A hybrid transistor with transcriptionally controlled computation and plasticity

Devices capable of translating biological and chemical activity into electrical signals are crucial for various applications, including biosensing, neuromorphic computing, cellular computing, and wearable electronics. Organic electrochemical transistors (OECTs) have emerged as ideal candidates due to their use of aqueous electrolytes, biocompatibility, and low operating voltages. Unlike conventional transistors, OECTs utilize ions within an electrolyte to modulate the doping state and conductivity of an organic mixed ionic-electronic conducting channel. This allows for exceptional transconductance and sensitivity. OECTs' ability to couple ionic and electronic transport makes them attractive for merging biological and traditional computation. While gate voltage typically controls ionic diffusion, redox reactions in the electrolyte can also alter channel doping, acting as a secondary gate. Examples include lipid bilayer functionalization, gated ion channels, and redox-active enzymes. However, these approaches are limited in computational complexity. Living cells, particularly engineered bacteria, can perform complex computations including Boolean logic, signal processing, and neuromorphic functions. While OECTs can detect bacteria or metabolites, directly coupling advanced bacterial computations to an electrical output through genetic circuits remains a challenge. Electroactive bacteria, capable of extracellular electron transfer (EET), offer a promising solution. EET, the process of transferring electrons across the cellular membrane, can be genetically controlled, providing a potential universal interface between bacteria and electronic devices. Compared to traditional bioelectrochemical cells, OECTs offer advantages like faster response times, controlled environments, signal amplification, and selective detection. Previous work demonstrated EET detection using OECTs, but a more comprehensive mechanistic understanding and genetic control are needed to enhance performance and enable complex biological computation.

The literature extensively documents the use of OECTs for various applications, including sensing and flexible electronics. Several studies highlight their potential in neuromorphic computing due to their ability to mimic synaptic weight changes via ion transport control. Researchers have explored using redox-active enzymes and biomolecules to interface with OECTs for chemical and biological sensing. However, the complexity of these sensing systems is limited. The use of electroactive bacteria and their EET capabilities for bioelectronics applications, particularly in microbial fuel cells, has also been explored. Recent advancements in synthetic biology enable tight genetic control over EET flux in model electroactive bacteria like *Shewanella oneidensis* and *Geobacter sulfurreducens*. This control is achieved through genetic circuits that regulate the expression of EET-relevant genes in response to various stimuli. This control provides a path to integrate bacterial computation with electronic devices. Existing research shows proof-of-concept for real-time cellular EET activity monitoring with p-type OECTs using *S. oneidensis*, demonstrating the feasibility of this approach. However, the current study aims to improve upon this by developing a deeper understanding of bacteria-OECT interactions, integrating genetic regulation, and exploring applications in biological computation and synaptic plasticity.

Planar p-type poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) OECTs were fabricated on quartz microscope slides with Ti/Au electrodes. *Shewanella oneidensis* MR-1 was used as the model electroactive bacterium. Experiments assessed cell growth and viability within the OECT electrolyte under anaerobic conditions, with and without fumarate as an electron acceptor. Cell distribution was visualized using fluorescence microscopy. Colony-forming units (CFUs) and optical density (OD600) were measured to quantify cell growth. Electrochemical measurements were performed by monitoring the drain-source current (*IDS*) and gate voltage (*VGS*) under various conditions. UV-Vis spectroscopy was used to characterize the doping state of the PEDOT:PSS channel. Genetically engineered *S. oneidensis* strains, including Δbfe (impaired flavin secretion), Δlysis (impaired biofilm formation), ΔmtrC (deficient in outer membrane EET proteins), and Δmtr (deficient in multiple EET proteins), were used to investigate the mechanisms of EET-driven channel de-doping. Plasmid vectors encoding key EET genes (*mtrC* and *mtrCAB*) were used to complement EET-deficient mutants, allowing for controlled expression of EET proteins through the addition of IPTG or OC6. Boolean logic gates (NAND and NOR) controlling *mtrC* expression in response to IPTG, OC6, and aTc were constructed and their effect on OECT current was measured. For synaptic plasticity studies, paired-pulse experiments were performed, applying pulsed gate voltages to monitor short-term plasticity (STP) as characterized by the paired-pulse facilitation (PPF) and paired-pulse depression (PPD) indices. Long-term plasticity (LTP) was assessed by continuous presynaptic stimuli. Statistical analysis was performed using unpaired two-tailed Student's t-tests and general linear hypothesis tests.

The study demonstrated that *S. oneidensis* MR-1 can de-dope the PEDOT:PSS channel in OECTs through EET, reducing channel conductance. This de-doping was confirmed by UV-Vis spectroscopy, showing spectral changes consistent with de-doping in abiotic devices. The rate of current decay was proportional to the number of cells and their metabolic activity, with lactate being the preferred carbon source. Experiments using EET-deficient strains showed that direct EET via the Mtr pathway is the primary mechanism for channel de-doping, although exogenous flavins could enhance the process. Genetically controlled EET using inducible promoters successfully translated Boolean logic gate outputs (NAND and NOR) into electrical signals (changes in *IDS*). OECTs containing *S. oneidensis* displayed distinct synaptic behaviors, exhibiting hysteresis in transfer curves, indicative of short-term memory. Paired-pulse experiments revealed asymmetric responses to positive and negative gate pulses, with positive pulses inducing significant conductance changes and a clear dependence on pulse interval. Continuous presynaptic stimuli demonstrated consistent and reversible synaptic modulation, with little evidence of LTP. Experiments with EET-deficient strains and complemented strains confirmed the link between EET and synaptic modulation. Finally, the integration of Boolean logic gates further demonstrated computational control over synaptic function.

This research successfully demonstrates the integration of living electroactive bacteria with OECTs to perform computation and mimic synaptic plasticity. The findings highlight the feasibility of using EET as a mechanism to translate genetically encoded biological computation into electrical readouts. The ability to control EET flux via transcriptional regulation allows for the design of modular and programmable biosensing and biocomputing systems. The observed synaptic behavior opens up possibilities for developing neuromorphic devices with bio-inspired features. The asymmetric response to positive and negative gate pulses and the spike-recovery behavior in the presence of *S. oneidensis* suggest an interesting interaction mechanism, which warrants further investigation to fully understand the underlying processes. The use of Boolean logic gates showcases the potential of the hybrid OECTs for complex computations. Future work could focus on optimizing the design and exploring various genetic circuits and regulatory elements. Overall, this study provides a significant advance in bioelectronics and opens doors for exploring sophisticated bio-inspired computational devices.

This study successfully integrated living electroactive bacteria with OECTs, demonstrating the transduction of bacterial computation into electrical signals. The system leveraged extracellular electron transfer (EET) controlled by genetically encoded Boolean logic gates, converting transcriptional logic into electrical outputs. Furthermore, the study demonstrated EET-driven synaptic plasticity in the hybrid transistors. Future research should focus on optimizing device designs, exploring diverse genetic circuits, and investigating the underlying mechanisms of the observed synaptic behavior. This work paves the way for advanced biosensing and biocomputing applications.

While the study provides strong evidence for the role of EET in OECT de-doping and synaptic modulation, the exact mechanisms are not fully elucidated. The study primarily focused on *S. oneidensis* MR-1; further research is needed to explore the generality of these findings across other electroactive bacteria. The influence of environmental factors beyond the tested inducers on EET and OECT performance requires further investigation. The long-term stability and reliability of the hybrid OECTs need further evaluation to assess their potential for practical applications.