Multimodal Characterization of Cardiac Organoids Using Integrations of Pressure-Sensitive Transistor Arrays with Three-Dimensional Liquid Metal Electrodes.

Herein, we present an unconventional method for multimodal characterization of three-dimensional cardiac organoids. This method can monitor and control the mechanophysiological parameters of organoids within a single device. In this method, local pressure distributions of human-induced pluripotent stem-cell-derived cardiac organoids are visualized spatiotemporally by an active-matrix array of pressure-sensitive transistors. This array is integrated with three-dimensional electrodes formed by the high-resolution printing of liquid metal. These liquid-metal electrodes are inserted inside an organoid to form the intraorganoid interface for simultaneous electrophysiological recording and stimulation. The low mechanical modulus and low impedance of the liquid-metal electrodes are compatible with organoids' soft biological tissue, which enables stable electric pacing at low thresholds. In contrast to conventional electrophysiological methods, this measurement of a cardiac organoid's beating pressures enabled simultaneous treatment of electrical therapeutics using a single device without any interference between the pressure signals and electrical pulses from pacing electrodes, even in wet organoid conditions.

Organoids are simplified and miniaturized in vitro model systems of organs and they recently have attracted a lot of attention because of their applications in tissue development, disease modeling, and drug screening.[1] Despite rigorous advances in the culturing of physiologically optimized organoids in the relevant field, there is limited applicability due to inadequate tools for practical clinical translation. The integration of engineering innovation into complex organoid systems for more in-depth control and recording of mechanophysiological parameters has proved to be a problematic task.[2−4] The high culturing variability of organoids can be managed by engineering a potential design for physical cues during the self-organization processes.[5−8] Modern organs-on-a-chip devices have established systemic characterization and controlled synthesis of target tissue elements, which has improved the replicability of functional readouts significantly and extended their lifespans.[9−11] However, these devices restrict the natural self-organizing process of growing cell aggregates. The expression of the physiological properties is hampered when using such devices with those constraining conditions. These cell lines fundamentally follow the mechanophysiological properties set out by the architectural design of the respective organs-on-a-chip devices. Hence, when paired with such devices, the physiological relevance of these organoids ultimately is reduced.[12−14]

For the translation of the mechanophysiological parameters of cardiac organoids, the physiological parameters associated with the beating motions of the organoid need to be captured. Current analytic models of cardiac organoids have been limited to the electrophysiological, genomic, and optical characterization of their physiologic functionality.[15] It is difficult for most commonly used optical monitoring methods to provide full, viable information about the morphological property of the organoids.[16,17] A design that controls the explicit locations of the organoids without restricting their natural structure during electrophysiological characterization is an optimal setup that resolves the limitations of translatability.

Herein, we present an unconventional method for the multimodal characterization and stimulation of human-induced pluripotent stem cell (hiPSC)-derived cardiac organoids. Our device establishes unrestricted development of cardiac organoids while comprehensively monitoring the readout of their functional activity. Our device can record and control essential cardiac functions that are indispensable for physiological analysis without the reductionist engineering approach presented in modern organs-on-a-chip devices. Unlike organs-on-a-chip devices, which comprise premeditated properties of the in vivo microenvironment, our device unrestrictedly monitors the complex cellular architecture of cardiac organoids. The variation of the pressure of the organoid’s beating motions that were recorded using the pressure-sensitive transistors can correspond to electrocardiogram (ECG) traces and can generate negligible artifacts in the sensing signal even when an electrical impulse was delivered for cardiac organoid stimulation. Table S1 compares and differentiates our device from previous works on pressure sensors for biomedical applications.

Integration of Pressure-Sensitive Transistors and 3D Liquid-Metal Electrodes

Figure a shows the schematic illustrations of the sensing mechanism of our multimodal device, which includes the pressure-sensitive transistor array, and the integrated 3D liquid-metal electrodes. When a cardiac organoid is located on this device, the 3D electrodes can be inserted naturally inside the cardiac organoid. Simultaneously, the active-matrix array of pressure-sensitive field-effect transistors (FET) can monitor the beating motions of this cardiac organoid mechanophysiologically by detecting the spatiotemporal change in compressive pressures. In contrast to previous surface-type electrodes, this 3D electrode can stimulate and record the interior area of an organoid directly by forming the intraorganoid interface. In addition, the 3D electrodes serve as physical cues that support the growth of the cardiac organoids. For example, Figure b shows immunofluorescence images of a hiPSC-derived cardiac organoid on 2D (flat) and 3D electrodes. We observed that cardiac cells cluster around these 3D electrodes during their early initiation and conformally surround the electrodes during their maturation process.

Figure 1

Multimodal sensing device using directly printed 3D electrodes and pressure-sensitive transistor arrays. (a) Schematic illustration of the sensing mechanism of the multimodal device (top). Illustration of the integrated 3D liquid-metal electrodes (bottom left) and pressure-sensitive transistor array (bottom right). (b) Immunofluorescence images of hiPSC-derived cardiac organoid that is grown separately on 2D (flat) and 3D electrodes. Scale bar, 100 μm. (c) Plot of the maximum height of the 3D electrode versus the speed of vertical elevation. Inset: Schematic illustration of the electrode printing system. (d) Plot of the diameter of the electrode versus the diameter of the nozzle. Inset: Schematic illustration of the 3D electrode printing. Error bars in (d) indicate the standard deviations. (e) Photograph of the 3D electrodes after high-resolution direct printing (top). Scale bar, 600 μm. The SEM image of the 3D electrode tip (indicated by the red circle, bottom left). Scale bar, 5 μm. The photograph of the directly printed 3D electrodes with PDMS microwell for cardiac organoid culturing (bottom right). Scale bar, 150 μm.

The use of liquid metals in deformable electronic devices has been prominent due to their good processability at ambient conditions and superb electrical conductivities.[18−23] Their intrinsically low modulus and superb stretchability also can be advantageous for their use as biointerfacing electrodes, as conventional rigid solid metals can damage cellular interfaces because of the modulus mismatch between metallic solids and biological cells.[24] We used eutectic gallium–indium (EGaIn) because of its low toxicity and negligible volatility. Figure S1 compares the Young’s modulus of EGaIn with conventional solid-phase conductive materials. The modulus of EGaIn is about 1000 times lower than the values of solid conductors and it is comparable to the modulus of human cardiac tissues.[25] Upon exposure to air, EGaIn instantaneously forms a thin solid layer (∼1 nm) of gallium oxide on its surfaces under atmospheric oxygen levels. This oxide skin is thin enough to avoid damaging the cellular interfaces substantially and solid enough to maintain its 3D shape against gravity and surface tension. The printing setup consists of a printing nozzle connected to an ink reservoir that is connected to an external pneumatic pressure controller (see Supporting Information). By adjusting the stage from a fixed position to the direction of the z-axis, the 3D electrode can be printed (Video S1). Figure c shows the speed of the vertical elevation of a nozzle that presets the height of the printed 3D electrode. Also, the diameter of 3D electrode can be determined by the inner diameter of a nozzle (Figure d). For example, 3D electrodes with a diameter of 12 μm were printed using our nozzle of 10 μm in diameter (Figure S2), and the electrode diameter did not significantly degrade the electrode performance (Figure S3). Figure S4 shows the printing duration required to form a single 3D electrode, and 10 s typically were required to print a single 3D electrode with a height of 250 μm. Figure e shows optical micrographs of the 3D electrodes printed on a thin flexible polyimide (PI) film. The bottom-left inset shows a scanning electron microscope (SEM) image of the conical tip part of an EGaIn electrode. As shown in the bottom-right inset, a polydimethylsiloxane (PDMS) layer was perforated with a hole using a laser ablation machine as a microwell. Here, 3D electrodes were placed within the vacant hole of this PDMS microwell to promote the localization of the cell aggregate around the 3D electrodes during organogenesis.

Performance Characterization of the Multimodal Sensing Device

The schematic layouts of this multimodal characterization device for a cardiac organoid are shown in Figure a. This device is composed of two main components: (i) the bottom part comprises an array of pressure-sensitive FETs (as pressure sensors) for detecting the mechanophysiological compressive pressures, and (ii) the top part comprises 3D electrodes for the recording of electrophysiological signals and delivering of electrical stimulation. The image of the device and its relevant components are shown in Figure S5. Figure S6 illustrates the overall fabrication process of this device. We used an ultrathin, single-crystalline Si membrane as the FET channel by etching the top Si layer of a silicon-on-insulator wafer (see Supporting Information for detailed fabrication steps). For the elastomer layer, the embedding of glycerol microdroplets inside PDMS can drastically lower the value of the modulus of the elastomeric layer, which leads to large elastic deformation even at low compressive pressures.[26] Then, hollow holes were produced inside the elastomer via laser ablation. Figure S7 shows the PI film with the gate electrodes and Pt interconnects. The 3D electrodes were printed directly onto the open area of the Pt interconnects.

Figure 2

Characterization of the multimodal sensing device for cardiac organoid detection. (a) Schematic layouts of the multimodal sensor that is composed of an active-matrix pressure-sensitive transistor array with integrated soft liquid-metal electrodes. (b) Optical micrograph of the two adjacent transistors. S and D denote the source electrode and drain electrode, respectively. Scale bar, 20 μm. (c) Optical micrograph after the gate electrode (denoted as G) assembly. Scale bar, 50 μm. (d) Representative transfer characteristics of Si FET (VD = 5 V) of the device at ambient conditions. (e) Output characteristic of an air-dielectric FET (VG = 0 to 40 V, 10 V step) at ambient conditions. (f) Real-time measurements of normalized ID changes for the applied compressive pressures during electrical stimulation at different electrical fields (VG = 20 V, VD = 5 V). (g) Plot of the relative changes in ID versus applied compressive pressure to define pressure sensitivity (denoted as S). Error bars indicate the standard deviations. (h) Plot of the pressure sensitivity versus the different magnitudes of the electric field. (i) Impedance of the 3D EGaIn electrode (height, 250 μm; diameter, 60 μm; red line) and flat electrode (black line) over the frequency range from 10 Hz to 10 MHz. (j) Cyclic voltammetry plot of 3D EGaIn electrode (height, 250 μm; diameter, 60 μm; red line) and flat electrode (black line) over a range from −0.6 to 0.8 V at a sweep rate of 50 mV s–1.

For the spatiotemporal pressure mapping, a 10 × 10 active-matrix array of pressure-sensitive air-dielectric FETs was fabricated. On the basis of the average size of the cardiac organoid, the total size of the pressure-sensing region was optimized to 500 × 500 μm. Also, the channel size of the individual FET was designed as 10 × 10 μm, and the total distance between adjacent FETs (i.e., pixel resolution) was 50 μm. This active-matrix circuitry consists of scan and data lines in which a targeted pressure sensor (or pixel) was designed to operate separately to give direct electrical responses, according to the selection of the combination of rows and columns.[27−30]Figure b shows an optical micrograph of the Si channel with the patterned source/drain electrodes. Accordingly, the assembly of the gate electrode results in the formation of the air-dielectric FETs (Figure c). Figure d,e plots the representative transfer and output curves of this transistor at ambient conditions, respectively. The on/off ratio (Ion/Ioff) and threshold voltage (Vth) were 105 and 10.5 V, respectively. As shown in Figure S8, the statistical measurement data fit the Gaussian distribution profile by presenting outstanding homogeneity of sensitivity (average sensitivity: 0.35 ± 0.01 kPa–1) in all pixels. The clean interface between the Si channel and the air dielectric can exhibit this outstanding homogeneity at ambient conditions.[31−33] In particular, the air-dielectric structure can result in negligible hysteresis during the operation of the FET.[34] Our air-dielectric FET exhibited a response time of 48 ms and a recovery time of 42 ms (Figure S9).

Figure f shows the real-time detection of the relative change in ID [ΔID/I0 (%)], where ID is the drain current, I0 is the current at zero Pascal, and ΔID= I – I0, which denotes the variation of the ID during stepwise pressure loading. The sensitivity (S) of the transistor was obtained by plotting the relative change in ID concerning the applied pressure (Figure g) and was expressed as [ΔID/I0 (%)]/ΔP, where ΔP denotes the applied pressure. ΔID/I0 linearly increased within the pressure range from 1 to 5 kPa, where the sensitivity was approximately calculated as ∼0.35 kPa–1. In addition, the ΔID/I0 values measured under desired pressure loading showed negligible variation when an electrical field was applied simultaneously to the 3D electrodes during pressure sensing (Figure f). As shown in Figure h, the application of an electric field did not change the S value.

The significance of the 3D electrodes was validated by using electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV) analyses to compare the impedance of the 3D EGaIn electrode (height: 250 μm) with a flat surface Pt electrode using a potentiostat. The areal dimensions of both electrodes, i.e., 3D EGaIn and flat Pt, were identical, but the 3D electrode exhibited 250 μm in height. Both electrodes were submerged in a phosphate-buffered saline (PBS) solution with a reference electrode (Ag/AgCl) and a counter electrode (Pt sheet) at ambient conditions (pH 7.4, 25 °C). As plotted in Figure i, the 3D electrode and flat electrode exhibited impedance values (at 103 Hz) of 10.5 and 20.7 kΩ, respectively. The charge storage characteristic was tested via CV analysis conducted at a scan rate of 50 mV/s with a potential limit range from −0.6 to 0.8 V (Figure j). The 3D electrode showed an outstanding charge storage characteristic, which was notable in the large area of the enclosed CV curve. Overall, the 3D electrodes had lower impedance and a better charge storage characteristic, which indicates efficient performance for electrode-to-cell signal transfer.

Formation of Human Cardiac Organoids Using Cardiomyocytes Differentiated from hiPSCs

The shape and duration of the action potentials are incoherent between animal and human heart. Hence, animal modeling is often inaccurate in translation to human clinical setting in cardiology research.[35,36] Cardiomyocyte differentiation of hiPSCs has been employed to overcome these limitations and to recapitulate action potentials and functional structures of human cardiomyocytes. It is more efficient to obtain cardiomyocytes from hiPSCs than other sources since human fetal cardiac tissues are difficult to obtain, and the ability of adult stem cells to proliferate and differentiate into functional cardiomyocytes is known to be limited.[37] More importantly, when hiPSC-derived cardiomyocytes are cultivated in a 3D condition, their electrophysiological properties are further enhanced. Their phenotypic characteristics become closer to that of actual human heart tissue as the 3D structural environment increases the cellular interactions and enhances the differentiation and maturation of the cardiomyocytes.[38]

hiPSCs were seeded on a 6-well plate to differentiate hiPSCs into cardiomyocytes, and cardiomyocyte differentiation was induced according to the protocol established by Hoang et al. (Figure a).[39] Four days after hiPSC seeding, the differentiation process was initiated by replacing the culture medium with a differentiation medium (day 0). The differentiated cells with cardiomyocyte-like morphology appeared over culture time up to 16 days (Figure b). The immunofluorescence staining to check cardiomyocyte differentiation revealed that three cardiac markers, α-actinin, cardiac troponin T (cTnT), and cardiac troponin I (cTnI), were highly expressed in the differentiated cardiomyocytes (Figure c). Then, cardiac organoids were generated using hiPSC-derived cardiomyocytes that had started beating on day 17 to day 21. The cardiac organoid development in the specially designed PDMS microwell was monitored over culture time (Figure d). We confirmed the high level of cardiac marker (cTnT) expression in human cardiac organoids at day 21 (Figure e). Moreover, human cardiac organoids showed autonomous beating motion in the microwell (Video S2). Additionally, a live/dead staining assay conducted 3 days after the generation of cardiac organoids with the 3D electrodes showed good biocompatibility (Figure S10).

Figure 3

Formation of human cardiac organoids using human-induced pluripotent stem cell (hiPSC). (a) Timeline and culture conditions for cardiomyocyte differentiation from hiPSC. (b) Light microscopic observation of differentiated cardiomyocytes over culture time. Scale bar, 300 μm. (c) Immunofluorescent images of α-actinin, cardiac troponin T (cTnT), cardiac troponin I (cTnl), and F-actin in hiPSC-derived cardiomyocytes. DAPI was used for nucleus staining. Scale bars, 150 μm. (d) Light microscopic observation of 3D human cardiac organoids generated with differentiated cardiomyocytes over culture time. Scale bar, 300 μm. (e) Immunofluorescent image of cTnT and F-actin in human cardiac organoid (left; scale bar, 200 μm) and high-magnified image of cTnT in human cardiac organoid (right; scale bar, 30 μm). DAPI was used for nucleus staining.

Multimodal Characterization of the Cardiac Organoids

Since the average size of the cardiac organoids was no larger than 400 μm, the organoid spanned over multiple FET, which allowed detection of spatiotemporal changes in local pressure distribution during its beating. As shown in Figure a, our device simultaneously monitored the beating activity of a cardiac organoid by ECG (blue), pressure (red), and calcium imaging (black). The ECG signal (unit: mV) was recorded using the 3D electrodes and an electrophysiological measurement system that was comprised of a multielectrode array and a data processor with a real-time controller. The measurement of beating pressure was obtained by using automated instrumentation with a custom-made jig (Figure S11) connected to separate modules using two sourcemeters, a system switch, and a relay card. The readouts of each FET were converted to corresponding compressive pressure values. The organoid was stained using Fluo-4 AM for calcium imaging. The normalized fluorescence intensity at a specific region of interest (F/F0) of the calcium flux was plotted against time. All modes of characterization showed a consistent beating frequency of 12.1 beats per minute (bpm) with a consecutive beating interval of 4.94 s. Figure b presents the relative changes in the three characterization data during a single beat (indicated as the green dashed box in Figure a). Figure c shows pseudocolor fluorescence images of this organoid, which represents the changes of the calcium indicator at the time phases indicated in Figure b. These images indicated the depolarization (top) and resting state (bottom) of the organoid. The ECG signal recorded using the 3D electrodes did not significantly deteriorate for 7 days (Figure S12). Figure d visualizes the spatiotemporal changes in the beating pressure distribution as 2D color gradation contour plots. Video S3 presents the synchronization of this spatiotemporal pressure mapping with the calcium imaging during beating motions.

Figure 4

Multimodal characterization of cardiac organoids. (a) Real-time characterization plot of the ECG amplitude (blue), compressive pressure (red), and normalized fluorescence intensity (black) of the hiPSC-derived cardiac organoid by the multimodal characterization device. (b) Magnified plot of multimodal characterization data versus time which defines one beating cycle of the cardiac organoid, denoted as the green dotted box in (a). (c) Pseudocolor fluorescence images of the cardiac organoid at the instance of depolarization (top), labeled as time frame 1 in (b) and 3.8 s afterward (bottom) labeled as time frame 2 in (b). Scale bar, 100 μm. (d) Time-lapsed spatiotemporal color maps of compressive pressure distribution of the hiPSC-derived cardiac organoid to capture the instances of movement during a single beating. Each motion frame of the cardiac organoid was taken every 174 ms.

Electrical Stimulation of the Cardiac Organoids

Unlike previous studies that measured electrophysiological signals of an organoid by attaching thin, flexible electrodes to the outer surfaces of the organoid (extraorganoid sensing),[5,6,40,41] the 3D electrodes of our device were embedded inside a cardiac organoid for the intraorganoid electrophysiological recording. Modern methods of confinement are microfluidic devices or arrays of microwells, which can diminish the quality of physiological signals or are ill-equipped to make high-quality characterizations.[42−44] In contrast, our device was advantageous for localizing the organoid development toward the sensing site without any elements of constraint that potentially diminish the significance of the readouts. These 3D electrodes do not force organoids into a final specific shape; rather, they naturally guide the formation of the shape of the hiPSC-derived cardiac organoids during their growth. The SEM image of the cardiac organoid developed on these 3D electrodes is shown in Figure a. The cross section schematic inset in Figure b (left) illustrates the intraorganoid interface between the cardiac organoid and the 3D electrodes. In Figure S13, the immunofluorescence images of the cryosectioned cardiac organoid showed the positional traces of the 3D electrodes in the cardiac organoid, which represent the intraorganoid interfaces with no signs of bending or displacement. Figure b shows the immunofluorescent images of the organoid with the 3D electrodes (middle inset) and close-up of the surface of the organoid (right).

Figure 5

Simultaneous electrical stimulation of cardiac organoids with multimodal sensing. (a) Colorized SEM image of the hiPSC-derived cardiac organoid developed on the 3D liquid-metal electrodes. Scale bar, 100 μm. (b) Cross section schematics of the intraorganoid interface between the cardiac organoid and the 3D electrodes (left). The immunofluorescence images of the matured cardiac organoid with the 3D electrodes (middle inset) and close-up of the surface of the hiPSC-derived cardiac organoid (right). Scale bar, 200 and 20 μm, respectively. (c) Real-time characterization plot of the ECG amplitude (blue), compressive pressure (red), and normalized fluorescence intensity (black) of the hiPSC-derived cardiac organoid before and during electrical stimulation (gray dotted line, 0.8 V/mm). (d) Magnified single ECG signal denoted as the black dotted box indicated in (c). (e) Electrical circuit of cardiac organoid with 3D electrodes for electrical stimulation and characterization. The C, RP, and RO represent the capacitance of the organoid, the resistance of the probe, and the resistance of the organoid, respectively. Inset: Aerial schematics of the cardiac organoid. (f) Finite element analysis (FEA) simulation of the electric field for different cases of electrodes for the optimization of the organoid stimulation. Case 1 and 2 define the electric field simulation of the flat electrode at the top and bottom position, respectively. Case 3 defines the electric field simulation of the 3D electrodes located at the bottom base.

For the modulation of the organoid’s beating, it was subjected to an electrical stimulation pulse of 0.8 V/mm (pulse width, 1 ms; frequency, 1 Hz) using an externally connected pulse generator. Figure c shows the simultaneous multimodal characterizations (ECG, pressure, and calcium flux) of the organoid before and after electrical stimulation (gray-dashed box). A single ECG peak before applying the electrical impulses (black-dashed box in Figure c) is shown in Figure d. As seen, the depolarization of a cardiac organoid was indicated by a distinct P wave in the PQRS complex of the ECG signal. However, during the electric pacing, the electric pulses significantly disrupted the ECG trace and essentially rendered the interpretation of the ECG impossible. There was an undesirable interference with the electric signals due to the leakage through the organoid’s electrical conduction system, thereby limiting the simultaneous functions of ECG recording and electrical pacing. After the end of electrical stimulation, the ECG signal was returned to the normal state (Figure S14). Figure e shows a diagram of the electrical circuit of this cardiac organoid with 3D electrodes. In contrast, our mechanophysiological sensing enabled the recording of the organoid’s beating motions with no signal disruption during the application of electrical impulses in electrical therapeutics. The intraorganoid interface between the 3D electrodes and the organoid enlarged the contact area relatively to ensure efficient charge transfer. Also, as shown in the computational simulation of the electric potential for the shape and placement of electrodes inside a microwell (Figure f), the 3D microstructure of the electrodes for our device can be effective to focus and locally distribute the electric field near the pacing region (case 3), compared with the flat geometries of the other electrodes (case 1 and case 2).

Recent research related to organoids has performed exceedingly well in the recapitulation of cellular diversity and microscale tissue architecture to bring out critical functions of their in vivo counterparts. However, the issues of translational relevance and high-throughput functional readout remain crucial challenges in practical utilization for clinical applications.[43,45] In traditional culture systems, there is little or no integration of biophysical and topological parameters that can support the self-organization of organoids.[46] Hence, the growth environment of these organoids usually is versatile and ill-defined, which leads to high variability in organoid phenotypes. Although optical analysis methods have been used mainly for cellular characterization, organoids have 3D shapes and versatile motion displacements, which makes automated live imaging very challenging. The optical imaging method also holds little functional insight and relies on reporter lines for the assessment of the expression of specific markers. The multimodal sensing mechanism can provide a capacity for deeper contextual understanding of physiological parameters for more profound analytical studies. Furthermore, the implementation of each of the integrated methods was conducted independently and does not effectively affect the performance of the other methods. The straightforward microfabrication steps and the simplicity of the direct printing of 3D liquid-metal electrodes allow a smooth transition into large-scale microcompartmentalization for automated screenings. This multimodal sensory device suggests the future promise of an all-in-one analysis platform for a higher level of biomedical modeling of cardiac organoids in various areas, such as personalized medicine, disease modeling, and drug cardiotoxicity.