Introduction

Various strategies have been used to realize cytokine (inflammation biomarker) biosensors including optical, impedimetric, and electrochemical means1. Integration of a biorecognition element (e.g., antibody, aptamer) with an electrode is a fundamental aspect of an impedimetric and electrochemical biosensor. A class of biosensors that has gained increasing attention over the last years is based on transistors, namely electrolyte gated organic transistors (EGOTs)2. Their use for sensing of inflammation biomarkers has been subject to reviews3. Inflammation is caused by pathogens4,5 (e.g., the pleiotropic cytokine interleukin 6, IL6) and tissue damage, among others6,7. Inflammation has also been linked to neurodegenerative diseases8,9 and coronavirus disease (COVID-19)5,7,10. IL6 in particular has been associated with inflammation11,12 as well as multiple sclerosis13,14,15 and COVID-1916,17,18,19. IL6, with a molecular weight of 19–26 kDa, has low levels (around 0.2–7.8 pg/ml, or approximately 10–300 fM) in human serum20. During inflammation these levels can increase a thousandfold11,21. Whilst IL6 detection with EGOTs has been demonstrated22,23, enzyme-linked immunosorbent assays (ELISA) and western blot remain the state-of-art quantification techniques24. These state-of-the-art methods rely on laborious protocols and bulky equipment compared to EGOTs, which limits their applicability for point-of-care or home application25. Electrochemical sensors26,27, and especially field effect transistors28,29, have gained interest due to their potential for rapid detection in miniaturized platforms. They facilitate a high signal-to-noise ratio while operating at low voltages, benefit from easy fabrication, and can be operated in aqueous environments30,31.

Organic electrochemical transistors31 (OECTs) are a special form of EGOTs, where ions of the electrolyte can penetrate into an organic mixed ionic–electronic conductor (OMIEC) transistor channel32. Poly(3,4-ethylenedioxythiophene) doped with polystyrene sulfonate (PEDOT:PSS) is by far the most commonly used OMIEC in OECTs. Since the ions can penetrate the OMIEC film, OECTs exhibit a volumetric transconductance increase and performance does not scale with the OMIEC/electrolyte interface area (channel)31,33,34. The comparably high transconductance31 compared to organic field-effect transistors, combined with easy integration into circuits as well as stable operation in aqueous environments, make them ideal candidates for biosensors. Indeed, various metabolites35,36,37,38,39 and neurotransmitters40,41 have been successfully measured with OECTs.

However, to our knowledge only a limited number of OECT-based cytokine sensors have been reported22,42. Gentili et al. investigated IL6 detection, but their strategy used a gold gate for immobilization of antibodies and an immuno-affinity regenerated cellulose membrane to increase the effectively measured IL6 concentration by an order of magnitude22. Diacci et al. used gold nanoparticles for immobilization of aptamers through electrochemical breaking of disulfide bonds allowing for simultaneous deposition of the aptamer and a blocking agent29. A strategy utilizing a PEDOT:PSS gate was proposed by Decataldo et al. combining (3-aminopropyl) triethoxysilane (APTES), biotin, streptavidin, and antibodies for detection of bone morphogenic protein 2 (BMP-2). Their study was a proof-of-concept as just one concentration of BMP-2 was measured42.

Diazonium salts have been studied in various forms for biosensors43,44 via electrochemical and spontaneous grafting, e.g., for C-reactive protein45 or tumor necrosis factor46. We propose to use diazonium salts to bind aptamers chemically onto PEDOT:PSS as shown in Fig. 1a. The diazonium 4-aminobenzoic acid (4-ABA) was attached to PSS through spontaneous diazonium chemistry43,47,48,49. This allowed for (1-ethyl-3-(3-dimethylaminopropyl)carbodiimide)/N-hydroxysuccinimide (EDC/NHS) activation of the carboxylic group and coupling with an amine-modified DNA aptamer. The overall idea and binding mechanism of aptamers to PSS through diazonium chemistry is depicted in Fig. 1a.

Fig. 1: Basic methodology and introduction to OECT
figure 1

a 4-ABA (spontaneous after HCl reduction) and aptamer functionalization. Dotted line indicates location of functionalization on the gate, b basic OECT diagram.

Full size image

Figure 1b shows the schematic circuit of an OECT, where the gate and channel are ionically connected through an electrolyte (depicted in yellow). OECTs are three terminal devices in which a voltage at the gate (VGS) modulates the current in the semiconductor channel (IDS), which is driven by a voltage between drain and source contacts (VDS). A small voltage change at the gate (such as upon an analyte binding event) will be amplified by the transistor and can be measured by the change of IDS. This can be summarized by a figure of merit, the transconductance, defined as ∂IDS/∂VGS. For biosensors, the change in effective voltage drop at the electrode/electrolyte interface is determined by the recognition element and analyte31,50. Both gate/electrolyte and electrolyte/channel interfaces can be used to attach the recognition element. When functionalizing the interface, the distance of the binding event to the electrode is important because of the Debye-screening length, even though sensing beyond the screening length has been observed51,52. Nevertheless, small linker and recognition elements are advantageous; hence, our choice of 4-ABA and synthetic oligonucleotide aptamers, which exhibit high specificity, low production costs, good storability/shelf-life, and can be modified for various immobilization strategies53.

Results and discussion

Screen printed OECTs have previously been demonstrated for the detection of uric acid37. For the experiments detailed below, the same architecture employing PEDOT:PSS as channel and gate material was used and an image of the device is shown in Fig. S1.

Functionalization and transfer curves

Activation of 4-ABA was carried out with NaNO2 in HCl, reducing and binding it (spontaneously and not electrochemically) to the surface. For each step of the functionalization strategy, transfer curves and electrical impedance spectroscopy (EIS) measurements were taken. The transfer curves show a loss of drain current modulation with 4-ABA (Fig. 2a, solid gray) and then aptamer (solid purple) compared to the pristine OECT before functionalization (solid blue). Furthermore, a prominent hysteresis is observable. To see whether the loss of modulation and increasing hysteresis were gate effects or possibly channel degradation, the “effective” voltage in the electrolyte (voltage drop at the electrolyte/channel interface) was recorded during the transfer curves with an additional Ag/AgCl electrode. These transfer curves against this effective voltage are shown in Fig. 2a (dashed, same color-code). While a shift of the drain current is still visible (less current at given voltage), which is often attributed to channel degradation, it shows that the main effect of modulation difference stems from different potentials in the electrolyte at the different stages of functionalization. Therefore, we can conclude that the voltage drop at the electrolyte/channel interface is altered during our functionalization and is a significant driving force behind the hysteresis behavior; in all three cases, linear behavior without significant hysteresis is observed vs the reference electrode compared to the transfer curves vs applied gate voltage, leading to the conclusion that these changes overwhelmingly originate from the gate and not the channel degradation. Modification of the gate electrode was also investigated with EIS measurements, as well as associate control experiments and comparison of different print batches, are shown in Fig. S2. These show an impedance shift with subsequent functionalization steps in accordance with Fig. 2a.

Fig. 2: OECT recordings.
figure 2

a Transfer curves during functionalization (blue: PEDOT:PSS; gray: 4-ABA; purple: 4-ABA+aptamer). b Dose curve of IL6 (black, n = 6) and BSA (brown, n = 3) in 1× PBS by constant OECT measurement. Linear fit for IL6 with slope (−9.5 × 10–4 ± 4 × 10–5) and offset (−4.3 × 10–2 ± 5 × 10–3). Control experiments (n = 1) without aptamer are indicated in gray. c Normalized response of constant measurement in bovine plasma after accounting for drift. Green areas indicate when bovine plasma was spiked with IL6 was measured.

Full size image

Constant measurements in PBS

Constant measurement (IDS vs time) was chosen as the measurement strategy and a protocol for stabilization the planar structure in 1× PBS was established. Before measuring IL6 concentrations, 1× PBS solution was changed multiple times to obtain a good estimate of device behavior and acquire a suitable fit for background subtraction (Eq. 1). Such a constant measurement can be seen in Fig. S4. Changing the electrolyte resulted in short transient spikes before continuing along the same trend (small alteration can occur after first change due to evaporation). With increasing concentration of IL6 in PBS from 1 pM to 1 nM the signal could be detected as drain current increases. Afterward, the electrolyte was changed again to PBS resulting in the drain current decrease, nearly back to the initial signal.

Bovine serum albumin (BSA) was chosen as control to elucidate how the sensor reacts to other proteins that are not specific to the aptamer. Furthermore, in a previous study the same aptamer sequence compared the detection of two cytokines to each other, namely IL6 and Tumor Necrosis Factor (TNF)29, and distinguishing the two. BSA is known to be the main interfering agent in blood (and thus plasma samples) and is often used as a blocking agent23,54. BSA was observed to cause only a small change in drain current, much smaller than elicited by IL6 (Fig. S4). For a better understanding of the effect of BSA (i.e., non-specific binding), the experiment was repeated with varying concentrations of BSA (1 pM to 1 nM) followed by measurement of 1 nM of IL6, indicating robust sensing capability even after non-specific binding (Fig. S5).

The dose curves of multiple samples with IL6 and BSA can be seen in Fig. 2b. A clear distinction between the two is possible even for concentrations down to 100 pM. While this does not allow measurements of physiological conditions, it is more than sufficient for detection in the case of inflammation. Furthermore, linear behavior can be observed in the range 10–1000 pM IL6, as shown with the linear fit. The dose curve for BSA does not show a significant change in this concentration range. A limit of detection (LOD) of 100 pM was calculated with linear regression55. Another control, without the aptamer, was performed to sense IL6 and can be seen to closely mimic the BSA results. Experiments were also carried out in bovine plasma, after incubation of the OECT overnight at 4 °C. The resulting constant measurement after accounting for drift (Fig. 2c, without drift Fig. S6) showed a similar behavior as a device in PBS, although the measurement threshold seems to be higher ( ≈ 1 nM of IL6 in spiked bovine plasma). Changing solution from spiked back to clean bovine plasma also returning to the same trend as before the spiking.

The constant current results are supported by impedance measurements at the OECT gate (Fig. S7). After measuring in PBS, concentrations of IL6 (100, 500, and 1000 pM) were measured before measuring PBS again. The curve for 100 pM IL6 is indistinguishable from the initial PBS signal and only higher concentrations elicited an impedance shift. The ensuing measurement with PBS resulted in a curve close to the initial PBS signal. This is in good agreement with the previously observed behavior in Fig. 2c, Figs. S4 and S5, where it was also observed that in switching from electrolyte to electrolyte-plus-IL6 to just the electrolyte, the signal nearly returned to its initial state. This suggests that only a small amount of the analyte is permanently bound and that the Kon (association reaction) and Koff (dissociation reaction) governing the binding reaction are similar, which is an unexpected result. In addition, measurements with added BSA (500 pM and 1 nM) overlapped with the PBS curve, showing that such impedance measurements could also distinguish BSA and IL6 (Fig. S7). A second device demonstrated the similar behavior (not shown).

Only a superficial comparison between EIS measurements and OECT in constant mode has been drawn, but it indicates a lower OECT detection level. Because of the observed behavior of Kon and Koff, the sensor could be measured multiple times, which was rather surprising. A linear behavior is retained but the lower concentrations do not show linear behavior in the third measurement performed 20 days after functionalization (Fig. S8).

Improving PEDOT:PSS functionalization

To further understand the functionalization, we performed time-of-flight secondary ion mass spectroscopy (ToF-SIMS, Fig. S9) and x-ray photoelectron spectroscopy (XPS, Fig. S10) of the screen-printed PEDOT:PSS surface. The ToF-SIMS data showed the presence of Si- and SiC3H9+ fragments (typical for Polydimethylsiloxane and its derivatives, likely from the encapsulation layer or from the PEDOT:PSS ink formulation itself e.g., crosslinker) on the surface and that detection of C8H7SO3- (PSS) was only possible when sputtering the surface layer away (squares in Fig. S9). The unexpected similarity of Kon and Koff, the ToF-SIMS data and unusual S2p peak of this PEDOT:PSS formulation observed in XPS, and inconsistent results of fluorescence experiments (not shown), led us to deeper investigation of the PEDOT:PSS and in particular the PSS content (where 4-ABA is intended to bind, Fig. 3). Approaches included changing the PEDOT:PSS component as they have different PEDOT to PSS ratios (PH1000 ratio PEDOT to PSS of 1:2.5, AL4083 ratio of 1:6, Fig. S11) or reducing the crosslinker56.

Fig. 3: Effect of PSS content on functionalization strategy.
figure 3

a AL4083 with 0.25% GOPS plus 4-ABA-binding fluorescent probe. b AL4083 with 1% GOPS and added 1% PSS plus 4-ABA-binding fluorescent probe (PLL-FITC). c, d Same formulations as a and b, respectively, without fluorescent probe, showing auto-fluorescence. Scale bar 1 mm.

Full size image

In Fig. 3, a low amount of crosslinker (0.25% 3-glycidyloxypropyl)trimethoxysilane (GOPS), Fig. 3a) is compared to starting with a higher PSS fraction (Fig. 3b). The higher fluorescence (after attaching a fluorescent probe which binds to 4-ABA/PSS) is evident, but auto-fluorescence also increases (Fig. 3c, d). The proposed strategy, based on AL4083 with added PSS, has been tested against high salt concentration (1 M NaCl) and alkaline solution (1 M ethanolamine, ETA) to show its stability with attaching PLL-FITC (Fig. S12) and aptamer with a fluorescent 6FAM-tag (Fig. S13).

This strategy can also be used to attach a redox probe (e.g., amino ferrocene) instead of a fluorescent moiety (Fig. S14). As a side note, using electrochemical strategies for diazonium based immobilization on the screen-printed sample, like the method proposed by ref. 49 could have also worked (Fig. S15). Comparing the fluorescent regions from Fig. S15 to the ToF-SIMS results (Fig. S9) shows that in this case even PSS poor regions—and thus relatively PEDOT rich and more conductive regions—of the film are functionalized (fluorescent).

Frequency dependent OECT measurements

The proposed sensing strategy is based on capacitive changes at the gate/electrolyte interface. Since during functionalization, the impedance increases (Fig. S2) and only small capacitive changes are expected, a different strategy than constant measurement (where VGS and VDS are constant) was adopted. We turned to a strategy based on frequency-dependent OECT measurements proposed previously57,58 for cell impedance studies. Through this approach we hoped to overcome limitations concerning detection of small capacitive changes because of the high initial capacitance at the gate compared to the sensing event and estimated low conductivity of the proposed PEDOT:PSS formulation.

The setup used by ref. 57 is shown in Fig. 4a and the circuit for our approach in Fig. 4b. In general, the system can be separated into low-pass and high-pass components57. The gate/electrolyte represents the high-pass filter, which is related to gate properties (e.g., capacitance) and the channel represents the low-pass filter which is dominated by slower processes (i.e., ion mobility). As demonstrated above (Fig. 2a) measuring voltage inside the electrolyte can indeed deliver information about the system (and more specifically the gate). Others used an additional reference electrode in an OECT setup to control the voltage at the gate/electrolyte interface59, with great success compared to classical cyclic voltammetry or square wave voltammetry (SWV). OECTs with SWV have shown superior to 3-electrode SWV, especially when reducing the gate electrode area60. At the same time as AC voltage is applied between gate/source, both interfaces (gate/electrolyte and electrolyte/channel) will affect the potential in the electrolyte. The system was tested against various transistor configurations. VCh was introduced to differentiate between applied (source-measure unit) and measured voltage (data acquisition module, DAQ) at the transistor channel, which is needed if an additional series resistance is used for measuring IDS. Our configuration substitutes the digital multimeters used by Rivnay et al. and utilizes the DAQ for measuring as it was already in place for applying the sinusoidal voltage at the gate electrode. The additional voltage measurement was also observed to aid in the case that the channel area was larger than the gate area (Fig. S16). In case of small (or zero) VDS, it is similar to an EIS spectrum with the gate as the working electrode and the channel as the counter electrode. It is noteworthy that the placement of the reference electrode should be kept constant for the duration of the experiment as its position relative to the gate can affect the measured VR (Fig. S17 for 10 kHz). However, we observed that sensitivity to reference electrode position was also dependent on the whole setup, as a smaller OECT channel seemed less affected by the placement of the electrode (Fig. S18).

Fig. 4: Frequency-based OECT measurement.
figure 4

a Schematic circuit used by ref. 57 b Schematic circuit for our frequency dependent OECT measurement. c Measurement of IL6 with interdigitated channel and functionalized gate. Black and gray curves show measurements in PBS and after incubation for 15 min in PBS respectively. Blue and purple curves show incubation for 15 min in various IL6 concentrations. d Nyquist representation (real and imaginary parts of VR for the same PBS and IL6 conditions as part c.

Full size image

For measuring IL6 we chose to use an interdigitated PEDOT:PSS channel (PH1000 formulation) with a larger area than the aptamer- functionalized gate (AL4083, 1% GOPS & 1% PSS). The results can be seen in Fig. 4c where the change of VGS to VR is plotted as “gain” (defined as ∂VR/∂VGS, Eq. 2) as a function of frequency. Having a small gate area can be beneficial by not only using small quantities of the recognition element but also for sensing low quantities61, as the relative change upon analyte binding is larger. Incubating for 15 min in PBS resulted in an increase of the measured voltage (sinewave amplitude). Consequent incubation (15 min) with increasing concentration IL6 (100 pM, 1 nM, 10 nM, and 100 nM) resulted in decreasing signal. The signal decrease was most pronounced at frequencies above 1 kHz. The results also indicate gain saturation for IL6 ≳1 nM. Similar behavior can be observed in the Nyquist plot (imaginary vs. real part) of the measured voltage (Fig. 4d) derived from Eqs. 3 and 4. Measurements in PBS after each incubation step were also carried out and similar trends were observed with increasing IL6 concentration (Figs. S20 and S21). This contrasts with the measurements using screen-printed devices where the IL6 measurement required IL6 in solution, and the signal was not retained through subsequent PBS rinse (e.g., EIS data Fig. S7). This “retainment” of the signal is expected from bound analyte rather than electrostatic interaction of analyte in solution. SPICE simulation with a simplified circuit in PySPICE indicates that the behavior of the device upon IL6 binding can be related to capacitive changes at the gate (Fig. S22). To tentatively compare the two approaches, we compared standard “constant” OECT measurement of the screen-printed OECTs to the normalized response at 10 kHz using the frequency method described above. The results (Fig. 5, log/log plot and Figure SI 21 as lin/log plot) show a significantly stronger response: approximately 100× at 100 pM and over 10× at 1 nM.

Fig. 5
figure 5

Comparison of constant voltage and frequency dependent (10 kHz) measurement of IL6 with screen-printed OECTs.

Full size image

A diazonium-based functionalization strategy for PEDOT:PSS was explored to attach aptamers to screen printable and other PEDOT:PSS formulations. Subsequent deposition of 4-ABA and ensuing aptamer immobilization were examined. Furthermore, the effect of PSS content and the composition of the blend was investigated. While IL6 concentrations could be measured with screen printed devices, even in bovine plasma, the underlying detection mechanism is not yet completely understood; it may arise from electrostatic interactions and was only observable with the screen printed PEDOT:PSS OECTs.

Other PEDOT:PSS compositions were investigated regarding their binding of 4-ABA and the PSS content was identified as a key parameter for the proposed binding mechanism. This was observed with fluorescent and electrochemical tags attached to the surface.

Finally, following similar approaches combining OECTs with electrochemical spectroscopy, a frequency-dependent measurement strategy utilizing an additional electrode provided IL6 quantification measurements reminiscent of EIS. With this method, distinguishing down to 100 pM of IL6 in PBS was possible, and exhibited a 100× higher normalized response than the screen-printed “standard” approach. Given the low- and high-pass components of the frequency-dependent measurement, in combination with the additional voltage measurement, future work using this platform could realize a passive band-pass/stop filter with a narrow frequency selection. This would depend on the system properties (e.g., capacitance, resistance) and more information like bandwidth and lower- and upper-cut-off frequency could be extracted that give further information on the measurement and the underlying sensing efficiency/kinetics. If the measured frequency range can be reduced, the measurement time could be even further shortened (currently ~20 s). While this allows for higher throughput and better cost effectiveness, the main advantage compared to ELISA remains the potential for point-of-care devices.

Methods

Chemicals were obtained from Sigma Aldrich unless otherwise stated.

OECT screen-printing

OECTs were screen printed on Hostaphan, a polyethylene terephthalate (PET) substrate (125 µm thick), using a DEK Horizon 03iX screen printing machine. The PET substrates were preheated at 120 °C for 5–10 min before screen printing the first layer. OECT fabrication included five sequentially screen printed layers. Step 1: screen printing silver-based paste (Ag 5000, DuPont) to create interconnects and contact pads. Steps 2 and 4: screen printing PEDOT:PSS-based channels and gate electrodes (Clevios S V3 paste, Hereaus), both 4 mm2 area (width = 2 mm; length = 2 mm). Step 3: screen printing carbon-based source and drain electrodes (carbon paste 7102, DuPont). Step 5: screen printing insulating material (encapsulation layer, 5018, DuPont), to ensure exposure of only channels and gate electrodes. Layers 1–4 were thermally cured using a conveyor belt (120 °C), while layer 5 was exposed to ultraviolet light (650–750 mJ/cm2).

PEDOT:PSS formulation and electrodes

We used two PEDOT:PSS formulations, Clevios PH1000 and Clevios AL4083. Our PH1000 solutions, unless otherwise stated, consisted of 5% v/v ethylene glycol (EG),1% v/v (3-glycidyloxypropyl)trimethoxysilane (GOPS) and dodecylbenzenesulfonic acid (1 drop per 5 ml). Solutions were spin-coated (1000 rpm for 30 s) on glass slides after cleaning with plasma at 50 W for 2 min.

AL4083 in various compositions was spin-coated and/or drop coated onto Au/Cr substrates or glass substrates depending on use. Samples for OECT measurements were spin-coated at 2500 rpm for 30 s and cured at 120 °C for 1 h on an electrode consisting of Au (50 nm)/Cr (5 nm) layers or on glass alone (for fluorescent images). Photolithography and Parylene C double layer coating was used for patterning of electrodes as previously described62,63. The geometry of the gate was 600 × 600 µm. Samples for fluorescent images were drop cast on glass slides and cured. For all cases the surfaces were plasma cleaned beforehand for 2 min at 50 W.

For frequency dependent measurements, an OECT channel on interdigitated electrode (ED-IDE1-Au, MicruX Technologies) spin coated with our PH1000 formulation. The spacing of the interdigitated electrodes was 10 µm, the electrode width was 10 µm, and overall diameter of the active/channel area 3.5 mm.

Device functionalization and characterization

4-ABA (1 mM in 0.5 M HCl) activation with sodium nitrite (5 mM in deionized water, DI) was performed in ice (0 °C) and then deposited on the PEDOT:PSS gate for 20 min at room temperature. After washing and curing at 90 °C for 30 min, it was treated with EDC/NHS (4:1 in 100 mM (2-(N-morpholino)ethanesulfonic acid), MES buffer, pH 5) for 30 min (room temperature) followed by aptamer (5 µM in 1× PBS) immobilization overnight at 4 °C. As a final step for the screen printed OECTs, free carboxylic groups were deactivated by ethanolamine (ETA, 1 M for 30 min) followed by immersing in 1× PBS for 15 min three times, to remove unbound entities and immediately measured or stored in dry condition if necessary.

The IL6-aptamer sequence(5′-3′) CTT CCA ACG CTC GTA TTG TCA GTC TTT AGT [AmC3] was resuspended in PBS pH 7.4. For fluorescent images with aptamers, the adapted sequence was [6FAM] CTT CCA ACG CTC GTA TTG TCA GTC TTT AGT [AmC3].

Transfer and output curves, for determination of the functionalization, were obtained using a Keithley 2612 (transfer: VDS = −0.2 V, VGS from 0 V to 0.8 V with rate 0.02 Vs−1) in 1× PBS with a custom LabView program.

EIS measurements for characterizing functionalization steps were performed after three cycles of CV (for stabilization from −0.5 V to 0.5 V) in [Fe(CN)6]3− in a three-electrode setup (Pt as counter and Ag/AgCl pellet as reference). Impedance measurements for IL6 detection were carried out in 1× PBS. Both were done with a Biologic SP 200 potentiostat.

Constant measurements

Constant measurements for screen printed electrodes were performed at VDS = −600 mV and VGS = 50 mV and the drop of electrolyte changed regularly after an equilibration time of about 2 h (voltages for measurements taken from Fig. S3). The time of equilibration before measuring IL6 was used to fit an exponential + linear trend (IFit) for background subtraction. For fitting, the following formula was used:

$${I}_{{Fit}}\left(t\right)={I}_{0}+{c}_{E}\cdot \exp (-t/{t}_{0})+{c}_{L}\cdot t$$
(1)

With I0, t0 being the current and time offset respectively, cE and cL constants of the exponential and linear parts respectively. For the dose curve the IDS – IFit was taken 480 s after changing electrolyte.

Sample with bovine plasma was incubated with bovine plasma overnight (4 °C) before constant measurement with VDS = −400 mV and VGS = 50 mV.

Fluorescence microscopy

Fluorescence microscopy images were obtained with a Nikon Eclipse Ni-E and either PLL-FITC or aptamers with a 6FAM-tag at the 5′ end. For images in the same figure (samples and controls) exposure and thresholds were set the same, but can vary between PLL-FITC and 6FAM-aptamer experiments. PEDOT:PSS formulations were functionalized with 4-ABA and PLL-FITC (0.2 mg/ml for 3 h) unless stated otherwise. Images were taken after 1 day in PBS and 30 min Tween 0.05%, except images in the SI which were taken immediately after a wash unless otherwise stated.

Frequency dependent OECT measurements and data analysis

For frequency dependent OECT measurements a National Instruments PXIe-1078 system with a PXIe-6259 DAQ and PXIe 4163 SMU with a Ag/AgCl pellet reference electrode (for VR) was used. For measuring IL6 on the interdigitated OECT channel, VDS = −50 mV was applied and a sinusoidal VGS (amplitude = 0.1 V, offset VGS = 100 mV) of various frequencies was applied for two wave periods. The reference electrode was placed close to the gate for the measurement. All electrodes were kept in place during the measurement, including during electrolyte changes. For each frequency, the voltage difference ΔV = (Vmax – Vmin) of the last period was calculated. With ΔVR the change of “output” voltage at the electrode in solution (VR) compared to the input voltage ΔVGS (nominally applied by the system, sine voltage) can be displayed as a system property gain:

$${gain}={\Delta V}_{R}/{\Delta V}_{{GS}}$$
(2)

The measured signal can also be fit for each frequency as a sine wave:

$${V}_{R}(t)=\sin (2\pi \cdot t\cdot f+\varphi )\cdot A+{V}_{0}$$
(3)

where A is the amplitude, f the frequency, φ the phase, and V0 the offset of the sine wave. For each frequency the voltage of the sine wave can also be represented as a complex number:

$${V}_{R}=A\cdot \cos \varphi +j\cdot A\cdot \sin \varphi \cdot \cos \varphi +j\cdot A\cdot \sin \varphi$$
(4)

From this complex form, the amplitude, phase, and real and complex components for Bode and Nyquist plots were obtained.