Starting with the end in mind
“Start with the end in mind.” So says businessman and leadership icon Stephen R. Covey in a statement that has permeated modern culture across disciplines. Turns out, it may also be one of the best ways to approach designing biological circuits.
That, at least, is the idea underpinning Brian Bachmann’s lab at Vanderbilt University. Dr. Bachmann recently sat down with the PLOS Synthetic Biology Community to talk about his background, current work, and thoughts on some of the approaches being tested and used within the synthetic biology field.
Making the decision: academia or industry?
Bachmann received his PhD from Johns Hopkins, where he worked on the biosynthesis of beta-lactams and discovered beta-lactam synthase. This was in the late ‘90s, and Bachmann was conflicted between a career in academia (which he had fallen in love with) and industry (which was his original plan). Either way, though, he was interested in the connection between genomics and natural products.
Once a year or so, someone would publish a completed gene cluster for biosynthesis of a natural product. Each paper that came out stimulated deep conversations between Bachmann and his colleagues about the data and consequences.
In relatively short order, the field reached a point where scientists could begin to predict the structure of products from their gene clusters. Today, he says,
“I think this field is probably the only one where we’ve been able to translate principles to actually predict concrete output of gene sequence data, down to very complex 3D structures including stereochemistry.”
Somewhere around the end of graduate school, Bachmann attended a conference and saw talk by Ecopia Biosciences. “They had developed a technology to very rapidly sequence gene clusters,” which reduced the time it took to sequence a gene cluster from a year or many months to about a week. “I got so excited by the database they had that I couldn’t resist the opportunity to […] go up to Montreal and work for them.” There, Bachmann worked on the genomics of secondary metabolism.
Finding opportunities from the crash
As with most of the industry, the 2001 tech crash had a dramatic effect on publicly held Ecopia. The company realized that selling genomic data “was not a business model.” A shift took place, and Bachmann was given the assignment of trying to isolate molecules predicted by sequence data. They wanted him to “put your money where your mouth is,” he says, and obtain real substances out of the predictive data he was working on. Bachmann led a group within the company, and they soon reached a point where they were isolated one compound per month.
It was an exciting time, but Bachmann started thinking a change might be in order.
“I realized that I missed academia. While I found the job very rewarding and stimulating, there were some things that I felt I needed to do scientifically.”
He started looking for academic jobs while continuing to build the team at Ecopia and isolating new molecules. In 2003, Bachmann took a position at Vanderbilt University.
“The reason I got into natural product biosynthesis in the first place was, coming from a synthetic background, an interest in ‘can we do chemical synthesis, total synthesis, within organisms?’ So I came to Vanderbilt specifically to engineer biosynthetic pathways to make non-natural products. The level of risk was really more appropriate for an academic setting.”
Evolving in the university
Today, part of his lab is continuing this synthetic biology project, working to make non-natural products through biosynthetic pathways.
“My vision was, could we take the lessons learned from reading the blueprints of natural product biosynthesis for twenty years to the point where we could predict accurately what they make? Can we take those lessons and apply them to making new compounds?”
Directed evolution and enzyme engineering play a huge role in this field, and Bachmann came of age scientifically as that work was coming online. In order to make non-natural products out of novel biosynthetic pathways, though, the pathways themselves obviously have to be optimized. Thinking about ways to do this led Bachmann to a position that, well, puts him somewhat at odds with a significant part of the synthetic biology movement.
Picking apart parts-based design
“I decided to focus on aspects of pathway assembly. I decided it was an unsolved problem. There is a big movement in synthetic biology called the ‘parts-device approach.’ We are going to build a toolkit of parts and then assemble them into devices. I believe that is a fundamentally flawed approach when it comes to biosynthetic pathway engineering. Enzymes are remarkably selective for their substrates, and fluxes need to be controlled and engineered in as you build. So it seemed to me that we would not be able to build the equivalent of electrical circuits that create compounds, because of the amount of evolutionary optimization [required for] each part.”
Bachmann believes that this specificity ultimately causes a breakdown in the parts-device method of synthetic biology, to the point where he suggests it should be abandoned.
Additionally, enzymes are still something of a black box, and scientists still don’t really understand how they work. “We have transition state theory, but that’s not enough. We don’t understand them enough to design them, yet. So I think that for the part-device approach to work, we’re going to need to design catalysts rapidly,” which will include evolutionary and computational technologies. If you have to make a custom part for every component, Bachmann wonders, is the part-device model worth it?
Finding an alternative
Instead, his group focused on processes derived from evolution. “The thinking based on looking at sequence data is that [biological systems] recruit enzymes from primary metabolism and start to concatenate them,” he says. “The products that are formed create value-added materials that give [the organisms] an evolutionary fitness advantage.” Eventually, over evolution-scale periods of time, pathways are assembled.
“This is the model of what I would say is done in pathway engineering. You have recruitment, sometimes there’s simultaneous recruitment of each enzyme for a given substrate, but ultimately they’re put together in a forward linear fashion to try to get to a product.”
However, this approach didn’t satisfy Bachmann, either. “We decided this was too inefficient for our scale of endeavor.” He went back to his roots in chemical synthesis and looked at a principle called “retrosynthesis,” essentially reverse engineering. And here is where the principles of a world-famous leadership expert converge with evolutionary biology: begin with the end product and work backwards to determine how to build the pathway. “Let’s start with the last step, evolve the last step to make the product from an available precursor and enzyme. Then we’ll go to the penultimate step and evolve that enzyme in the presence of the optimized enzyme, again selecting for product.” And so on, always selecting for product in a process termed “bioretrosynthesis.”
While problems can arise regardless of which direction you might be working in, Bachmann thinks that retrosynthesis leads to fewer dead ends:
“The number of possibilities starting from the beginning is much, much larger, and the number of branch points is much larger, that could lead to potential dead ends. But if you’re always using that product selection, then you always have your eyes on the prize and you only need one selection method to evolve all these enzymes […] The efficiency is much greater.”
Following in the footsteps of an icon
Bachmann built his group around testing this model of design and quickly realized that he was late to the party. Norman Horowitz originally developed the idea of bioretrosynthesis at Caltech in the 1940’s as an evolutionary theory, suggesting that it was how biosynthetic pathways actually came about. (For more on the fascinating Norman Horowitz, see his 2005 obituary in the LA Times and this interview he did in 1984 for Caltech.)
“Perhaps inadvisably,” Bachmann says with a smile, his group decided to apply principles from Horowitz to an entirely new pathway. The goal? To synthesize nucleoside analogues – completely unnatural products that are used in the pharmaceutical industry as anti-viral reverse transcriptase inhibitors.
“The reason we chose them was because we could imagine using the toolkit that primary metabolism has supplied and adapting it to make them. We also determined that they’re very expensive to make chemically […] due to the high cost of the starting materials.”
However, nucleoside analogues are relatively straightforward to synthesize, only taking a handful of steps. “We thought that if we could make the molecule biosynthetically rather than chemically, we could lower the cost and make it less financially toxic,” he says. “It was worth doing and it was possible.”
Validating backwards thinking
According to the plan, Bachmann’s group started from the end and worked backwards, with design and optimization each enzyme in the pathway becoming a PhD project. They published a paper last year describing the process, which to Bachmann’s knowledge is the first example of a pathway like this being applied to make a compound.
“I was excited to get it out there, I want people to adopt it.” When asked if he thought having the method published and somewhat validated would indeed lead to wider adoption, Bachmann said, “I’ve been getting a lot of good feedback, a lot of buzz. When I talk to people, they say, ‘I’m going to do that!’ So I’m optimistic.”
However, he points out that not everyone is sold, and he is unsure about how the engineering community will respond.
“There are some disadvantages to the approach, namely, it’s linearly contingent and you can only optimize one enzyme at a time. With the other method, you can evolve all of the enzymes simultaneously.”
Even so, to date there have been very few examples of a successfully engineered pathway built to produce unnatural products. According to Bachmann, “there are lots and lots of pathways for recapitulating natural products, there are biofuel pathways, but for making a totally non-natural compound in a multistep pathway…it’s very rare.”
Working the middle ground
Taking a broader perspective of the field, Bachmann groups small molecule synthetic biology into three main categories: recapitulation of existing pathways, development of de novo pathways, and “tweaking,” which sits somewhere between the first two. In this last area, researchers take existing pathways and make modifications to produce a desired outcome. In addition to their work in bioretrosynthesis, Bachmann’s group is also interested in the tweaking process, which he points out is far more feasible than creating a pathway de novo.
“The big excitement in that area is that […] with recombinant technologies, we can do SAR [structure-activity relationship] to some extent by changing the gene clusters that encode natural products to make analogues not found in nature. I think the potential of that is huge […] we should be able to generate dozens or hundreds of scaffolds and optimize them as human drugs.”
It seems to be working, and Bachmann’s group has developed analogues to a complex molecule that had previously been pulled from the clinical trial pipeline due to issues of clearance from the body in different populations. The hope, then, is to take something “so close to being a drug” and tweak its biosynthetic pathway to make favorable structural changes that will eliminate the problems. At the same time, the team is gaining deep insight into the enzymology of the natural pathways. Thus, Bachmann points out, the project is also valuable from a basic science standpoint.
Looking to the future
So where does Bachmann see synthetic biology headed?
“I think the future of the field will be the successful ability to engineer enzymes that do non-natural reactions, which is the holy grail of the field […] I feel strongly that if we can build on chemistry and physics, from the ground up, that that’s the solution to engineering in biological systems. Biological reductionism, on the other hand, does not work. Representing proteins and circles, and reactions as arrows, is not predictive, it’s descriptive. You can’t engineer with it.”
What is exciting for Bachmann, and encouraging for the field as a whole, is that there has been progress in this area. He points to groups designing reasonably functional enzymes and developing better computational modeling systems that help the whole process.
“They’ll engineer enzymes to do a non-natural reaction, and then they still have to use directed evolution to optimize it.” “But,” he says putting in one final plug for his method of choice, “I still think that doing it retro might have some advantages.” For Bachmann, at least, forward thinking will continue to require looking backwards.
For more on bioretrosynthesis, check out Bachmann’s 2010 review “Biosynthesis: is it time to go retro?” in Nature Chemical Biology. Then, let us know what you think in the comments. We’d love to hear your questions and comments on the pros and cons of various pathway engineering methods. So please, jump in!