As a Berkeley bioengineering undergraduate, I was imbued with the gospel of metabolic engineering. The potential to use self-replicating microorganisms to produce useful chemical and therapeutic compounds in an eco-friendly way has not only attracted vast government funding, but also venture capital to commercialize these biological processes. For me, the success story of synthetic artimisinin production from the Keasling lab at UC Berkeley was a shining example of the tangible impact scientific discovery can have on industry and society.
The current vice president of R&D programs, Tim Gardner, at Amyris, the company that first developed and optimized artimisinin production in genetically engineered yeast strains, gave an early morning talk Tuesday. I was both excited and curious to learn more about this company. Gardner presented a clear overview of the pipeline involved in selecting and optimizing yeast strains for production of the company’s current product, farnesene, a compound that can be converted to biofuels. Gardner highlighted that the R&D process uses much of the fundamental tools developed in academic research such as directed evolution and screening techniques. But the main difference is that these processes are mostly standardized and automated at Amyris. Gardner listed four broad principles that have facilitated the productivity of the company.
- Systematic storage and tracking
- Quality control
Standardization of the “DNA parts” used to construct yeast strains lends the system to automation. Automation of the cloning process improves throughput of strain construction and screening. Careful tracking and storage of information in databases enables more efficient tracking and communication of information. And lastly, rigorous quality control helps to ensure reliability of results.
Gardner moreover made an interesting point about reproducibility of results in the development pipeline, saying that the limitation to programming cells is not the ability to manipulate DNA but the ability to make accurate measurements. Gardner’s points on standardization and automation seemed like the bread and butter protocols for many industries, but his comment on the nature of engineering microorganisms was more particular to the biological enterprise. Unlike traditional industries, the metabolic engineering companies rely on performance metrics that are inherently noisier due to biological and technical errors. To emphasize his point, Gardner told an anecdote about increasing farnesene yield. A reduction in the measurement error from 2.3% to 0.5% made a dramatic difference in production because it revealed the location in the development pipeline that required optimization and this revelation resulted in an increase in both reproducibility and the yield of farnesene.
Having previously been an academic who authored one of the seminal papers in the field of synthetic biology, Gardner keenly noted that industry has a more ascetic tolerance for error than academia. For example, a relatively small error in data collection is generally acceptable in academia, but might not make the cut in industry. Gardner attributes the irreproducibility of experiments some times observed in the biological literature to the lack of precision in measurements. While the amount of precision required in an experiment is dependent on various factors such as the signal to noise ratio of the experiment, the uncertainty in measurements clearly obscures the ability to extract more nuanced determinants of a biological phenomenon, and thus lead to irreproducibility of experiment when performed different environments. This brings up the question of whether there should be greater emphasis in academia for tighter error bars to improve the reproducibility of biological research. The title of the symposium was “Biophysics in Industry”, and highlights the impact of biological research on the biotech industry. But perhaps there is also something academia can learn from industry.