We had a retirement dinner for one in the “Original Family” of the event selection team that I had the privilege to lead. This led me to note six lessons learned in hopes that it will help future crop technology entrepreneurs.
Constructs are key. Think of Constructs as new ideas and Events as variations on a theme of that idea. Tweaking a bad idea still results in a bad idea, and inserting a poorly performing construct in different places in the genome will have disappointing results.
Event generation is the Siren’s Song of trait development. Being tempted to find the needle in a haystack of your favorite construct is a sure path to shipwreck.
The purpose of field trials is to quickly identify and discard non-commercial constructs. Spend resources and careful planning on data quality. Any construct showing significant Genotype Environment interaction (GE) must be discarded along with all events in that construct. Save your resources for new Construct
All GMO and Gene Editing events should have a qualitative, single-gene dominant phenotype. If they don’t, discard the event because the marketplace expects the effect of the trait to always be present.
The genome of almost all crop plants contains repetitive sequences, and any assumption of single insert must be rigorously challenged. Either you will do it, or your regulatory group will, and both should kill multi-insert events.
Be patient for your first released variety. Nothing kills a trait launch like the performance of old genetics and customers no longer accept average performance from a new trait. No trait added yield or crop value, it only protects the genetic gain from the breeder. Corn Enogen is the only exception to this rule.
We evaluated thousands of events and commercialized more field traits than most companies. The work of the trait research scientists was creative and generally successful when a target protein was known somewhere in Nature. From my experience, I would develop a trait pipeline as follows:
Generate 15-20 constructs per functional gene. Thoughtfully modify promoters, terminator sequences, regulatory sequences and remember to take advantage of signal sequences that may sequester your protein in useful places.
Transform into the best possible true breeding line available.
Generate no more than 100 T1>T2 plants representing events in the greenhouse per construct. Select strongly for trait expression and seed production capability.
Field test no more than 25 events per construct. If you don’t have a commercial quality event in that cohort, you likely won’t generate one for that construct.
Do not plan a return to the transformation lab to generate more events of a moderately successful construct. You have not got the time and the likelihood of improved performance is low.
Hire a top-notch statistician which reports to the Head of Trait Development, and not the scientists nor the business. This role is the equivalent of an auditor and must be independent of those with vested interests in the success of the project pipeline. Keep the data analysis simple. The first indication of a poor cohort of constructs is a sophisticated data analysis with irrelevant statistical precision.
Beware of statistically significant results caused by high numbers of replications. Field variability on the farm scale requires at least 20 bushels for corn and 5 bushels for soybeans to be seen under your stress conditions.
Be ruthless but not brutish when culling constructs and events. Show empathy to those who worked hard to bring them to the field.
I always resisted those who thought a trait brought market value and this notion has hurt the industry. The only thing that a GMO input trait can do is protect what the breeders have accomplished. The only thing that any gene editing technique can do is create variability that can be utilized by the breeder to improve the genetic base. The only exceptions to this are Golden Rice and Corn Enogen, which are not input traits. The lesson is to fully fund your breeding group and concentrate on where your sales staff has critical mass to win.
Finally, a trait should be deployed in only one crop. Glyphosate tolerant soybeans were great. Glyphosate tolerant corn and cotton was a step too far and we are paying the price. Greed brought on resistance in weeds and insects. Not so much corporate greed (although that likely played a role), but the greed of efficiency when applied to the entire crop rotation. Keep an eye on Nature to monitor what the next stressor will be, and if your solution to a stressor is maintaining its effectiveness.
Take these thoughts for what they are worth but ignore them at your own peril. In the end, it is the unsexy, stepwise genetic improvement of the breeders that will feed the world. All input traits can do is protect what they have accomplished, or in the case of gene editing give them a source of variation where none existed.
Comments