A prototype electronic sensor that emits current when it detects the anticancer drug afimoxifene. Rice University synthetic biologists created the device to demonstrate a new method that could slash the costs of creating wearable monitors for precision, automated drug dosing of chemotherapies and other drugs. (Image: Jeff Fitlow/Rice University)

Rice University synthetic biologists have found a way to piggyback on the glucose-monitoring technology used in automated insulin-dosing systems and make it universally applicable for the monitoring and dosing of virtually any drug.

In a recently published study in Nature Communications, researchers in the lab of Caroline Ajo-Franklin, Professor of biosciences, demonstrated the technique by modifying a blood-glucose sensor to detect the anticancer drug afimoxifene, an estrogen inhibitor that patient’s bodies also make after they take the chemotherapy tamoxifen.

By building on mature biosensing technology that’s commercially available at most drug stores for under $20, Ajo-Franklin’s team hopes to speed the development of automated dosing systems for chemotherapies and other drugs as well as other technologies for real-time monitoring of biomarkers in the blood.

“The dream is to have technology similar to what’s available today for monitoring and treating variations in blood glucose, and have that be true for basically any drug,” said Ajo-Franklin. “Millions of people use blood-glucose monitors every day. If we can use that same basic technology to monitor other drugs and biomarkers, we could move away from the one-size-fits-all dosing regimens that we’re stuck with today.”

The heart of blood-glucose monitoring technology is a biochemical reaction in which specific proteins bind to glucose molecules and release electrons. Millions of these reactions take place within seconds, creating a small electrical current that is proportional to the amount of glucose in the blood sample.

Rong Cai, a postdoctoral research associate and the lead author of the study, tested more than 400 slightly modified versions of the electron-releasing protein and found a version that reacted with afimoxifene, reducing the current output from the glucose reaction in the blood. This allowed the team to detect the presence of afimoxifene by comparing the current produced by the regular glucose test to the reduced current from the modified test.

To demonstrate the technology in an electronic device, Ajo-Franklin’s team worked with the research group of Rice engineer and materials scientist Rafael Verduzco to create an afimoxifene sensor that emitted a current when the drug was detected.

Rice University synthetic biologists (from right to left) Caroline Ajo-Franklin, Chiagoziem Ngwadom, and Rong Cai worked with Rice engineer Rafael Verduzco (left) to create and demonstrate a method of universalizing blood-glucose detection technology as a way of rapidly and inexpensively creating sensors that can monitor the dosing of chemotherapies and other drugs in real time. (Image: Jeff Fitlow/Rice University)

Here is an exclusive Tech Briefs interview — edited for length and clarity — with Ajo-Franklin.

Tech Briefs: What was the biggest technical challenge you faced while developing this sensor and technology?

Ajo-Franklin: The biggest one by far was our limited ability to predictively engineer proteins to act as sensors. And that's honestly a grand challenge that the entire field — actually, multiple fields — is facing.

We recognized that we didn't know how to do it rationally, so we took a comprehensive approach. That was just courageous, blood, sweat, and tears from my postdoc Rong Cai and her colleague.

Tech Briefs: Can you explain in simple terms how it works?

Ajo-Franklin: It's really based off of how glucose sensors work, the ones you buy at the drug store all the time. Basically, in a glucose sensor, you have a protein that will break down glucose, and, in so doing, will create an electrical current. What we said is, ‘Oh, great, that's an amazing sensor.’ But we want to sense more than glucose. So, what we did is basically engineered that protein so that that breakdown of glucose and production of the current would only happen when an analyte — an estrogen mimic — bound to our engineered protein.

We literally slammed together this protein that breaks down glucose and makes electricity with a protein from humans that detects estrogen in your body. So, we took those two things and figured out: How can we get this protein from a human and a protein from a microorganism, slam them together, and get them to work together?

Tech Briefs: The article I read says your lab is already working on both ways to improve the sensitivity of glucose-based drug tests and methods to rapidly identify glucose-oxidizing proteins that can detect drugs other than afimoxifene. My question is, how is that coming along? Is there any news or updates you can share?

Ajo-Franklin: We really created a method to make chimeric proteins that can both detect an analyte and essentially transduce that into an electrical signal. We're going to use that same idea and expand it.

We did something that can detect estrogen, but we were thinking: Why not other therapeutics of interest? Or what about other proteins that detect, for example, other human steroids? Because that was of course, greatly interesting.

Tech Briefs: Are there any other updates you can share?

Ajo-Franklin: What I can say is, and again, my team just reinforces this. So, we want to find out if there is a way that we don't have to reinforce it. Is there a way that we can use evolution to actually do some of the work for us? That's what we're working on.

Tech Briefs: Do you have any advice for engineers or researchers aiming to bring their ideas to fruition, broadly speaking?

Ajo-Franklin: I think one of the things that the reviewers of this paper and our colleagues have recognized that's very valuable is we went end-to-end. We didn't just do the protein engineering, although that was probably the hardest part. We also involved material scientists and even electrical engineers — that allowed us to make an integrated device. It also allowed us to have more tools in the toolbox to solve this problem.

So, for example, we worked really hard to engineer the protein and there were some problems with it. It couldn't distinguish if it was the glucose concentration that was changing or our therapeutic concentration exchange. So, instead of trying to engineer the protein to overcome that problem, we used a kind of electrochemistry electrical engineering approach to solve that problem. We made two electrodes that could make measurements — one would measure glucose and the other one would measure the glucose plus therapeutic. And by comparing them we're able to isolate the therapeutic. So, use all the tools in the toolbox to solve the problem.

Tech Briefs: Is there anything else you'd like to add that I didn't touch upon?

Ajo-Franklin: I think this is a super-exciting time for this field because, as I've alluded to, the protein engineering is the most challenging part. But with the advent of new machine learning and AI approaches, there's optimism that we won't have to brute force this, that we can use those algorithms to say, ‘Hey, here are five things. Try those five things and you'll probably get at least one.’ Instead of what we did now, which is we tried 400 things and got a couple to work.

So, many people in the field are excited that maybe we can not only overcome these challenges, but maybe these machine learning tools will help teach us humans . There's also optimism that the algorithms can teach us humans how to do this by looking at a level of complexity and distilling it down in a way that we struggle with. It's hard to look at 450 variables all at once as a human.