By Deborah Borfitz
January 6, 2022 | Researchers in Australia have succeeded in building a chemically induced dimerization (CID) system that can detect the concentration of the antimetabolite drug methotrexate in serum. Coupled with a functionally related biosensor of rapamycin, commonly used to prevent organ transplant rejection, they also looked at lab prototypes of a multiplexed bioelectronic system for repeated measurements of multiple analytes, according to Kirill Alexandrov, professor of synthetic biology at Queensland University of Technology’s Centre for Agriculture and Bioeconomy.
CID systems, in which two proteins associate only in the presence of a certain small molecule or other dimerizing agent, have been used as a tool for modulating biological processes since their invention nearly three decades ago. Here, they are being used to construct small molecule biosensors designed to capture biomarkers of choice and produce measurable responses.
Alexandrov’s personal goal is to establish a pipeline of clinical analytes for which a synthetic CID system could be rapidly produced. Building the methotrexate-controlled device took about six months, he says, but in the future “designer biosensors” that switch on color or electrical responses to different drugs could be producible in a matter of weeks.
The technology is based on reengineering of PQQ-glucose dehydrogenase (PQQ-GDH) protein—the principal component of electrochemical point-of-care glucose monitoring systems. GDH breaks down glucose and generates electrons as by-products to produce an electrical current proportional to the amount of the captured target molecule. It was redesigned by the research team so that the analyte of choice controls the enzyme’s activity.
A device based on such biosensors would be valuable for cancer patients put on high doses of methotrexate because their organs may start to fail if the drug is not excreted as expected, explains Alexandrov. Serious organ damage can result unless an antidote is administered in a timely fashion.
The long-term intention here is to aid development of point-of-care tests for therapeutic drug monitoring that could be performed outside of a central lab and potentially in a physician’s office or even a patient’s home, he says. Patients being treated with toxic drugs currently go to a specialized medical facility for monitoring by immunochemical and mass spectrometric methods and this data informs decisions about drug dosing and the introduction of rescue therapies.
One of the chief challenges that has been overcome, as covered in a substantial study in Nature Communications (DOI: 10.1038/s41467-021-27184-w), is how to build enough specificity in a large receptor protein to selectively and efficiently recognize small molecules. The answer was to evolve an artificial CID system that uses two different proteins that interact in the presence of methotrexate.
In a series of experiments using methotrexate, researchers demonstrated that the nature of the binding domains strongly influences the effectiveness of the selection process. Their artificial CID system was used to construct electrochemical protein biosensors and the resulting sensory electrodes, suggesting it could serve as the backbone of diagnostic devices offering fast information relay.
Diagnostic assays currently used for detecting drug molecules are predominantly based on “competition” between an antibody and a drug-reporter conjugate. This puts limitations on further development and miniaturization of such systems. In contrast, CIDs can be integrated in a variety of sensing architectures and molecular machines, says Alexandrov. The caveat there is that CID systems are not easy to engineer with methotrexate-controlled CID being one of perhaps five synthetic CID systems in existence.
Construction of multiplexed sensors will be important, says Alexandrov, as most diseases cannot be unambiguously assessed with a single biomarker. The question has been whether it is possible to resolve the signals coming from two or more biosensors in the same sample and the answer is “most likely yes.”
The research team was able to show the signals coming from methotrexate and rapamycin in their experimental multiplex system didn’t interfere with one another. “That opens the way to making systems where you have individual electrodes to signal-analyze but you can combine them in various ways and create panels,” he says.
The critical issue in bringing bioelectronic devices to the point of care for therapeutic drug monitoring will be whether a biomarker panel can be built for detecting analytes across all orders of magnitude, Alexandrov continues, noting that, for instance, glucose and insulin are present in vastly different concentration levels in the body. This is of course not a problem in a central lab, where scientists can easily dilute a sample or deploy any combination of assays.
Designing signal amplification methods and amalgamating results in a stand-alone application that is as easy to use as point-of-care glucose testing will not happen overnight, he stresses. But if protein biosensors can enable testing in small, regional or remote labs and hospitals using less sophisticated equipment, the effort could be well rewarded. “Proteins are at the core of a $70-billion-dollar global diagnostic market that relies heavily on central lab processing,” he notes.
Alexandrov is an inventor on patents covering electrochemical biosensor technology used in the study and holds equity in Molecular Warehouse Ltd that owns one of those patents. He is also aware that the road to market can be long and bumpy.
Much more than the core technology is needed to produce a widely used, real-world diagnostic test, he says. This includes use cases for the technology as well as regulatory approval and protein biosensor manufacturability and scalability.
The newly described diagnostic approach could be termed a “major incremental advancement” in the broader field of protein engineering, says Alexandrov. The basis of the field is the observation that proteins yield real-time energy as well as information as they switch on and off in the body on a millisecond to second scale.
Molecular biologists can engineer synthetic versions of naturally occurring proteins, making it possible to extract information from biochemicals in approximately real time, Alexandrov says. Biological systems can be redesigned out of standard components to give them new and useful abilities.
In Alexandrov’s case, the focus is on building a system to recognize a molecule, rapidly generating a signal that can be easily measured. Many molecules currently used in diagnostics—in particular, chemical pathology—are based on a similar principle except they use naturally occurring biosensors.
Nature hasn’t provided molecules to do everything scientists would like, he says, including the properties that are compatible with their intended use. “Therefore, we are building molecules with the desired inputs to recognize the desired analytes… [so they] behave like robots, making them easy to use.”
Building The Switch
In a prior research article published in Angewandte Chemie (DOI: 10.1002/anie.202109005), Alexandrov and his colleagues dealt with the question of how to build the PQQ-GDH switch so a protein molecule can detect the amount of current being produced in the presence of the desired analytes—in this instance, the drugs cyclosporine A, rapamycin, and tacrolimus that are all used for organ transplant management.
Using protein engineering and modular design, the research team built a system that could be switched from one clinically relevant analyte to another, he says. They also showed that those biosensors could be attached to an electrode surface, providing a prototype of a point-of-care diagnostic device.
When tested in colorimetric and electrochemical assays, the developed biosensors allowed cyclosporine A to be measured in a small volume (1 microliter) of undiluted blood with accuracy on par with the leading diagnostic technique using much more sample. The same technology was used to construct highly porous gold bioelectrodes capable of detecting cyclosporine A at concentrations as low as 20 picomolar.
A fingerprick volume of blood is ideal for point-of-care diagnostics because that would eliminate the need for a trained phlebotomist, says Alexandrov. “Yet the traditional diagnostic industry has also become a lot better at extracting more information out of a much smaller sample, so it is kind of a race.”
Research focused on the biosensors’ switching mechanism, as well as creating receptors that recognize methotrexate, were done in collaboration with Clarkson University (Potsdam, New York), Pathology Queensland, and Commonwealth Industry Research Organization.
The tests were designed for monitoring the concentration of a drug in patients’ blood to make sure it stays within the desired therapeutic range and doesn’t become toxic, Alexandrov says. Therapeutic drug monitoring is currently a $1.6 billion industry, driven by the lifelong need for testing among organ transplant recipients as well as patients being treated for cancer, epilepsy, and some autoimmune and cardiovascular diseases.
A certain proportion of patients needing therapeutic drug monitoring simply don’t comply with the advice to have it done, Alexandrov adds. This was one motivation for developing tests that could be more ubiquitously deployed on a smaller device.
Modular design of the novel biosensors alleviates difficulties in working with proteins, which are complex and fragile, he says. They could be adapted to potentially target any small molecule—not just therapeutic drugs.
In addition to diagnostics and pharmacokinetic testing of drugs under development, Alexandrov’s envisioned analyte pipeline would touch multiple disciplines including wellness and various environmental and military applications. Since the biosensors are based on PQQ-GDH, they would be compatible with disposable sensor electrodes.