Creation of a point-of-care therapeutics sensor using protein engineering, electrochemical sensing and electronic integration

The study addresses the challenge of rapidly detecting clinically relevant biomarkers directly from blood using point-of-care (POC) devices despite the complexity and opacity of blood. Leveraging the widespread availability and manufacturing base of glucometers, the authors aim to repurpose glucose-oxidation-driven readouts to sense a therapeutic metabolite. Specifically, they seek to develop a glucometer-based allosteric sensor (GBAS) that detects 4-hydroxytamoxifen (4-HT), a key tamoxifen metabolite used in hormone receptor-positive breast cancer therapy. The hypothesis is that engineering an allosteric linkage between a 4-HT-binding domain and glucose dehydrogenase can modulate glucose oxidation currents in response to 4-HT, and that appropriate electrochemical interfacing and signal processing can decode the 4-HT signal independent of varying glucose levels. The work aims to create a rapid, selective, miniaturizable, and potentially self-powered POC sensor with integrated signal amplification.

The authors highlight the clinical importance of extracting biochemical information from blood and the paucity of POC devices capable of doing so. Prior efforts to adapt glucometers for diverse analytes have been explored to exploit existing manufacturing pipelines. Alexandrov and colleagues demonstrated engineered PQQ-dependent glucose dehydrogenase (GDH) as a platform by inserting analyte-recognition proteins to detect compounds like rapamycin, tacrolimus, amylase, and cyclosporine A. Other protein switch designs have inserted ligand-binding domains into proteins such as fluorescent proteins, Cas9, aminoacyl-tRNA synthetases, and ferredoxins to regulate outputs, though predicting stable insertion sites remains challenging. These studies motivate using GDH as a robust scaffold for allosteric modulation, while underscoring the need for effective electrochemical signal transmission and integration into self-contained POC devices.

- Protein engineering: The ligand-binding domain (LBD) of human estrogen receptor alpha (ERα) was inserted into pyrroloquinoline quinone-dependent glucose dehydrogenase (PQQ-GDH) from Acinetobacter calcoaceticus to couple 4-HT binding to GDH redox activity. A comprehensive domain insertion library was generated using the SPINE algorithm to target all 454 amino acid positions of GDH. The library was assembled via Golden Gate cloning and verified by sequencing. - Library characterization and screening: Deep sequencing confirmed broad insertion coverage. The library was expressed in E. coli BL21, and 7392 colonies were screened using a whole-cell DCPIP reduction colorimetric assay (with PMS mediator) to identify variants capable of glucose oxidation in the presence of 4-HT. Active variants were further screened for differential activity with 4-HT versus DMSO to identify allosteric regulation. - Mapping permissive/allosteric sites: Identified insertion sites were mapped onto the GDH β-propeller structure and homodimer interface to analyze structural determinants of permissive and allosteric sites. - Variant selection and optimization: A top-performing variant, GDH-5E (LBD inserted after Thr5), was purified and characterized. To enhance allostery, a flexible linker (SGRPGSLS) was inserted after Lys104, yielding GDH-5E+, which showed increased baseline activity and stronger 4-HT repression. - In vitro enzymology: DCPIP-based assays quantified catalytic rates of wild-type GDH, GDH-5E, and GDH-5E+ with varying glucose and 4-HT concentrations; selectivity was tested with endocrine therapeutics. - Electrode interfacing: GDH-5E+ was immobilized in ferrocene-modified linear poly(ethylenimine) (Fc-LPEI) hydrogels to wire PQQ to electrodes while preserving protein conformation. Cyclic voltammetry and amperometry measured current responses to glucose with/without 4-HT and controls (e.g., 17β-estradiol), including in human whole blood. - Ratiometric electrochemical algorithm: Electrochemical kinetics of wt-GDH and GDH-5E+ were measured across glucose concentrations with/without 4-HT. A ratiometric metric i_GDH-5E+/i_GDH was developed to normalize out glucose concentration variability, implemented via a two-electrode setup (GDH-5E+ and wt-GDH working electrodes) to yield a binary Yes/No outcome for 4-HT. - Self-powered sensor: A glucose/O2 enzymatic fuel cell (EFC) was built using GDH-5E+/Fc-LPEI anode and a laccase (Trametes versicolor) cathode on anthracene-modified MWCNT/Nafion, separated by a Nafion 212 membrane. Polarization and power curves with human blood evaluated 4-HT-induced repression on current and power output. - Signal amplification via OECT: An organic electrochemical transistor (OECT) with PEDOT:PSS channel operated in depletion mode was connected in parallel with the EFC (cathode to gate; anode to source), with a source-measure unit imposing drain bias. Since 4-HT affects Vmax rather than OCP, a derivative-based analysis of drain current (dIsd/dt) linked to Vmax was used to resolve 4-HT effects, yielding amplified milliamp-range signals. - Human samples and statistics: Single-donor human whole blood was used under approved biosafety protocols. Experiments included biological triplicates or n≥3 replicates; unpaired two-tailed t-tests assessed significance. Detailed materials, device fabrication, and protocols are provided in Methods.

- Library screening and mapping: 232 (~51%) GDH insertion sites supported glucose oxidation in the presence of 4-HT; 71 (~16%) sites were 4-HT-regulatable. Allosteric sites preferentially located in flexible loops and at the homodimer interface, suggesting roles for secondary and quaternary structure in signal propagation. - Engineered variant performance: GDH-5E+ (with added linker) showed ~2× higher glucose oxidation than wt-GDH and exhibited 4-HT-dependent repression: 18% decrease in activity (112 ± 2 vs 92 ± 2 units; mean ± sd; p = 0.0005; n = 3), effective across 338 pM to 2 µM 4-HT (clinically relevant range). Selective regulation was observed for hexestrol, diethylstilbestrol, and lasofoxifene over 17β-estradiol. - Electrochemical sensing: GDH-5E+/Fc-LPEI retained 4-HT repression on electrodes. Amperometry showed significant inhibition with 4-HT vs. 17β-estradiol after ~4 min. In human whole blood (no added glucose), 4-HT reduced current by ~18% versus estradiol (3.2 ± 0.2 µA vs 3.9 ± 0.3 µA after background; p = 0.0475). - Ratiometric algorithm: i_GDH-5E+/i_GDH distinguished 4-HT presence independently of glucose concentration above 0.4 mM. In buffer: 6.37 ± 0.1 (blank, n = 8) vs 4.85 ± 0.1 (with 1 µM 4-HT, n = 9). In blood: 6.1 ± 0.5 (blank) vs 4.6 ± 0.4 (1 µM 4-HT); n = 3; p = 0.039, enabling Yes/No calls. - Self-powered sensor (EFC): Blood glucose/O2 EFC exhibited OCP 690 ± 22 mV. 4-HT reduced max current density from 179 ± 5 to 157 ± 6 µA cm⁻² and max power density from 32 ± 2 to 26 ± 1 µW cm⁻² versus estradiol controls. - OECT coupling and amplification: EFC-gated OECT produced milliamp-level drain currents. While OCP was insensitive to 4-HT, derivative analysis showed that 4-HT reduced dIsd/dt from 127 ± 16 to 53 ± 2 µA s⁻¹ (n = 3; p = 0.0038), achieving 58% modulation and robust amplification of the enzymatic signal.

The results validate the concept of a glucometer-based allosteric sensor (GBAS) where engineered protein allostery encodes the presence of a therapeutic into the glucose oxidation current. Mapping insertion permissibility and allosteric hotspots in GDH revealed structural principles enabling allosteric signal propagation, particularly via flexible loops and the dimer interface. The optimized GDH-5E+ variant provided selective, rapid 4-HT repression in both buffered systems and complex matrices like human blood. Critically, the ratiometric electrochemical algorithm i_GDH-5E+/i_GDH decoupled analyte detection from variable glucose levels, enabling reliable binary decisions across physiologically relevant glucose concentrations. Integrating the allosteric bioanode into a glucose/O2 enzymatic fuel cell created a self-powered sensor in which 4-HT decreased both current and power, and coupling this with an OECT yielded large, milliamp-scale signals with enhanced modulation via derivative-based analysis of drain current. Collectively, these advances demonstrate a practical pathway to POC therapeutic monitoring using existing glucometer paradigms, with robust electrochemical interfacing and electronic amplification addressing key challenges of sensitivity, selectivity, and miniaturization.

This work establishes a generalizable framework to convert glucometers into POC therapeutics sensors by engineering allostery into PQQ-GDH, decoding analyte-induced modulation via a ratiometric electrochemical algorithm, and integrating self-powered operation with OECT-based amplification. The comprehensive GDH insertion map informs future protein switch designs, while the GDH-5E+ variant demonstrates selective, rapid detection of 4-hydroxytamoxifen in buffered solutions and human blood. The self-powered glucose/O2 EFC and its coupling to an OECT enable significant signal amplification, achieving milliamp-level outputs and 58% modulation in derivative-based metrics. Future work could scale power via multiple EFC units or alternative wireless powering schemes and translate the GBAS into wearable or implantable formats (e.g., skin patches or ingestible/wireless devices) for real-time monitoring of endocrine therapeutics and other biomarkers.

A single enzymatic fuel cell unit does not provide sufficient power for conventional wireless communication (e.g., Bluetooth or RFID), limiting fully self-contained real-time operation. While multiple EFCs in series or alternative power transfer (e.g., magnetic body communication) can address this, such integration adds system complexity. Additionally, while the ratiometric algorithm mitigates glucose variability, broader clinical validation across diverse patient samples, interfering substances, and long-term stability/biocompatibility assessments will be required for translation.