile still fell within the normal range of biochemical diversity exhibited by the larger group of varieties.
The finding suggests little or no accidental biochemical change due to genetic modification in this case, as well as a useful way to address consumer concerns about unintended effects in general, Hoekenga says.
He explains that the U.S. Food and Drug Administration already requires developers of genetically modified crops to compare a handful of key nutritional compounds in modified varieties relative to their non- modified parents. Part of biotechnology risk assessment, the process is designed to catch instances where genetic manipulation may have affected nutritional quality, for example.
The approach of Hoekenga’s team, in contrast, doesn’t decide ahead of time which metabolites are important to measure, suggesting it could be more likely to snare a truly unexpected impact. “We throw a net in the water and try to get as many fish as we can,” Hoekenga says.
Moreover, comparing a modified variety to diverse cultivars can help both scientists and consumers put into context any biochemical changes that are observed. “We accept that there isn’t just one kind of tomato at the farmer’s market. We look for diverse food experiences,” Hoekenga says. “So we think that establishing the range of acceptable metabolic variability can be useful for examining GM varieties.”
At the same time, this brand of non-targeted metabolomics is expensive, and the chemistry methods it employs aren’t robust enough yet to be used in official safety assessments, Hoekenga acknowledges.
Most importantly, making statistical comparisons of metabolic fingerprints is no easy task. In their study, Hoekenga’s group adapted a style of statistics, called network analysis, which was developed to compare overall patterns of gene expression, or transcription, in mice. The reason for this choice, he explains, is that just like gene transcripts, metabolites that participate in the same biochemical pathways or fall under the same regulatory control are expected to cluster together. And as the researchers hypothesized, network analysis allowed them to detect metabolic clusters in tomato and compare those patterns across different varieties.
But the techniques don’t apply only to tomato. “The method can be applied to any plant or crop,” Hoekenga says. “We’ve made something fundamentally useful that anyone can use and improve on.” His group has already characterized the corn metabolome, and he hopes plant breeders will begin to see the utility of metabolomics, as well.
When crossing parent plants, for example, breeders often like to track the genes underlying their trait of interest, such as resistance to a pathogen. That’s because pinpointing offspring that carry the right genes is often faster and easier than examining plants for the trait itself.
But sometimes, so many genes contribute to a single trait that figuring out which genes are involved in the first place becomes onerous. This is where Hoekenga thinks metabolomics and network analysis might one day help.