The Foundry: Scaling Up Biological Design
Above: Depicted, from left to right: Chris Voigt, Ben Gordon, Rob Nicol. Photo by Lillie Paquette / MIT School of Engineering.
Living organisms are amazing feats of engineering: By following instructions encoded entirely in DNA, living systems can sense and respond to their environment, build intricate structures and materials, and churn out complex chemicals. How these abilities are encoded is undeniably complicated, but figuring out how to embrace this complexity is at the heart of the synthetic-biology facility known as the Foundry.
Located in Cambridge, Massachusetts, the Foundry is taking on the enormous task of designing, fabricating, and testing large sequences of DNA—20,000 bases long and more—at never-before-seen scales. Such capabilities will ultimately allow an academic, commercial, or government lab to produce new research tools, unique pharmaceuticals, novel materials, and just about anything in between both quickly and inexpensively.
With a five-year, US$32 million contract from the Defense Advanced Research Projects Agency (DARPA), the Foundry is making progress toward realizing these ambitious objectives and helping researchers from many backgrounds to begin broadening their views of what is possible through synthetic biology.
From the Ground Up
Back in 2003, the Human Genome Project revealed the genetic blueprint for human beings, and in so doing, put into high gear the ambitions of myriad research groups that hoped to begin modifying the DNA of organisms, much as electronics engineers were altering the function of semiconductor chips by changing their programming. Those plans had one major hitch however, according to Ben Gordon, director of the Foundry. “There has long been the promise to be able to engineer biology just by combining the right genetic parts. But that is where analogies to other engineering fields fall apart: It may work on the drawing board, but once we start working inside a living organism, it’s kind of a mess.”
That “mess” translates into a slow and costly debugging process, he says. “It would take months for every iteration of designing a genetic program, constructing the requisite DNA, evaluating the program’s performance, and deciding what to do next.” Moreover, it usually took many trial rounds to tune the program to behave as desired. “The process could easily end up costing tens of thousands of dollars each time you make a little change,” he says.
Such a path was just too slow and costly for Christopher Voigt, MIT biological engineering professor and co-director of MIT’s Synthetic Biology Center. Voigt started thinking about how to accelerate the process by breaking it down into different areas for scale. Regarding genetic design, Voigt says, “It was clear that we were getting bogged down by this pen-on-paper type of design process. People were actually using Microsoft Word—in fact, I had this big Word file that had all my little pieces of DNA in it. So you had to become kind of a ‘DNA whisperer’ to put those pieces together to design a living system.” Design was just the start. The researcher would then have to go to the lab to painstakingly construct the actual living system, and even if that went well, he or she would still have no way to ensure that the system worked exactly as intended because the testing consisted of indirect measurements, he remarks. “So if something goes wrong, you don’t know exactly what went wrong.” That, in turn, led to more experiments to try to correct the problem, and each of those experiments were similarly expensive, time-consuming, and often non-conclusive.
Rather than reinventing the wheel, Voigt looked to other industries to see how they injected economies of scale to help with complicated engineering challenges. His attention fell on SEMATECH, which was a DARPA-funded facility for chip manufacturing. The SEMATECH facility began operations in 1988 as a research and development plant that used rapid prototyping and high-throughput manufacturing to make chip design, testing, and production faster, cheaper, easier, and more practical—all things that would elevate the field of synthetic biology.
Voigt then approached Rob Nicol, the director of Technology Labs at the Broad Institute of Harvard and MIT. The Broad had a reputation as the world leader in DNA sequencing, and Nicol was one of the key architects of the Institute’s Genomics Platform, where he established manufacturing processes for biological research. Nicol liked what he heard, and he and Voigt got to work setting up the Foundry in 2013. Nicol recalls, “The intent was to combine all of the amazing synthetic biology Chris envisioned, the high-throughput expertise that I had in my group, and the advanced technology developments that also existed at the Broad to create a technology infrastructure that would take synthetic biology to the next level.”
Work is well under way, and with the major infusion of DARPA funding they received in May 2015, the Foundry has brought in about three dozen new staff to develop the complete design-build-test cycle that Voigt and Nicol envisioned. “The Foundry is a genetic-design institute that is developing the computer-aided design packages to simplify decision making; has the robotics and the processes that can take the DNA sequence off the computer and physically construct it; and then once it’s built, can combine all the techniques that are necessary to see what happens after that DNA is put into the cell,” Voigt says. “The whole idea of the Foundry to serve as a genetic design institute that offers capabilities to help separate the designer from the construction and the evaluation of the biology that was designed.”
Doing the Work
Such a segregation of design from construction and testing will free up researchers to concentrate on their projects’ objectives. According to Voigt, this is the same evolution that nanotechnology, chemical engineering, and chip R&D went through: While the rudimentary principles of the design-build-test cycles are being worked out, researchers have to perform every aspect of the cycle hands-on. As the stages are perfected and finally automated, the researcher is liberated from those aspects of the cycle. In other words, the researcher can start spending much more time in the driver’s seat and far less under the hood. “This would really turn biology into an information science, so that anything that a designer can imagine, he or she can access,” he says.
To accomplish that, the Foundry is developing innovative technologies to streamline all parts of the design-build-test cycle, Nicol says. “Around design, for instance, we have advanced new software that you can basically use as the genetic equivalent of a CAD tool to research your designs efficiently, and in ways that you simply cannot do on Excel or in Word.” On the build side, he points to advanced molecular-biology and automation tools that access the designs from that software, and connect to robotic systems, which realize them physically as DNA circuits and constructs that instruct cellular pathways.
One of the ways the Foundry can operate so nimbly is that it takes a combinatorial approach to building large pathways. “We are able to build and test lots of permutations of genetic programs because rather than having to synthesize each 20,000-base pathway, we can build individual smaller modules and reuse them in different ways to produce variants of the entire pathway,” Nicol says.
“The Foundry’s pipeline also includes massively parallel assays. So rather than evaluating one genetic program at a time, we are developing technologies to test 100,000 or even a million at once,” he adds. This is possible because the Foundry has high-throughput systems using microfluidics and droplet microfluidics that can build and evaluate at large scale. “We can test all these designs efficiently in one round essentially, and get vast amounts of data that will show which pathway is most successful as well as inform the algorithms and software,” Nicol explains. “That means we can close the loop and just iterate quickly through this cycle with lots of simultaneous experiments, and in the end enable rapid progress.”
Applying the Capabilities
Although the Foundry has just a three-year history, it is already allowing advances in the synthetic-biology field. Voigt’s research group is one of those benefiting. “We actually created the first programming language for bacteria, but without the capabilities of the Foundry in fully evaluating it, it would have been exceptionally challenging to do,” he says. This project grew out of his group’s interest in building new sensors and signal-processing algorithms that can be placed in cells, so that the cells will sense and respond in set ways to certain conditions in their environment. The research team uses a programming language from semiconductor chip design called Verilog to describe a circuit, and uses custom software to convert the Verilog code into a DNA sequence to put into cells. “With this technology, we were trying to control exactly when cells turn on different responses under different conditions and at different times, and we’ve done that by creating this programming language,” he says.
Voigt’s group is also using the Foundry to engineer the pathway that certain bacteria use to make iron oxide nanocrystals. “The microorganisms build these iron oxide nanocrystals with a precision that you can’t get in chemistry, and it involves an enormous pathway,” he says. His group has been working in collaboration with the Department of Defense, which is interested in the crystals for antennas and other applications. Voigt’s goal is to manipulate the pathway. He describes, “We want to control the size, the shape, and the composition of the nanocrystals, and do it for our purposes versus what the pathway in a biological system had in mind.”
Foundational technology development has been a primary part of the Foundry’s activities, but always with commercial and other users in mind, Gordon asserts. “Rather than confining ourselves to the ivory tower, we are formally structured to partner with companies through our industrial consortium so that our technologies will specifically feed into corporate-research or corporate-product pipelines. That keeps us on the straight and narrow, so we can be sure our research is headed in relevant directions,” he says.
As an example, he refers to a project to explore the microbiome, the diverse community of bacteria and other microbes living in and on the human body. “We know that microbes can affect our physiology through molecules that they produce, but we do not know what those molecules are. Moreover, the microbes do not grow in the lab, so traditional ways of studying them just don’t work,” Gordon says.
In response, the Foundry is working with Professor Michael Fischbach from the University of California San Francisco, who has a bioinformatics pipeline that can scan the many megabases of genome sequencing data available for the microbiome, and determine which clusters of genes are likely making the health-related molecules. Fischbach sends that information to the Foundry, which then reprograms the clusters for compatibility with laboratory strains, such as E. coli or yeast, and builds the requisite DNA. The strains follow the new genetic program and begin pumping out the molecules of interest, he says. “At the Foundry, we can not only measure which new molecules are being made, but we can also characterize them, and even try to evaluate the physiological effects they may have.”
For extended characterization, the Foundry is also partnering with a pharmaceutical company to scan the molecules for interesting immunological, anti-inflammatory, or other properties that might be useful as therapeutics. In the end, Gordon says, the basic science aspect advances the understanding of how the microbiome works, while the application aspect seeks out potential new pharmaceuticals.
Applications also encompass materials, agricultural, and other industries. For instance, he notes that the Foundry is working with a large European chemical company to engineer yeast to make green plastics, and with global researchers to re-engineer the bacterial processes behind nitrogen fixation that is essential to plant growth.
“The Grand Vision”
“The grand vision of the Foundry is to tie everything together: the design, the building, and the testing,” Gordon says. He uses the analogy of a huge test kitchen: Rather than refining a recipe over many years by preparing one dish at a time, a better approach is to equip a kitchen capable of making thousands of different recipes simultaneously, each with slightly different ingredients, cooking times, or other alterations. This allows for the study of the entire set together to find common aspects between the best and worst results as a means of arriving at the perfect recipe. “That’s what we’re doing at the Foundry,” he affirms. “We may not be able to approach the complexity of biology from purely a reductionist, engineering standpoint, but we can be smart about figuring out how to address the complexity head on.”
This approach will give synthetic biologists the ability to undertake projects that seemed out of reach just a few short years ago. Nicol comments, “At the Foundry, we’re most interested in projects that are impossible in academia or impossible in industry, and asking really big questions using the platforms that we have: Can we make every molecule that every organism that’s living on us makes? Can we access new chemicals that aren’t possible in materials science or chemistry?”
He adds, “Ultimately, we want to be able to design living systems at a complexity and a level of sophistication that we know is possible but we just don’t have the capability yet to do.”