A hybrid transistor with transcriptionally controlled computation and plasticity

Organic electrochemical transistors (OECTs) are ideal devices for translating biological signals into electrical readouts and have applications in bioelectronics, biosensing, and neuromorphic computing. Despite their potential, developing programmable and modular methods for living systems to interface with OECTs has proven challenging. Here we describe hybrid OECTs containing the model electroactive bacterium Shewanella oneidensis that enable the transduction of biological computations to electrical responses. Specifically, we fabricated planar p-type OECTs and demonstrated that channel de-doping is driven by extracellular electron transfer (EET) from S. oneidensis. Leveraging this mechanistic understanding and our ability to control EET flux via transcriptional regulation, we used plasmid-based Boolean logic gates to translate biological computation into current changes within the OECT. Finally, we demonstrated EET-driven changes to OECT synaptic plasticity. This work enables fundamental EET studies and OECT-based biosensing and biocomputing systems with genetically controllable and modular design elements.


Device Fabrication
Three versions of the OECTs were used: small channel OECTs, 2-electrode versions of the small channel OECT without the gate, and a large channel OECT.The small channel OECTs were used for the majority of experiments with the following exceptions.In direct channel reduction experiments (Figure 2h, Figure S3g), the 2-electrode versions of the small channel OECT were used and noted as 2-electrode devices or no gate.In UV-vis spectroscopy (Figure 2g, Figure S3c -S3e), the large channel OECTs were used to fit the laser aperture of the instrument.The large channel OECTs were labeled as 'large channel OECTs' and the small channel OECTs were noted as original three-terminal OECTs or without any specific naming.OECTs were fabricated according to prior work 1 .Quartz microscopic slides (FQ-S-003, AdValue Technology) were cleaned with soapy water, acetone, and isopropyl alcohol, and dried with nitrogen before oxygen reactive-ion etching (RIE, 150 W, 50 sccm, 120 s).Quartz slides were coated with a photoresist layer (AZ5209E) and lithographically patterned to define the electrode layout.Subsequently, Au electrodes (100 nm) with a Ti adhesion layer (10 nm) were thermally evaporated on the quartz slides, and excess materials were removed with acetone lift-off.Another photolithographic pattern was formed to define the PEDOT:PSS region over the channel and the tip of the gate to obtain a width/length of 150 μm x 10 μm and 500 μm x 500 μm, respectively.The PEDOT:PSS (Clevios TM PH1000) was filtered (0.22 μm PES filters) and spun cast on the pattered quartz slides, followed by hot plate drying at 90 °C for 15 min and acetone lift-off.To increase the conductivity, PEDOT:PSS films were immersed in ethylene glycol at 90 °C for 3 min over a hot plate.The asfabricated PEDOT:PSS film had an average thickness of 26.5 nm for the channel and 38.3 nm for the gate-tip layer.Polydimethylsiloxane (PDMS, Sylgard TM 184) with 9 % wt curing agent was drop cast and cured for 48 hours at room temperature to ensure surface smoothness.The OECT chambers and fluid access ports in the PDMS sheets were manually cut with a hole punch.The OECTs with the larger channel size were only used for UV-vis measurement.The electrodes were fabricated in the same way as the smaller OECT.The PEDOT:PSS channel was fabricated by drop-casing 80 μL PEDOT:PSS over the entire slide, followed by 20 minutes of air-drying and heating at 90°C for 20 minutes.Afterward, the excessive PEDOT:PSS film was removed with cotton swabs soaked with 70% ethanol.Conductivity enhancement was achieved by EG treatment of 10 min at 90 °C.Silicon spacers (0.5 mm, GBL664581, Sigma-Aldrich) were hand cut to create the OECT chambers and only used with the large channel OECTs.

Device Operation and Electrochemistry
Shewanella Basal Medium (SBM) amended with 0.05% casamino acids and 1X Wolfe's Trace Mineral Mix was used as the base electrolyte except for the carbon source comparison experiments (Figure 2e, Figure S2d) where the SBM was only amended with 1X Wolfe's Trace Mineral Mix.When noted, the SBM was further supplemented with 40 mM sodium fumarate to support cell growth.Before each experiment, the OECT slides and PDMS sheets were autoclaved separately and assembled in the biosafety cabinet.OECT experiments were conducted in a nitrogen-filled glovebox to create the anaerobic condition except for UV-vis spectroscopy which was conducted under ambient conditions.Media and solutions stocks were purged with argon for 15 min and stored in the glovebox.A multichannel potentiostat (MultiPalmSens4, PalmSens BV) was used for the electrochemical measurements.During continuous OECT operation, unless otherwise noted, the gate (VGS) and drain (VDS) voltages were biased at 0.2V and -0.05 V, respectively.Transfer curves were measured with a gate scan rate of 20 mV/s, except in the artificial synaptic hysteresis measurement where a rate of 10 mV/s was used.For all the experiments, OECTs were stabilized in the glovebox with abiotic electrolytes and constant bias voltages for 72 hours before inoculation or measurements.In the electrode potential measurements with Ag/AgCl pellet reference electrode (550010, A-M Systems), the Ag/AgCl electrodes were directly inserted into the OECT chamber without salt bridges.The same Ag/AgCl pellet electrodes were used as the gate for the large channel OECTs in abiotic UV-vis measurements.Hybrid OECT experiments with electroactive bacteria were conducted between 18 -36 hours after inoculation (initial OD600 at 0.01).Synaptic measurements were conducted after IDS was stabilized for at least 1 minute at VGS = 0 V.

Inoculation Procedure
For aerobic culture growth, SBM supplemented with 20 mM lactate was used for cell growth.Following aerobic overnight growth, the cultures were triple-washed by centrifugation using the same growth medium to remove byproducts.The optical density of the washed cultures was measured before transferring them into the glovebox.side the glovebox, the cell cultures were diluted to create an intermediate inoculum with a cell density 10-fold higher than the target OD for final inoculation.This dilution used the same electrolyte solution that filled the OECT.Finally, the OECTs were inoculated using a 1:9 volume ratio of the intermediate cell culture to the OECT electrolyte.For instance, S. oneidensis supplemented with 1 μM exogenous FMN with inoculum OD600 of 0.05 was prepared from aerobically grown cell cultures.The triple-washed cell cultures (average OD600 at 3.38) were brought into the glovebox and diluted to the intended OD600 of 0.5 with SBM supplemented with 1 μM exogenous FMN, creating the intermediate cultures.Then 5 μL of these intermediate cultures were inoculated into the OECTs containing 45 μL of SBM supplemented with 1 μM FMN, achieving a final inoculation OD600 at 0.05.For steady-state expression conditions, cells were anaerobically cultured in SBM supplemented with 20 mM lactate and 40 mM fumarate, along with specific inducer(s).To maintain consistent induction strength and to prevent cell growth from affecting OECT readouts, electrolyte in the OECTs were SBM supplemented with 20 mM lactate and specific inducer(s), but lacked fumarate.Prior to induction, strains were grown statically and anaerobically without inducer(s) for 6 hours.Subsequently, cell cultures were diluted 1:25 into media containing inducers (prepared from 1000x concentrated stocks) and incubated for an additional 18-24 hours.After this induction period, the cell cultures were directly inoculated into the OECTs at a 100-fold overall dilution without prior washing.Specifically, cells were first diluted 10-fold with the fumarate-free OECT electrolyte forming the intermediate cultures.The final inoculation of the OECTs was achieved by adding the intermediate cell culture to the OECT electrolyte at a 1:9 volume ratio.In the carbon source experiment, SBM amended with 1X Wolfe's Trace Mineral Mix was used as the base electrolyte.Aerobically grown S. oneidensis MR-1 cells culture were triple-washed with SBM without a carbon source, then the cell cultures were kept at room temperature for 3 hours to induce starvation conditions 2 .Afterward, cell cultures were washed again and their OD600 was measured.Subsequently, cell cultures were brought into the glovebox and diluted to obtain the intermediate stocks with an intended OD600 of 0.1.The dilutions were performed with SBM supplemented with either 20 mM lactate, 20 mM pyruvate, 20 mM acetate, or no carbon source.Finally, the intermediate stocks were inoculated into OECTs containing the SBM supplemented with the respective carbon source or no carbon source at a ratio of 1:9, achieving a final inoculation OD600 of 0.01.In cell viability experiments, aerobically grown S. oneidensis MR-1 cell cultures were triplewashed with SBM supplemented with 20 mM lactate.For E. coli cultures, the lactate was replaced with 20 mM glucose.The densities for the washed cell cultures were measured by OD600 and the S. oneidensis cultures were allocated to 3 parts: live, heat-killed, and lysed cells.Heat-killed cells were obtained by incubating at 80 °C for 20 min.Lysed cells were obtained by sonication (Qsonica 55, Qsonica LLC) for 90 s at 4 °C.The output power was set to 15 W with "ON" and "OFF" intervals of 10 s and 5 s, respectively.Then cell cultures were brought into the glovebox and diluted with SBM supplemented with 20 mM lactate to obtain the intermediate cultures at an intended OD600 of 0.1.The intermediate cultures were inoculated into the OECTs at a ratio of 1:9, achieving a final inoculation OD600 of 0.01.The S. oneidensis supernatants were obtained by filtering (0.22 μm PES filters) the cell cultures (initial OD600 at 0.01) anaerobically grown in the glovebox with SBM supplemented with 20 mM lactate.The entire volume of OECT electrolyte was replaced with the supernatant during inoculation.

Spectroscopy
The UV-Vis spectroscopy of the PEDOT:PSS channel was measured from 190 nm to 1100 nm (Agilent 8453 UV-Visible Spectroscopy System).A custom sample holder was 3D printed to fit the OECT slides to the instrument.Measurements were blanked with devices lacking the PEDOT:PSS channel.When bacteria cells were present in the sample, the blank devices were likewise inoculated with the same inoculum.

Fluorescence Microscopy
Microscopy was performed using a Nikon Ti2 Eclipse inverted epifluorescence microscope.Immediately after the 24-hour operation in the glovebox, OECTs were gently washed 2x by refreshing the electrolyte with the sterile SBM supplemented with 0.05% trace mineral supplement, 0.05% casamino acids.Then, the PDMS sheets were replaced with a 0.5 mm silicon spacer to ensure the sample thickness was compatible with the working distance of the microscope.Subsequently, the OECTs were gently washed with SBM containing 0.05% trace mineral supplement, 0.05% casamino acids, and LIVE/DEAD ® BacLight TM Stain mix (3 μL of SYTO 9 and propidium stocks per 1 mL) at a final solution volume of 10 μL per OECT chamber.The OECTs were then sealed with coverslips, covered with aluminum foil, and transferred out of the glovebox for microscope imaging.Bacterial counts were performed using ImageJ software.

Atomic Force Microscopy
The AFM scans were conducted using a DriveAFM (Nanosurf AG).The cantilevers (Dyn190Al) were driven with the photothermal laser (CleanDrive) under dynamic mode.OECTs were randomly selected from two fabrication batches: 3 slides of the as-fabricated OECTs (8 OECTs per slide) were used as pre-inoculation samples, and 3 cleaned OECT slides post the 48-hour inoculation served as post-inoculation samples.Two OECTs per slide were randomly chosen for AFM scans using Nanosurf CX software for data acquisition.Images for PEDOT:PSS film thickness were acquired at 90 μm x 90 μm and processed with Nanosurf CX software to correct background variations.Topology and phase images were acquired at 500 nm x 500 nm and analyzed without further process.Grain sizes were fitted by the Watershed method with the resulting histograms generated using Gwyddion software.

OECT Data Processing
Measured OECT data were processed using GraphPad Prism9 and MATLAB (R2021b update 1).The measured IDS data were normalized to the initial value before inoculation (IDS0) before fitting.The IDS decay rate constants for all samples except the lysed S. oneidensis were obtained by fitting IDS/IDS0 data with an exponential decay model: Where  is time in hours,  is the fitted IDS decay rate constant.Fitting was performed with the built-in one phase decay function in GraphPad Prism9.The IDS/IDS0 data for lysed S. oneidensis samples were fitted to a simple linear regression model: Where  is time in hours, the slope  is used as the rate of change.Fitting was performed with the built-in simple linear regression function in GraphPad Prism9.The response function of FMN concentrations was modeled with a four-parameter logistic regression function in terms of IDS decay rate constants (noted here as ):

Figure S3 .
Figure S3.OECT de-doping investigated using electrochemical and spectroscopy methods.(a) Transfer curves plotted using effective gate voltage (VG eff ) values with respect to Ag/AgCl pellet reference electrodes.(b) Gate (VG)and source (VS) potentials with respect to the Ag/AgCl pellet reference electrode plotted with the channel current IDS for an OECT inoculated with S. oneidensis MR-1.(c, left) Schematic and (c, right) photo-image of the large channel OECT used exclusively for the UV-Vis instrument.UV-Vis spectra were collected for (d) abiotic PEDOT:PSS channel under different Ag/AgCl pellet gate bias voltages, and (e) for S. oneidensis MR-1 inoculated channel overlayed with abiotic channels.(f) Cartoon illustrations comparing the

Figure S4 .
Figure S4.OECT responses to △mtrC strains carrying NAND and NOR Boolean gate plasmids.Cartoon illustrations of the plasmid architecture of the (a) NAND and (b) NOR Boolean gates expressing mtrC.The △mtrC mutants carrying the corresponding Boolean gate plasmids were brought to steady-state MtrC expression with combinations of 500 μM IPTG, 200 nM OC6, and 10 nM aTc inducers.Transfer curves of the induced (c) NAND and (d) NOR gates samples.The channel current IDS/IDS0 curves for (e) NAND and (f) NOR gates samples with different inducer combinations.Data in panels (e) and (f) show the mean ± SD of 3 biological replicates.Figure (a) and (b) created with BioRender.com.

Figure S5 .
Figure S5.Modulation of the OECT synaptic behavior with varying pulse conditions and strains.(a) A2/A1 index plotted with varying pules voltage VP, while pules duration tP and pules interval △t were fixed at 80 ms.(b) OECT channel conductance changes △G with varying pulse interval △t, fixed tP = 80 ms and VP = -0.5 V. A2/A1 index for OECTs with varying pulse interval △t, fixed tP = 80 ms and VP of (c) 0.5 V or (d) -0.5 V. (e) A2/A1 index of △mtrC strain carrying mtrC Buffer gate (+mtrC) or empty vector plasmid (+empty) under steady-state protein expression.A2/A1 index p values for pairs indicated from top to bottom p = 0.0046, p = 0.0045, and p = 0.0149.(f) Channel current IDS responding to the continuous voltage pulses with VP of 0.5 V or -0.5 V. Faded lines represent the raw IDS, while the bolded lines represent IDS baselines after filtering out the pulses.(g, h) One-phase exponential fitting of the IDS baselines for each continuous 4-pulse session, with VP equal to (g) 0.5 V or (h) -0.5 V. (i) The corresponding time constants of the fitting results.Time constant p values for the indicated pair p =1.0 × 10 -5 .In panel (a) to (e), data show the mean ± SD of 3 biological replicates.In panel (e) and (i), unpaired two-tailed Student's t-tests were performed without adjustments for multiple comparisons, n.s.represents p > 0.05.

Figure S6 .
Figure S6.Cartoon illustration of the major OECT channel fabrication and assembly steps.Created with BioRender.com.

Figure S7 .
Figure S7.Morphological characterization of PEDOT:PSS channel and gate-tip coating.(a) Thickness of PEDOT:PSS films derived from AFM scans.OECTs were inoculated with S. oneidensis and operated with constant bias voltages at the gate VGS = 0.2 V and drain VDS = -0.05V for 48 hours.(b) Histogram and Gaussian fit (lines) of PEDOT grain size in the channel region based on (e) and (f).Representative surface topologies of the PEDOT:PSS channel (c) prior to and (d) after the S. oneidensis incubation.Representative phase images of the PEDOT:PSS channel (e) prior to and (f) after the incubation, scale bars represent 50 nm.AFM measurements were performed on 6 independent samples.Panel (c) to (f) presents a zoomed-in morphology graph extracted from 2 representative samples.In panel (a), unpaired two-tailed Student's t-tests were performed without adjustments for multiple comparisons, n.s.represents p > 0.05.

Figure S8 .
Figure S8.OECT response to different extracellular electron transfer (EET) mechanisms.(a) The IDS/IDS0 curves of knockout strains with and without the addition of exogenous flavin mononucleotide (FMN) (1μM), and (b) S. oneidensis MR-1 cells with varying exogenous FMN concentrations.Initial inocula were adjusted to OD600 of 0.05.(c) The IDS/IDS0 curves of △mtrC, △Mtr, and MR-1 strains with initial inoculation OD600 at 0.1.(d) Cartoon illustration of the △mtrC strain and (d, insert) diagram of the mtrC Buffer gate controlled by the IPTG inducer.Faded shapes indicate removed proteins with genomic deletion and inhibition of electron transfer through the inner (IM) and outer (OM) membranes.Data show the mean ± SD of 3 biological replicates.Figure (d) created with BioRender.com.

Figure S9 .Figure S11 .
Figure S9.Channel current IDS/IDS0 curves of strains carrying the Boolean logic gates plasmids.(a) △mtrC strain carrying NAND Boolean mtrC plasmids.(b) △mtrC strain carrying NOR Boolean mtrC plasmids.MR-1 strain and △mtrC strain carrying empty vectors were used as positive and negative controls, respectively.Inducible mutants were brought to steady-state expression under various inducer combinations and concentrations before inoculation.Data show the mean ± SD of 3 biological replicates.

Figure S12 .
Figure S12.Paired-pulse responses of abiotic OECTs at varying electrochemical doping states.The source electrode potential (VS) was constantly biased at the specified voltages against the Ag/AgCl pseudo-reference electrode.(a) Channel current IDS of OECT with VS biased from 0.2 V to -0.7 V in increment of 0.1 V. (b) Normalized IDS for VS biased at -0.5 V and -0.6 V.

Figure S13 .
Figure S13.Electrochemical response of the hybrid OECTs under varying oxygen conditions.(a) Channel current IDS were normalized to the pre-inoculation aerobic baseline values, with gate voltage VGS = 0 V and drain voltage VDS = -0.05V. Paired pulse responses of pulse voltage VP = 0.5 V and drain voltage VDS = -0.05Vwere shown with the measured source electrode potentials against Ag/AgCl pseudo-reference electrode during (b) aerobic pre-inoculation condition, (c) anaerobic stabilization post-inoculation, and (d) subsequent re-oxygenation to ambient conditions.Data in panel (a) show the mean ± SD of 3 biological replicates.

Table S1 .
Strains and plasmids used in this study.