American Scientist

James J. Collins is the Termeer Professor of Medical Engineering and Science, and a professor of biological engineering, at the Massachusetts Institute of Technology, as well as a member of the Harvard-MIT Health Sciences and Technology Faculty. He is also a core founding faculty member of the Wyss Institute for Biologically Inspired Engineering at Harvard University, and an institute member of the Broad Institute of MIT and Harvard. He has helped launch a number of companies, including Synlogic, Senti Biosciences, Cellarity, and Sherlock Biosciences. He is one of the founders of the field of synthetic biology, and his research group is currently focused on using synthetic biology to create next-generation diagnostics and therapeutics, with a particular focus on using network biology approaches to study antibiotic action, bacterial defense mechanisms, and the emergence of resistance. Collins was a keynote speaker at Sigma Xi’s International Forum on Research Excellence (IFoRE) in November 2022, and after his talk he spoke with editor-in-chief Fenella Saunders in more depth about his research. (This interview has been edited for length and clarity.)

How do you define synthetic biology? What is the line between it and genetic engineering?

I define synthetic biology as a maturing field that’s bringing together engineers with molecular biologists to use engineering principles to model, design, and build synthetic gene circuits and other molecular components, and use these circuits and components to rewire or reprogram living cells and cell-free systems, endowing them with novel functions, and enabling a broad range of applications.

Looking now to the distinction between synthetic biology and genetic engineering, I think in many ways synthetic biology puts engineering into genetic engineering. Genetic engineering is very much about taking a gene from one organism and introducing it to another organism, and producing the protein from that gene primarily for industrial or therapeutic purposes. In many ways, it’s the notion of replacing a red light bulb with a green light bulb. Although there may be many jokes about how many engineers it takes to change a light bulb, that’s really not an example of engineering.

What synthetic biology does is create the circuit to control the expression of that light bulb. It might be that you can flip it on stably when you enter the room and flip it off stably when you leave the room. It might be that you have it so it’s controlled to flash on and off in an oscillatory fashion, similar to a Christmas light. It might be that it turns on and off at a set time of day. You’re actually introducing control. That’s what synthetic biology is about. It’s the underlying control circuitry that now enables one to program a living cell or cell-free system to sense its environment, make a decision on its environment, and act on its environment by producing an output. You’re introducing some biological element of sensing, decision-making, and acting, which in many cases had not existed before for that particular function.

So the distinction is that genetic engineering puts in new parts, but synthetic biology adds a control element?

I’ll give a contemporary example that has fallen under the banner of synthetic biology, but I put it in genetic engineering. That’s Impossible Foods or Beyond Meat. In this case, they’ve taken the heme gene and introduced it into plants, or in some cases in yeast, and it will now express heme to give a taste like meat to that plant or yeast. I think it’s marvelous. It has had an impact and changed the way we eat certain foods. But I view that as genetic engineering and not synthetic biology. It’s similar to what’s been done since the mid-1970s when recombinant DNA techniques were introduced. It was a brilliant insight into what you could introduce and have a meaningful alteration in its features and properties. But there’s no element of real control there.

One of your research projects uses bacteria as sensors. How does that work?

In our very early work on synthetic gene circuits 22 years ago, we built into bacteria and showed, for example, that we could create toggle switches, these stable memory elements that give you addressable, programmable memory in a living cell—in this case, bacteria.

We then explored whether we could create a programmable cell that could sense its environment, decide on what it sensed, and then give an output, which may just be, in this case, a biosensor. Our initial efforts, starting in 2004, were to engineer highly sensitive DNA damage-sensing systems to function in E. coli. We did it by coupling the toggle switch to the DNA damage-sensing response inside E. coli, so that when the DNA damage-sensing system was triggered, it would actually flip in the toggle switch and you could record the event.

Going forward almost a decade, we were challenged in the beginning of 2010s by the Gates Foundation to engineer bacteria to detect cholera. In this case, we modified Lactococcus lactis, generally regarded as a safer organism and found in dairy products. We reengineered it with synthetic circuitry and hybrid receptors that were based in part on repurposed components from the quorum-sensing system, the intracellular systems from cholera. We reprogrammed L. lactis to eavesdrop on the presence of cholera, so it could detect the small molecules given off by cholera and flip on circuitry inside L. lactis, which would then produce enzymes that could change the substrate to a different color.

What other places would you want to have this kind of switch?

It could be in bioreactors, where you only flip on the production of a protein of interest when you reach a certain density. You can envision it in therapeutic purposes, where you only want your living cell to produce the therapeutic, which might have some level of toxicity, at the site of disease. You could envision having the switch in an agricultural setting, where you might have a drought-resistant gene that may limit crop yield, but you want it flipped on in the face of a drought. You could have it off as you plant your crops and they grow and you have plenty of water, to maximize your yield, but in a drought you could flip it on stably with an induced signal that enables your plants to survive.

How can these devices work in theranostics, in which therapy and diagnostics are combined?

On the heels of our efforts to create these living diagnostics in L. lactis for detecting cholera, we also made efforts to create a living therapeutic, uncovering along the way that we didn’t need to actually engineer L. lactis. We were able to use its natural ability to produce lactate, and thus lower the pH in its microenvironment, as a means to both prevent and treat cholera infections. But we found, intriguingly, that when we introduced the diagnostic component to create, in a single cell, this theranostic capability, both to detect and treat, we actually eliminated the ability of those cells to produce lactate, because of the metabolic cost of the diagnostic circuit. Thus, our theranostic capability was actually accomplished in a population mixture, in which we mixed the engineered diagnostic cells with the natural therapeutic cells to both report out on infection state and treat infection.

How are you getting a readout from these various diagnostics?

On the living therapeutic that you would take orally, it would be a matter of changing your stool a different color with the enzymatic substrate. For other applications such as environmental detection, using the paper-based ones, it would be, in most cases, a color change on the paper that you could detect by eye.

What work are you doing in the area of antibiotic resistance?

Initially, we went after antibiotic resistance in our paper-based diagnostics. We showed that it’s possible to freeze-dry cell-free extracts along with a synthetic biology sensor onto paper that could then be rehydrated some time later with a patient sample or water. That enabled what had been freeze-dried to become reactivated, such that you could have transcription and translation function on a piece of paper the same as inside a test tube or inside a living cell. We had developed RNA sensors that could detect RNA given off by a pathogen, for example, and initially developed these sensors to detect RNA associated with antibiotic resistance. We showed that you could get readouts in 25 to 30 minutes for various seminal or critical markers of antibiotic resistance, and thus give important information to physicians in that golden window of an hour or two that they might have with a patient presenting in an emergency room with an injury that might be infected.

Going back to the late 2000s, we began working on engineered phage, a virus that specifically targets bacteria. The notion we had was, could we engineer phage to express proteins that could serve to boost the effectiveness of resisted antibiotics? Specifically, we decided to see if we could target the DNA damage response, similar to what we had done on the biosensors. Many commonly used antibiotics, downstream of their drug target interaction, will induce strong metabolic demands that lead to toxic by-products that cause damage to DNA proteins. We showed that with this engineered phage, we could deliver a protein that could basically keep the DNA damage response off, and thereby boost the effectiveness of antibiotics 100-fold to 10,000-fold. And most interesting, we showed that you could resensitize resistant strains to the antibiotic.

Some of your work uses other genetic material or enzymes and not whole cells?

This is this effort we helped pioneer around freeze-dried cell-free synthetic biology. Living cells will require special handling for storage or distribution. So we turned to cell-free systems that have been used for decades in molecular biology. The idea is you can open up a living cell, remove the cell’s inner machinery, and play with it in a test tube or petri dish. This machinery would include DNA, RNA, molecular machines such as ribosomes, as well as other biomolecules. We were able to show that you could take a cell-free extract along with a synthetic biology network or sensor, freeze-dry it onto paper or clothing, or without a substrate as a pellet, and then sometime later rehydrate what had been freeze-dried, and it now would function as if it was inside a test tube or a living cell.

This opened up possibilities to create paper-based diagnostics, wearable diagnostics, and portable on-demand biomolecular manufacturing. When the pandemic hit, we realized we had an opportunity to do something relevant. We came up with the idea that we’d create a wearable face mask diagnostic, a small insert that could be added to any face mask, which could use this freeze-dried cell-free synthetic biology to report out on infection state. We were able to design and build such a system quite quickly during the pandemic.

What are the limitations that remain in place for synthetic biology?

In many cases, biology is not yet an engineering discipline: We don’t have the design principles to create the objects of interest with the desired functionality, and we don’t have enough parts or components to engineer biology as readily as we would like. We don’t have the tools to measure the behavior of our components in real time as an electrical engineer would, nor the ability to create those components as quickly as we’d like, so that the time frame to design, build, test, learn, and do it again is much slower in biology. Harnessing machine learning will help us to better infer design principles and dramatically expand the number of well-characterized parts we have for engineering biology. I think these developments will enable us to make biology more of an engineering discipline.

A podcast interview with the scientist: