Hydrogen evolution and dynamics in hydrogel electrochemical cells for ischemia–reperfusion therapy

Abstract

Molecular hydrogen (H₂) protects organs from reactive oxygen species damage associated with ischemia–reperfusion (I/R) injury. Existing H₂ delivery methods, such as gas inhalation and H₂-rich water consumption, target the entire body and experience leakage during administration. Here we engineer a portable hydrogel electrochemical cell that enables on-demand H₂ production via the hydrogen evolution reaction. The system enables H₂ controlled generation, localized storage and sustained diffusion to the tissue–device interface, with better controllability and sustainability. We conduct a thorough study of H₂ evolution and dynamics in the hydrogel system, evaluating the influence of hydrogel polymer composition on the hydrogen evolution reaction kinetics, bubble morphologies and storage. We validate its protective effects (1) in vitro with cardiomyocytes and keratinocytes, (2) ex vivo in I/R hearts and (3) in vivo in skin I/R pressure ulcers. These findings demonstrate the potential of the hydrogel electrochemical cell design for efficient and sustainable H₂ delivery in I/R therapy, which could be broadly applied in other gas-based therapies and drug delivery research.

Hydrogen (H₂) gas is integral to numerous applications, including fuel cells¹, industrial chemical synthesis² and biomedical therapies³. The small molecular size and flammability of H₂ poses challenges in storage and release for both industrial and biomedical applications. Traditional production methods, such as steam reforming or electrolysis⁴, require harsh physical or chemical conditions, bulky equipment and storage under low temperatures or high pressures⁴. Therapeutic applications, which leverage the ability of H₂ to neutralize reactive oxygen species (ROS) for protection against ischemia–reperfusion (I/R) injury^3,5, are incompatible with such processes and demand mild, portable and controllable delivery systems.

Current H₂ delivery methods in clinical trials, such as gas inhalation⁵ and H₂-rich water drinking⁶ and H₂-rich saline injection⁷, face major limitations. Inhalation approaches are constrained by the percentage of H₂ delivered in the inhalation gas, and they also lack control over continuous, tissue-specific dosing. Drinking H₂-rich water is non-invasive and convenient but suffers from poor control over the administered H₂ concentration and similarly lacks spatial and temporal precision in delivery. H₂-rich saline injections offer more accurate and localized delivery; however, they are inherently invasive and not well suited for repeated or long-term treatment (Supplementary Table 1)⁸. Alternative approaches like photocatalysis⁹, algae–bacteria systems¹⁰, magnesium-based implants¹¹ and galvanic cells¹² are promising in mice, but exhibit limited external control and inconsistent production rates under dynamic biological conditions, rendering them unsuitable for precise and sustained therapy. These challenges underscore the need for advanced electrochemical engineering approaches that leverage material engineering, interfacial kinetics and molecular transport to enable controlled H₂ production and sustainable delivery for biomedical applications.

In this study, we introduce a hydrogel electrochemical cell designed for the controlled generation, localized storage and sustained diffusion of H₂ under mild, portable and regulated conditions. Hydrogels, that are water-rich, elastic and adaptable materials, are promising candidates for bioelectronics interfacing with biological systems¹³ and present an underexplored system for the H₂ evolution reaction (HER). Hydrogels, as soft materials, can be highly permeable to gases, making them ideal for gas therapy purposes^14,15. In the traditional liquid-electrolyte-based HER, H₂ bubbles escape readily, leading to inefficient diffusion and utilization in biological systems (Fig. 1a, top). By contrast, the three-dimensional (3D) water–polymer network of the hydrogel supports the HER and effectively and immediately traps H₂ bubbles, for safe and sustainable diffusion and improved biointerface delivery (Fig. 1a, bottom). We thoroughly study the H₂ evolution and dynamics in a hydrogel electrolyte system compared with a liquid electrolyte system, evaluating the influence of hydrogel polymer composition on electrochemical kinetics, gas morphologies and gas storage.

Fig. 1 |. H₂ generation, storage and diffusion in the hydrogel electrochemical cell.

a, H₂ generation (via the HER), storage and diffusion in solution and hydrogel electrolytes. b. Organ-specific device platforms. (i) and (ii) In CMs, hydrogel-electrochemical-cell-generated H₂ protects against I/R injury by neutralizing ROS. (iii) MEA–hydrogel attached on a pig heart. (iv) Structure of the MEA–hydrogel device and the HER process. (v) The H-Pad for protection against pressure ulcer injuries integrates the MEA–hydrogel with a flexible and wireless PCB. (vi) System design of the flexible PCB that supplies a constant current to the MEA-hydrogel device. DAC, digital-to-analog converter. (vii) and (viii) A portable H-Pad device patch comprising the Tegaderm film, MEA–hydrogel, PI film, battery, flexible PCB and 3D printed ring. Panel b created with BioRender.com.

We demonstrate the efficacy of the system in neutralizing ROS and preventing oxidative damage in a cell-free system and in cardiomyocytes (CMs) and keratinocytes in vitro (Fig. 1b(ii)). For tissue–hydrogel evaluation, we develop a membrane electrode assembly–hydrogel (MEA–hydrogel) device by attaching two Ti mesh electrodes on either side of a Nafion proton exchange membrane and embedding them in a hydrogel to create a compact and portable electrochemical cell (Fig. 1b(iv)). The hydrogel, serving as a biocompatible scaffold, traps H₂ and controls its release to the tissue–device interface, enabling localized and sustained H₂ delivery in ex vivo I/R heart models (Fig. 1b(i),(iii)) and in vivo skin pressure ulcer models (Fig. 1b(v)). We integrate a wireless flexible printed circuit board (PCB) to supply a constant current to the MEA–hydrogel device, which we envision will allow the on-demand control of H₂ production and dosage adjustment by patients or clinicians (Fig. 1b(vi)–(viii)).

The hydrogel electrochemical cell platform offers a promising and streamlined approach to the treatment of I/R injuries. The hydrogel itself fulfills three critical functions: H₂ generation, storage and diffusion, effectively integrating these processes into a single system. This multifunctionality not only simplifies traditional H₂-based therapies but also reduces costs as well as enables a portable and wireless therapeutic solution. Beyond H₂ therapy, the platform’s ability to facilitate electrochemical reactions within a hydrogel, coupled with its capacity for localized gas trapping and sustained diffusion, positions it as a versatile system for broader applications in other gas-based therapies and drug delivery research.

Results

H₂ generation, storage and diffusion in hydrogel electrolytes

Platinum (Pt) was selected as the electrocatalyst for HER due to its superior catalytic efficiency and biocompatibility^16–18. Scanning electron microscopy/energy dispersive X-ray spectroscopy analysis confirmed the successful attachment of Pt nanoparticles on the titanium (Ti) electrode surface (Supplementary Fig. 1a). Electrochemical characterizations, including linear sweep voltammetry (LSV) and electrochemical impedance spectroscopy (EIS), showed higher current density and lower impedance for Ti–Pt electrodes compared with bare Ti (Supplementary Fig. 1b,c). The reduced Tafel slope of Ti–Pt electrodes indicated enhanced reaction kinetics and catalytic activity (Supplementary Fig. 1d). Chronopotentiometry (CP) tests showed stable reaction kinetics over 12 h (Supplementary Fig. 1e).

To enable H₂ evolution in a soft and biocompatible platform, we selected a polyvinyl alcohol (PVA) hydrogel as the electrolyte matrix. PVA can be physically crosslinked through a freeze–thaw process without the need for chemical crosslinkers, which is critical for electrochemical stability as chemical crosslinkers may undergo oxidation or reduction during the HER operation, compromising both electrode integrity and hydrogel mechanical properties. We further evaluated the interaction between PVA hydrogels and H₂ gas by measuring the H₂ solubility in polymer solutions. The addition of polymers such as PVA, gelatin or alginate decreased the H₂ saturation concentration compared with Na₂SO₄ solution without polymers (Supplementary Fig. 2), which is probably attributed to the reduction in free water content available for gas dissolution. This decrease in solubility became more pronounced at higher polymer concentrations, as increasing the PVA content from 3% to 6% further lowered the dissolved H₂ concentration. In particular, among the different polymer types, the differences in H₂ solubility at the same mass ratio were relatively small.

To assess the electrode performance in hydrogel environments, PVA hydrogels with varying polymer concentrations (3%, 6% and 10% PVA in 0.1-M Na₂SO₄ solution) were synthesized using a freeze–thaw method to physically crosslink the polymer. A key distinction between solution-based and hydrogel electrolytes is bubble dynamics during the HER. In 0.1-M Na₂SO₄ solutions without PVA, bubbles escape when buoyancy exceeds adhesion to the electrode (Supplementary Video 1)¹⁹. In hydrogels, elastic forces trap bubbles either within the hydrogel matrix or at the electrode–hydrogel interface (Supplementary Videos 2–4). Using a PVA precursor solution without freeze–thaw crosslinking does not effectively trap H₂ gas on a sustained minutes to hours timescale (Supplementary Video 5), indicating that the polymer matrix plays an important role in long-term H₂ bubble retention. Gas trapping and restricted electrolyte diffusion in the hydrogel network causes reduced current density, increased impedance and dynamic instability (Fig. 2a–d).

Fig. 2 |. Characterization of the Ti–Pt electrode in solution and hydrogel electrolytes.

a, LSV of the Ti–Pt electrode in Na₂SO₄ solution and hydrogels. b, EIS of the Ti–Pt electrode at −600 mV (versus RHE) from 100 kHz to 0.1 Hz. CPE, constant-phase element. c, EITS of the Ti–Pt electrode at −600 mV (versus RHE) and 10 kHz. d, CP test of the reaction kinetics of the Ti–Pt electrode; −4.4 mA cm⁻² current density for 5 min. e, Images showing the structure of bubbles and Ti–Pt wires in different electrolytes; −4.4 mA cm⁻² current density for 5 min. f, Distance from the electrode of the distalmost bubbles in various hydrogels (independent experiment, n = 4). P_3%-6% = 0.0440. g, H₂ leakage from Na₂SO₄ solution and hydrogels into the air; −0.5-mA charge for 5 min (independent experiment, n = 4). P_3%-6% = 0.0006. P_3%-solution < 0.0001. h, Micro-CT results showing H₂ trapping in the 3% hydrogel compared with the solution electrolyte. i, H₂ diffusion with solution or hydrogel coating on a Ti mesh; −5-mA charge for 6 min. This figure shows the concentration of dissolved H₂ sensed by the H₂ sensor at different times (independent experiment, n = 3). j, Concentration of dissolved H₂ with different charging currents for 6 min (independent experiment, n = 3). k, H₂ diffusion modeling at 0 h, 1 h, 12 h and 24 h with 9.33 μmol of H₂ gas trapped inside the hydrogel (top) and heat map of the H₂ profile from 0 h to 24 h between 0 cm and 1 cm above the Tegaderm film (bottom). 0 mm marks the Tegaderm film. Data are presented as mean ± s.d. P values (shown in f and g) are determined by ordinary one-way ANOVA with Tukey’s multiple comparisons test.

Bubble morphology depends on the hydrogel polymer content and modulus. In 3% hydrogels (lowest modulus; Supplementary Figs. 3 and 4), bubbles remained discrete. Hydrogels with a higher polymer content (and therefore higher modulus) exhibit smaller strain changes under the same applied stress (Supplementary Fig. 5), compressing bubbles closer to the electrode surface and facilitating the debonding of hydrogels from the electrode and H₂ gas channel formation (Fig. 2e,f), as confirmed by mechanical simulations (Supplementary Fig. 6). Gas channel formation obstructs the catalytic sites and reduces the available electrode surface area, leading to further reduced current density and increased impedance as the hydrogel polymer content increases (Fig. 2a,b and Supplementary Fig. 7)^20,21.

The EIS curves (from Fig. 2b) were modeled using the Randles electrical equivalent circuit, with R_s, R_CT and CPE representing the solution resistance, charge-transfer resistance and constant-phase element, respectively. The simulation results show increases in R_s and R_CT and a decrease in CPE as the polymer content of the hydrogel increases (Supplementary Table 2), attributed to the reduced electrode surface area and restricted electrolyte diffusion in denser polymer networks. The interfacial capacitance also decreases due to the lower dielectric constant of trapped gas (~1) compared with water (~78.2). Over time, bubble accumulation in the hydrogel electrolytes leads to increasing real, imaginary and total impedance, unlike the stable behavior in solution electrolytes, as observed in electrochemical impedance time spectroscopy (EITS; Fig. 2c and Supplementary Fig. 8).

In the 6% and 10% hydrogel samples, voltage oscillations were observed when a constant current was applied to the Ti–Pt electrode (Fig. 2d). Similar impedance oscillations in the EITS tests were also observed for the 10% hydrogel sample (Fig. 2c and Supplementary Fig. 8). Video analysis revealed corresponding mechanical oscillations (Supplementary Videos 3 and 4 and Supplementary Fig. 9), attributed to dynamic instability caused by quasiperiodic gas leakage from the gas channels at the Ti–hydrogel–air interface (Supplementary Video 6 and Supplementary Fig. 10). By contrast, the 3% hydrogel showed no voltage or mechanical oscillations, as gas bubbles remained distinct at the interface without forming interconnected gas channels (Supplementary Video 2 and Supplementary Fig. 11). This stability minimized the gas leakage compared with the 6% and 10% hydrogels under identical charging conditions (Fig. 2g and Supplementary Figs. 12 and 13). Further reducing the polymer content to 2% resulted in a liquid rather than gel following the freeze–thaw process, which was unable to trap gas effectively (Supplementary Fig. 14). Using the 3% PVA precursor solution without the crosslinked matrix also did not trap gas effectively (Supplementary Fig. 15). Further increasing the number of freeze–thaw cycles in 3% PVA hydrogel reduced the pore size, resulting in slower H₂ diffusion and decreased H₂ retention (Supplementary Fig. 16). On the basis of the electrochemical performance and gas storage capability, the 3% hydrogel with one freeze–thaw cycle was, therefore, selected as the electrolyte for subsequent studies.

Current H₂ therapies rely on closed environments, confining cells or animals in closed chambers with specific H₂ and O₂ concentrations^3,5,22,23. This approach has limited clinical relevance due to the need for patients to be immobilized and the potential safety risks from H₂ flammability. Leveraging a hydrogel-based gas trapping mechanism, we developed an open-system electrochemical device for sustained H₂ delivery. Micro-computed tomography (micro-CT) analysis of a Ti–Pt wire immersed in 3% hydrogel or Na₂SO₄ solution electrolyte under −0.5 mA for 5 min revealed that 17.6 mm³ of gas was trapped in the hydrogel, compared with minimal retention in the solution (Fig. 2h), closely matching the theoretical yield of 18.7 mm³ at room temperature and pressure. However, it is also important to note that the trapped gas visualized by micro-CT may include dissolved ambient gases (for example, O₂, N₂ and CO₂), not only H₂, leading to slight deviations from the gas chromatography (GC) H₂ leakage result.

The gas retention properties of the hydrogel result in a distinct H₂ diffusion profile. To gain more insights into hydrogel-enabled diffusion, we designed a H₂-sensing setup utilizing a Ti mesh electrode to enhance H₂ production and a 3M Tegaderm-sealed chamber to separate the electrochemical cell from a sensor-containing solution (Supplementary Fig. 17a). The Ti mesh–hydrogel electrochemical cell was engineered by electrodepositing Pt nanoparticles onto the Ti mesh and coating it with a 1.3-mm-thick 3% PVA hydrogel. As expected, our simulation results (Supplementary Fig. 18) revealed that the hydrogel thickness would influence both H² peak concentration and protection time. Real measurements of H₂ diffusion through the Tegaderm film revealed a higher H₂ concentration peak and prolonged release with the hydrogel coating, compared with Na₂SO₄ solution, extending the reported effective treatment ([H₂] > 25 μM)³ duration nearly 36-fold, from 0.65 h to 23.5 h, with a small energy input of 0.5 mAh (Fig. 2i). The sustained diffusion and therapeutic concentration of H₂ enabled by the hydrogel indicates the potential for greater H₂ utilization efficiency. Optical imaging of the H₂ bubbles on the Ti mesh reveals that the bubble size gradually decreased over time (Supplementary Fig. 19). Interestingly, once the bubbles shrank below a certain threshold, their size stabilized. This stabilization may be due to residual elastic deformation in the hydrogel network, which does not fully recover after prolonged mechanical stress from the trapped gas. The H₂ delivery profile may be further tuned by adjusting the applied current, demonstrating the versatility of the hydrogel for clinical H₂ therapies (Fig. 2j). Figure 2k presents a numerical simulation of H₂ diffusion during a representative therapy scenario. On the basis of the generation of 9.33 μmol of H₂ (equivalent to 5 mA for 6 min), the model predicts that the H₂ concentration at the Tegaderm film surface reaches a peak of approximately 480 μM at 85 min, followed by a gradual decline over time (Supplementary Fig. 17b). Figure 2k (bottom) further illustrates the spatiotemporal H₂ concentration profiles between the Tegaderm surface and at a location 1 cm above it. These results demonstrate that the closer to the hydrogel interface, the earlier and higher the H₂ concentration peak occurs, whereas more distant regions experience slower peak and lower peak concentrations. This highlights the inherently localized nature of H₂ delivery, with limited penetration depth in the absence of external convection or pressure-driven transport. In particular, residual dissolved H₂ persisted in the hydrogel after 24 h, highlighting its potential utility in prolonged therapeutic applications.

H₂ delivery via hydrogel mitigates oxidative damage

Hydroxyl radicals (•OH) are a highly reactive and damaging ROS with strong affinity for electron-rich biomolecules, including proteins and DNA²⁴. To test the efficiency of the Ti mesh–hydrogel electrochemical cell in reducing •OH, we utilized the Fenton reaction (Fe²⁺ + H₂O₂ = Fe³⁺ + •OH + OH⁻) in a cell-free system, where •OH is generated on the introduction of hydrogen peroxide (H₂O₂) into an Fe(ClO₄)₂ solution at a specific time point³. Hydroxyl phenyl fluorescein was used as a fluorescent probe to monitor changes in •OH concentration. As shown in Fig. 3a,b, the addition of H₂O₂ to the Fe(ClO₄)₂ solution caused a rapid increase in •OH levels, whereas the addition of an equivalent volume of phosphate-buffered saline (PBS) had no effect. Pretreatment of the Fe(ClO₄)₂ solution with H₂, delivered using the Ti mesh–hydrogel electrochemical cell, significantly reduced the •OH levels, demonstrating the effectiveness of H₂ in neutralizing •OH.

Fig. 3 |. H₂ delivery via the hydrogel electrochemical cell mitigates oxidative damage in cell-free and in vitro cellular environments.

a,b, •OH the Fenton reaction in a cell-free system, with and without H₂ pretreatment, as shown by fluorescence intensity fold changes (a) and quantification of fluorescence intensity fold changes (b; independent experiment, n = 4). P_{H2O2−H2+H2O2} = 0.0030. Fluorescence levels measured using hydroxyl phenyl fluorescein. H₂+H₂O₂ group underwent a 30-min H₂ pretreatment using the Ti mesh–hydrogel device before the addition of H₂O₂. c, Schematic of Ti mesh–hydrogel electrochemical cells producing H₂ to scavenge the ROS generated by the copper-based Fenton reaction in CMs. d, Mechanism of the copper-based Fenton reaction, which converts cellular H₂O₂ into •OH, inducing oxidative stress. e, CMs stained with CellROX demonstrate that H₂ treatment reduces the levels of cellular ROS induced by the copper-based Fenton reaction. f, Quantification of CellROX fluorescence intensity in (Cu+Vc)-treated CMs with or without H₂ pretreatment (independent experiment, n = 4). P_{Cu+Vc−H2+Cu+Vc} = 0.0088. g, Viability of CMs treated with or without Cu+Vc and H₂, assessed by PI and Hoechst staining. Cell viability was quantified from the staining images (independent experiment, n = 5). P_{w/o-w-without Cu+Vc} = 0.3411. P_{w/o-w-with Cu+Vc} < 0.0001. h, Side scatter area (SSC-A) flow cytometry analysis of PI staining in CMs treated with or without Cu+Vc and H₂. i, Flow cytometry histograms of PI staining in CMs treated with or without Cu+Vc and H₂, corresponding to the data in h. j, NLRP3 immunofluorescence staining in CMs treated with or without Cu+Vc and H₂. Right: quantitative fluorescence analysis of NLRP3 expression (independent experiment, n = 4). P_{w/o-w-without Cu+Vc} = 0.9298. P_{w/o-w-with Cu+Vc} = 0.0122. k, Caspase-1 immunofluorescence staining in CMs treated with or without Cu+Vc and H₂. Right: quantitative fluorescence analysis of caspase-1 expression (independent experiment, n = 4). P_{w/o-w-without Cu+Vc} = 0.5948. P_{w/o-w-with Cu+Vc} = 0.0002. l, Schematic depicting the influence of H₂ treatment on inflammasome expression and caspase-1 activity, highlighting its impact on cell inflammation and viability. Data are presented as mean ± s.d. P values in g, j and k are determined by a two-sample t-test (two tailed). P values in b and f are determined by ordinary one-way ANOVA with Tukey’s multiple comparisons test. Panels d and l created with BioRender.com.

To verify the H₂ effect in vitro before organ-level application, we developed a cell model using a copper-based Fenton reaction to convert endogenous H₂O₂ into •OH, inducing severe oxidative stress (Fig. 3c,d)³. CMs or HaCaT keratinocytes were seeded on a Tegaderm membrane, which permits H₂ diffusion but prevents electrolyte exchange and electricity leakage (Supplementary Fig. 20). The Ti mesh–hydrogel setup generated, stored and released H₂ to cells (Fig. 3c). It is worth noting that with the gas-permeable but liquid-impermeable Tegaderm membrane, the pH in the Tegaderm chamber remained stable during and after device operation (Supplementary Fig. 21a) and the O₂ level decreased as H₂ started to diffuse and replace the O₂ in the solution (Supplementary Fig. 21b). This design allows H₂ gas to diffuse into the tissue and effectively isolate the hydrogel from direct exposure to biological electrolytes, proteins and interstitial fluids. Intracellular ROS levels, measured by CellROX and H2DCFDA probes, were elevated in cells treated with copper ions and vitamin C (Cu+Vc) relative to controls but were markedly reduced with H₂ exposure (Fig. 3e,f and Supplementary Figs. 22–24).

We evaluated cell viability using propidium iodide (PI)/Hoechst staining and PI flow cytometry in both CMs and HaCaT cells under Cu+Vc-induced oxidative stress. H₂ treatment reduced CM death from ~60% to ~25% (Fig. 3g and Supplementary Fig. 25) and HaCaT cell death from ~70% to ~30% (Supplementary Figs. 26 and 27). Flow cytometry confirmed these findings in CMs (Fig. 3h,i). To address the potential contribution of electrical stimulation, we further conducted control experiments using two Ti meshes without catalyst and applying 2 V across these two electrodes. Results showed no significant difference in CM death rates between the Cu+Vc and Cu+Vc+electrical stimulation groups (Supplementary Fig. 28).

Expression levels of inflammatory markers NLRP3 and caspase-1 were markedly decreased in H₂-treated CMs subjected to Cu+Vc-induced oxidative stress relative to untreated controls (Fig. 3j,k and Supplementary Figs. 29 and 30). No significant difference emerged between the sham and H₂-only groups. These findings suggest that H₂, delivered by a Ti mesh–hydrogel bioelectronic system, modulates the NF-κB signaling pathway, thereby reducing inflammation, protecting cells from oxidative damage and improving cell survival (Fig. 3l)²⁵.

H₂ delivery via MEA–hydrogel mitigates heart I/R injury

To explore the therapeutic potential of our hydrogel bioelectronic device at the tissue level, we used the clinically relevant myocardial I/R injury model characterized by excessive ROS generation and exacerbated tissue damage^26,27.

We designed an MEA–hydrogel configuration as a self-contained hydrogel bioelectronic device without external reference and counter electrodes (Fig. 4a, bottom). This compact system incorporates Ti–Pt and Ti–IrO₂ mesh electrodes on opposite faces of a Nafion proton exchange membrane, all embedded in a hydrogel matrix, resulting in a portable bioelectronic device for water splitting (Supplementary Fig. 31). We thoroughly evaluated and confirmed the stability and oxygen evolution reaction efficiency of the IrO₂ catalyst (Supplementary Fig. 32). Proton conduction through the proton exchange membrane drives H₂ generation on the cathodic Ti–Pt side. Electrodeposition of Pt and IrO₂ on the respective electrodes improved the electrochemical performance and reduced the impedance (Fig. 4b and Supplementary Fig. 33). Although the hydrogel coating slightly reduces the MEA device performance, it provides beneficial gas trapping and diffusion properties essential for therapeutic H₂ delivery. We evaluated the maximum H₂ storage capacity of the MEA–hydrogel device, confirming that it is sufficient to accommodate the amount of H₂ produced by a 5-mA, 6-min charging cycle (Supplementary Fig. 34a). We further demonstrated the device’s repeated H₂ production (Supplementary Fig. 34b) and trapping efficiency, showing that even after repeated charging, the hydrogel maintained around 80% gas-trapping efficiency and stable mechanical properties (Supplementary Figs. 34c,d and 35). To assess the operational safety of the device, we conducted a flame exposure test (Supplementary Video 7 and Supplementary Fig. 36), in which no continuous flame was observed following H₂ generation, indicating that the hydrogel matrix safely buffers and releases H₂ in a non-combustible manner.

Fig. 4 |. H₂ delivery via MEA–hydrogel protects ex vivo heart tissue and function against I/R injury.

a, Schematic and photograph of the MEA–hydrogel device and experimental setup for H₂ production to treat an isolated ischemic heart. b, LSV of three MEA samples (without Pt or IrO₂ electrodeposition, immersed in Na₂SO₄ solution, coated with 3% PVA hydrogel) from 1 V to 3 V. This figure plots the apparent current density against the voltage across the cathode and anode. c, Isolated rat heart experimental setup and protocols. For H₂ treatment, isolated rat hearts were immersed in 37 °C Tyrode’s buffer to maintain physiological temperature and placed on a Tegaderm film. MEA–hydrogel device was placed underneath the Tegaderm film and charged at a current of 5 mA for 6 min at the beginning of ischemia. d, Representative photographs of TTC-stained heart sections. The white area indicates the size of the infarcted area. e, Quantification of infarction size in each group (independent experiment, n = 5). Data are presented as mean ± s.d. P values are determined by ordinary one-way ANOVA with Tukey’s multiple comparisons test. P_I/_R-I/_R+H2 = 0.0007. f, LVP measurements in isolated rat hearts following the start of reperfusion, with or without H₂ treatment (independent experiment, n = 3). g, Spectrogram of the LVP signal shown in f, highlighted by the dashed-line box. h, Photograph of the multichannel electrode array that records the isolated rat heart ECG. i, LVP, ECG and ECG propagation delays recorded from sham hearts, I/R hearts and H₂-treated I/R hearts (independent experiment, n = 3). Panels a and c created with BioRender.com.

We assessed the therapeutic impact of the MEA–hydrogel device using a Langendorff I/R model (Fig. 4a,c)²⁸. A saline-filled balloon was inserted into the left ventricle of the isolated Langendorff heart to enable real-time pressure monitoring of the left ventricular pressure (LVP). Hearts were perfused with Tyrode’s buffer for 1 h. Ischemic injury was induced by halting the perfusion flow for 30 min, followed by reperfusion for 45 min. During ischemia, the isolated heart was placed onto the Tegaderm chamber, with the MEA–hydrogel device underneath. A brief 6-min, 5-mA charge generated H₂ within the hydrogel, enabling sustained release into the cardiac tissue. We performed a H₂ diffusion simulation with a 3D scanned rat heart model inside the Tegaderm chamber. The result revealed that the region of the heart in contact with the Tegaderm film received a large amount of H₂ during the 30-min ischemia (Supplementary Fig. 37), which could eliminate the localized •OH ROS that surges post-reperfusion^26,29. After I/R injury, heart slices were stained with triphenyltetrazolium chloride (TTC) to detect the infarcted areas. Hearts treated with H₂ exhibited a significant reduction in infarct size compared with the control group without H₂ treatment (Fig. 4d,e).

Functional assessments confirmed these histological findings. LVP measurements and spectrogram analysis showed that H₂-treated hearts recovered regular contractile activity with robust contraction pressure and stable contraction period within 3 min after the start of reperfusion (Fig. 4f,g). After 45 min of reperfusion, H₂-treated hearts exhibited heart rates, LVP and electrocardiogram (ECG) patterns similar to healthy, preinjury controls. By contrast, untreated I/R hearts displayed arrhythmias and diminished contractility (Fig. 4i). To show the practicality of our approach, we further demonstrated that our device can be attached with the epicardium, supplying H₂ to cardiac tissue and improving cardiac function (Supplementary Fig. 38).

To directly assess electrical conduction, a flexible 16-channel microelectrode array recorded epicardial potentials (Fig. 4h). Isochrone maps after reperfusion demonstrated markedly delayed activation in I/R hearts, whereas H₂-treated hearts showed conduction patterns with smaller delay time (Fig. 4i).

Collectively, these results confirm that H₂ generated via our MEA–hydrogel bioelectronic device protects against I/R injury. This treatment reduces infarct size, enhances cardiac contractile recovery and restores electrical conduction. Such advances highlight the device’s potential as a clinically relevant, bioelectronic solution for managing oxidative-stress-induced cardiac damage, beneficial for patients at high risk of heart tissue damage, such as severe coronary artery disease or undergoing major heart surgeries³⁰.

H₂ delivery via portable H-Pad for I/R skin pressure ulcers

Pressure ulcers (bedsores) present a global healthcare challenge, causing severe patient discomfort and reduced quality of life. However, this complication has received relatively limited public health attention. Advancements in novel dressings and therapies have shown limited additional benefits, and the treatment remains vague and time-consuming, underscoring an urgent need for improved preventive and therapeutic strategies³¹. Here we introduce a hydrogel bioelectronic device designed to mitigate pressure injuries and enhance clinical outcomes.

The pathophysiology of pressure ulcers involves mechanical loading (for example, pressure, shear and friction) and prolonged immobility, resulting in restricted blood flow and localized ischemia³². Ischemia impairs mitochondrial electron transport, leading to excessive ROS generation. When the patient’s body position changes and blood flow is restored, reperfusion further increases ROS production³³. This surge in ROS overwhelms the tissue’s antioxidant defenses, initiating oxidative stress, inflammation, disrupting cytokine signaling and delaying wound healing, ultimately leading to tissue necrosis.

To enable clinically translatable solutions, we developed the H-Pad system (Fig. 5), a sustainable H₂ delivery platform that integrates the MEA-hydrogel with a flexible and wireless PCB. The PCB provides stable current delivery, voltage measurement and Bluetooth-based communication (Fig. 5a,b and Supplementary Figs. 39 and 40). The functional block diagram of the PCB (Fig. 5a and Supplementary Fig. 41) includes a microchip that controls a digital-to-analog converter and regulates current through a Howland circuit, ensuring controlled H₂ production.

Fig. 5 |. H₂ delivery via a portable H-Pad bioelectronic device for the in vivo treatment of I/R skin pressure ulcers.

a, Functional block diagram of the major PCB components. b, Major components on the PCB. BLE, Bluetooth low energy. c, Digital images showing how ischemia, reperfusion and H₂ therapy are performed. d, H-Pad system to treat skin pressure ulcer. From bottom to top: Tegaderm film, MEA–hydrogel device, PI film and PCB. e, Representative images of dorsal skin pressure ulcers at different time points. f, Thermal images of H₂-treated and control groups on day 7. g, Minimal temperature of the ulcer site on day 7 in H₂-treated and control groups (independent experiment, n = 5). P_control–H2 = 0.0005. h, Comparison of wound area on day 7 in H₂-treated and control groups (independent experiment, n = 5). P_control–H2 = 0.0115. i, Representative H&E staining images showing higher dermis thickness and less edema on day 7 with H₂ treatment. j, Dermal thickness of the dorsal ulcer site on day 7 in H₂-treated and control groups (independent experiment, n = 5). P_control–H2 = 0.0015. k, Comparison of the inflammation lesion area of the dorsal ulcer site on day 7 in H₂-treated groups and control groups (independent experiment, n = 5). P_control–H2 = 0.0018. l, CD4 histochemical staining in the ulcer tissue after H₂ treatment shows a modulated and mild immune response. m, Number of CD4-positive cells on day 7 in H₂-treated and control groups (independent experiment, n = 5). P_control–H2 <0.0001. n, CD31 histochemical staining in ulcer tissue after H₂ treatment shows enhanced neovascularization. o, Number of blood vessels on day 7 in H₂-treated and control groups (independent experiment, n = 5). P_control–H2 = 0.0037. Data are presented as mean ± s.d. P values in g, h, j, k, m and o are determined by a two-sample t-test (two tailed).

We adopted a previously reported protocol to induce pressure ulcers on the mouse dorsal skin³⁴, using two magnets to compress the skin for 6 h, followed by release to induce I/R injury (Fig. 5c and Supplementary Fig. 42). The experimental group (H₂ group, five mice) received H-Pad treatment (Fig. 5d), whereas the control group (five mice) received only Tegaderm film coverage. A user interface (Supplementary Fig. 43) was developed to control the system, delivering a 5-mA current for 6 min, repeated thrice every 12 h during the first 36 h post-injury (Supplementary Fig. 42). This treatment schedule specifically targeted the critical ROS surge post-reperfusion. Visual confirmation of active H₂ generation was observed via bubble formation beneath the PI film (Fig. 5c). Under the protection of a Tegaderm film, the hydrogel would not swell as the film is waterproof (Supplementary Fig. 44).

Wound assessment over 7 days revealed that the H₂-treated sites showed transient redness on days 2 and 4, followed by reduced wound area and recovery (Fig. 5e,h). By contrast, the control sites exhibited progressive blanching, ringed inflammation and necrosis. Thermal imaging demonstrated higher local temperatures in H₂-treated wounds (Fig. 5f,g and Supplementary Fig. 45), indicating improved perfusion and metabolic function.

Hematoxylin and eosin (H&E) staining revealed pronounced differences in tissue architecture and inflammatory responses between the H₂-treated and control groups (Fig. 5i). The control group displayed severe dermal and epidermal damage with extensive edema and fluid accumulation, indicative of an exacerbated inflammatory response. By contrast, the H₂-treated group exhibited markedly reduced tissue disruption, diminished edema, smaller inflammatory lesions (Fig. 5k and Supplementary Fig. 46), increased dermal thickness (Fig. 5j) and evidence of epidermal regeneration (Supplementary Fig. 47). To assess whether electrical stimulation alone contributed to the observed therapeutic outcomes, a second control group was introduced, in which MEA–hydrogel devices without electrocatalysts were placed onto the skin. An external constant voltage of 2 V was applied for 6 min, repeated three times every 12 h during the first 36 h post-injury. The MEA without catalyst group exhibited wound outcomes similar to the untreated control group, including progressive skin blanching, the formation of red inflammatory rings and significantly lower skin surface temperatures. Histological analysis further revealed larger inflammatory lesions and thinner dermis, indicating the absence of protective or regenerative effects in the absence of H₂ generation (Supplementary Fig. 48).

Immunohistochemical staining for CD4 revealed fewer CD4⁺ T-cells in the H₂ group, indicating reduced inflammation (Fig. 5l,m and Supplementary Fig. 49), whereas immunohistochemical staining for CD31 showed significantly increased neovascularization in H₂-treated tissue (Fig. 5n,o and Supplementary Fig. 49). Further staining with NQO1, a downstream target of oxidative stress signaling, showed a markedly decreased NQO1 signal in the H₂ group compared with the control group, supporting that H₂ delivery effectively alleviated ROS-induced stress (Supplementary Fig. 50). Collectively, these results demonstrate that H₂-based therapy attenuates ROS-induced damage, moderates inflammation and promotes vascularization, offering a promising intervention for pressure ulcer management.

Discussion

This study presents a hydrogel-based electrochemical system that integrates controlled HER with localized gas storage and sustained release, providing a platform in gas storage and transport for biomedical applications. We reveal the influence of hydrogel polymer composition and mechanical properties on electrochemical kinetics, gas morphologies and gas storage. In vitro, this electrochemical configuration significantly reduces ROS damage, enhancing the viability of both CMs and keratinocytes under oxidative stress. Ex vivo, the system mitigates myocardial infarction in I/R injured hearts, restoring electrical and contractile functions. In vivo, a wireless and portable H-Pad system minimizes skin damage in a pressure ulcer model, improving metabolic activity, reducing inflammation and promoting neovascularization.

These findings underscore the potential of a hydrogel electrochemical system for continuously delivering therapeutic chemicals. Beyond H₂ therapy, this strategy could be adapted for other bioactive gases, such as oxygen therapy^35,36 or combined with drug-loaded hydrogels, which can be engineered to incorporate a wide range of therapeutic agents through physical encapsulation or chemical conjugation, enabling controlled and sustained drug release alongside H₂. Recent advances in wearable and wireless bioelectronics, implantable fuel cells and biosensing platforms could be integrated to provide long-term power sources, monitor skin metabolism and assess therapeutic conditions^37–39. Future investigations could also leverage this electrochemical hydrogel platform to systematically explore the influence of therapeutic gases on fibrotic signaling pathways. Such studies may help reveal whether electrochemically generated H₂ can modulate fibroblast activation or extracellular matrix remodeling in chronic injury models. Since bedsores frequently arise in immobile patients during prolonged hospitalization or long-term care, a smart mattress capable of delivering controlled H₂ therapy could play a role in mitigating ROS damage and enhancing wound healing. Ongoing efforts are directed toward scaling up the device to an electronic therapeutic mattress that integrates H₂ sensors with the MEA–hydrogel, incorporating proportional–integral–derivative control algorithms for closed-loop steady H₂ supply, and ensuring long-term operational stability.

Methods

Further methods are provided in the Supplementary Information.

Animal protocols

All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Chicago in protocol nos. 72378 (rat) and 72621 (mice).

Chemicals

Ti wire (diameter, 0.25 mm) and Ti mesh (80 mesh; diameter, 0.13 mm) were purchased from Fisher Scientific. Iridium(IV) chloride hydrate (IrCl₄·H₂O), potassium tetrachloroplatinate(II) (K₂PtCl₄) and PVA (99%+ hydrolyzed; molecular weight (M_w), 89,000–98,000) were purchased from Sigma-Aldrich. Tegaderm Ag Alginate Silver Dressing was purchased from 3M. Anti-NLRP3 monoclonal antibody was purchased from Biotechne and anti-caspase-1 monoclonal antibody was purchased from Proteintech. Unless otherwise noted, all other chemicals were purchased from Sigma-Aldrich without further purification.

Electrodeposition of electrocatalyst on Ti electrode

To prepare the electrodeposition solution, 0.1-M iridium chloride (IrCl₄·H₂O) was dissolved in deionized water and stirred for 30 min. Then, 40-mM oxalic acid [(COOH)₂·2H₂O] and 100-mM H₂O₂ were added, and the mixture was homogenized for 10 min. To adjust the pH to 10.5, 340-mM potassium carbonate (34.5% K₂CO₃) was added, followed by stirring for 3 days to ensure the stability of the prepared electrodeposition solution⁴⁰. Ti electrode was electrodeposited inside the solution with 0.7 V (versus saturated Ag/AgCl electrode) for 180 s to yield the Ti–IrO₂ electrode.

For the electrodeposition of Pt, a solution with 2.0-mM K₂PtCl₄ and 0.5-M HCl was prepared. The Ti electrode was electrodeposited inside the solution with −0.25 V (versus saturated Ag/AgCl electrode) for 400 s to yield the Ti–Pt electrode.

Preparation and coating of hydrogel on electrode

3 wt%, 6 wt% and 10 wt% PVA hydrogel precursor solution was prepared by dissolving 3 g, 6 g and 10 g of PVA powder in 0.1-M Na₂SO₄ solution, respectively. The mixture was stirred and heated to 80 °C to ensure homogeneity and then cooled to room temperature. The hydrogel precursor solution was poured onto the electrode and frozen in a −20 °C freezer for 16 h, after which time it was placed at room temperature to thaw for 8 h to achieve a hydrogel coating of 1.3 mm thickness on electrode.

Electrochemistry tests of Ti–Pt electrode in solution and hydrogel

LSV, EIS, EITS and CP were conducted with the aid of a potentiostat (Corrtest electrochemical workstation). For these experiments, a Pt wire served as the counter electrode, whereas an Ag/AgCl electrode immersed in saturated KCl acted as the reference electrode. The working electrodes were Ti–Pt electrodes. Electrolytes used were 0.1-M Na₂SO₄ solution or 3%, 6% and 10% hydrogel. For the LSV measurement, the sweep rate was set at 10 mV s⁻¹. EIS measurements were taken in the potentiostatic mode at −1.2 V (versus saturated Ag/AgCl electrode), applying a sinusoidal voltage with an amplitude of 25 mV and frequency from 100 kHz to 0.1 Hz. EITS measurements were taken in the potentiostatic mode at −1.2 V (versus saturated Ag/AgCl electrode), applying a sinusoidal voltage with an amplitude of 25 mV and frequency of 10 kHz. CP measurements were set at −0.5 mA for 0.114-cm² electrode surface area for 5 min. Current density was calculated to be −4.4 mA cm⁻².

The calculation of E (versus RHE) is based on the following equation:

$$E(\text{vs.RHE})=E(\text{vs.Ag/AgCl})+E(\text{Ag/AgCl})+\text{PH}\times 0.059.$$

Quantification of H₂ leaking, GC and micro-CT analyses

As shown in Supplementary Fig. 11, two hollow needles were inserted into a 2-ml vial through silicone septa. Two electrodes (Ti–Pt and Ti–IrO₂) were inserted into the hollow core of the needles and any gaps were sealed with epoxy. Then, 1 ml of 0.1-M Na₂SO₄ solution or hydrogel precursor solution was injected into the vial. The vials underwent 16 h freeze and 8 h thaw, after which the system was sealed tightly. A 0.5-mA constant current was supplied to the two electrodes with Ti–Pt as the cathode and Ti–IrO₂ as the anode for 5 min to electrolyze the water. The air inside the vial was then sampled for GC runs on the Agilent 7890B system equipped with both flame ionization detector and a thermal conductivity detector with N₂ flow as the reference gas. Whole vials with different electrolytes (solution or hydrogel) were used for the micro-CT analysis (X-Cube micro-CT scanner).

Soluble H₂ test with and without the hydrogel coating

One 1.8 cm × 1.8 cm Ti mesh was cut, wired and electrodeposited with Pt. The mesh was then placed onto a 1-mm glass slide in a 58-cm² Petri dish and 14.8-ml Na₂SO₄ solution or 3% hydrogel precursor solution was poured in. For the Ti mesh–hydrogel electrochemical cell, a freeze–thaw was performed to achieve a 1.3-mm-thick hydrogel coating on the Ti mesh electrodes. A hollow acrylic model (2.2 cm × 2.2 cm × 3.6 cm) was made by laser cutting. A gas-permeable and liquid-impermeable 3M Tegaderm film was attached to one side of the acrylic model and sealed with polydimethylsiloxane. This Tegaderm chamber was then placed onto the Ti mesh leaving a 1.3-mm gap between the electrode and film. Then, 10-ml Na₂SO₄ solution was added into the Tegaderm chamber. A 6-min, –5-mA constant-current charge was supplied to the Ti–Pt electrode with saturated Ag/AgCl as the reference electrode and Pt wire as the counter electrode. To measure how much H₂ was dissolved in the solution, a H₂ water sensor (Yewhick) was placed into the Tegaderm chamber, leaving a 9-mm distance from the bottom of the Tegaderm film.

H₂ diffusion modeling inside the hydrogel

H₂ diffusion modeling was performed using COMSOL Multiphysics with a 3D finite element analysis framework to simulate the time-dependent diffusion of H₂ through the hydrogel and surrounding media. H₂ diffusion modeling used the same setup as the soluble H₂ tests. The hydrogel was modeled as a cylinder (“bubble”) with a diameter of 8.8 cm and a height of 1.3 mm. The H₂-containing bubble was modeled as a cuboid with a width of 1.8 cm, depth of 1.8 cm and height of 0.7 mm. The solution inside the Tegaderm chamber was modeled as a cube with a width of 2.2 cm, a depth of 2.2 cm and a height of 3.6 cm. The bubble was located inside the hydrogel at the bottom. The distance from the bottom of the hydrogel to the Tegaderm was 1.3 mm. The height of the H₂ bubble was calculated such that the number of moles of H₂ (9.33 μmol) in the bubble volume (at atmospheric pressure) was equal to the H₂ generated by the device after 5-mA charging for 6 min. Open-boundary conditions were used for the boundaries of the modeling domain. “Thin impermeable barrier” boundary conditions were set for the bottom of the hydrogel and the side walls of the Tegaderm chamber. The initial H₂ concentration in the bubble was set as 40.3 mol m⁻³, which is equivalent to pure H₂ concentration at atmospheric pressure. The boundaries of the model were fixed, such that as the H₂ diffused away from the bubble, the concentration reduced but the size of the bubble was fixed. The diffusion coefficient of H₂ was set as bubble and air, 0.756 cm² s⁻¹; hydrogel, 1.26 × 10⁻⁵ cm² s⁻¹; solution, 5.11 × 10⁻⁵ cm² s⁻¹. The determination of diffusion coefficient in the hydrogel is based on a previous paper on gas diffusion inside the PVA hydrogel¹⁵:

$$P=P_{\text{p}}(1-\epsilon )+P_{\text{w}}\epsilon /\mathrm{\tau },$$

where P_p and P_w are the permeabilities of the polymer matrix and water, respectively, ε is the volume fraction of water in the hydrogel membrane and τ is the tortuousity of the water passage ways, taken to be 3.92. Since the polymer content is as low as 3% and P_p generally has a much lower value than P_w, the hydrogel membrane permeability is primarily determined by the water content.

Air–liquid interfaces were set using the “Partition Condition” boundary condition. The partition coefficient value was set as 0.019 for the bubble–hydrogel interface, and 52.4 for the hydrogel–air and medium–air interfaces. These numbers were calculated using Henry’s coefficient for H₂ in water. The simulation data are plotted into a heat map using customized Python scripts.

For the H₂ diffusion of a rat heart model placed within the Tegaderm chamber (Supplementary Fig. 27), a 3D stereolithography (STL) model of a rat heart was prepared by photogrammetry. The heart was isolated from an adult rat (400–500 g), cleaned with PBS and fixed in 4% formaldehyde. The heart was then suspended on a needle on a small turntable, ~150 pictures were taken and an STL model was prepared using Agisoft Metashape. The model was then post-processed on Fusion360 by removing artefacts, cropping the model, repairing holes and remeshing to a simpler model. The STL model was then imported into COMSOL for the diffusion simulation.

Tegaderm containers for cell-free system ROS verification and CM and HaCaT cell cultures

Gas-permeable and liquid-impermeable 3M Tegaderm was used to culture the cells. Acrylic was cut into pieces with an inner empty 1 × 1 cm² square. The pieces were placed onto Tegaderm and a polydimethylsiloxane precursor was used to seal the boundary between Tegaderm and acrylic. The assembled acrylic containers were placed in a 75 °C oven and cured for 3 h. The containers were then washed with 70% ethanol overnight to remove all polydimethylsiloxane monomers and catalyst residue and then rinsed with water and isopropanol. Containers were treated with 350-W oxygen plasma for 180 s at 70 °C and coated with fibronectin/gelatin solution for 30 min for use. Neonatal rat CMs and HaCaT cells were cultured on the containers with 1 × 10⁵ cm⁻² cell density. Culture methods for CMs and HaCaT cells are available in the Supplementary Methods.

Reaction of H₂ with •OH ROS in a cell-free system

To verify the reaction of H₂ with •OH, we adopted a previously reported Fenton reaction method³. We made a PBS solution with 0.1-mM ferrous perchlorate and 1-μM 2-[6-(4′-hydroxy)phenoxy-3H-xanthen-3-on-9-yl] benzoate dye. Then, 0.5 ml of the solution was added into the cell–Tegaderm container. For the H₂-treated group, the container was placed onto a Ti mesh–hydrogel device for a total of 30 min. A −5-mA constant-current charge was supplied to the Ti mesh for 6 min with Ag/AgCl as the reference electrode and Pt wire as the counter electrode. The container was then placed under a fluorescence microscope to monitor the fluorescence intensity. Then, 10 μl of 250-μM H₂O₂ was added into the container. The control group received H₂O₂ but no H₂ treatment. For the PBS group, 10-μl PBS instead of H₂O₂ was added into the Tegaderm container.

Intracellular Fenton reaction to induce •OH ROS in cells

We used the intracellular Fenton reaction as described in a previous paper³. We charged the Ti mesh–hydrogel electrochemical cell as described above with −5 mA for 6 min with saturated Ag/AgCl as the reference electrode and Pt wire as the counter electrode. For the H₂+Cu+Vc and H₂-only groups, the Tegaderm cell containers were placed onto the Ti mesh–hydrogel device for constant H₂ delivery. For the H₂+Cu+Vc and Cu+Vc groups, we removed the culture media and preincubated CMs and HaCaT cells with 1-mM CuSO₄ for 30 min with the corresponding cell culture media containing 10% fetal bovine serum. The Cu media were removed and cells were washed once with PBS containing CaCl₂ (0.1 g l⁻¹), MgCl₂ · 6H₂O (0.1 g l⁻¹), glucose (1 g l⁻¹) and sodium pyruvate (0.036 g l⁻¹), and then exposed to 100-mM (for CMs) or 150-mM (for HaCaT cells) ascorbate (vitamin C) for 1 h in PBS, as described above. Note that Cu²⁺ is reduced by ascorbate to Cu⁺, which catalyzes the Fenton reaction to produce •OH from H₂O₂ that is spontaneously produced in the cells.

Immunofluorescence

CMs were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature, permeabilized with 0.3% Triton X-100 in PBS for 15 min at room temperature and blocked with 5% bovine serum albumin for 1 h at room temperature. Samples were incubated with primary antibodies in a blocking buffer overnight at 4 °C. Subsequently, samples were washed three times with PBS and incubated with secondary antibodies for 1 h at room temperature. The primary antibodies used were anti-NLRP3 (Biotechne, NBP2-12446) and anti-caspase-1 (Proteintech, 22915-1-AP). Stained cells were imaged using a Nikon inverted microscope with ×20 or ×60 objectives, and images were analyzed using ImageJ. To minimize background interference, the NLRP3 fluorescent intensity and caspase-1 fluorescent intensity were calculated using the following formula: Fluorescence intensity (%) = (Mean_ROI – Mean_Background)/Mean_Background × 100%. The region of interest refers to the area imaged using a ×20 objective, with uniform cell confluence.

MEA-hydrogel fabrication

The insulation coating on the end of thin wires (A-M system) was removed and weaved into the Ti meshes. The meshes were then electrodeposited with IrO₂ or Pt to serve as the anode or cathode, respectively. As shown in Supplementary Fig. 24, the Ti–Pt mesh was then immersed into a Nafion solution and was hot pressed onto one side of a Nafion 117 membrane. Ti–IrO₂ was immersed into the Nafion solution and was hot-pressed onto the other side to form the MEA. A 3% hydrogel precursor solution was poured onto the MEA device, with the Pt side facing up to create an MEA–hydrogel device with 3% hydrogel of 1.3 mm thickness on top of the cathode side. A small amount of 3% hydrogel precursor solution will leak into the IrO₂ side and form a thin layer of hydrogel on the anodic side.

Ex vivo isolated heart ischemia and reperfusion model

The rat heart isolation procedure was carried out according to our previously established protocols^41,42. In summary, adult male rats weighing between 400 g and 500 g were treated with heparin and then anesthetized with isoflurane using a bell jar. Subsequently, the heart was removed and quickly placed into a cold Hank’s balanced salt solution. A cannula was inserted into the aorta to set up for Langendorff perfusion. A heated and oxygen-enriched Tyrode’s solution, buffered with HEPES, was circulated through the aorta, utilizing a system that included a heating coil and a bubble trap (Radnoti). The heart was then positioned within a water-jacketed container, ensuring the temperature remained steady at 37 °C. Perfusion pressure was maintained between 80 mmHg and 100 mmHg. Monitoring of perfusion and LVP values was achieved through a BP-100 probe (iWorx) attached to the perfusion line and a water-filled balloon placed inside the left ventricle, respectively. For ECG recordings, needle electrodes were placed on the left ventricular wall and the aorta, grounded through the cannula and linked to a C-ISO-256 preamplifier (iWorx). All signals (perfusion, LVP and ECG) were amplified by 400 times using an IA-400D amplifier (iWorx) and interfaced with a computer using a Digidata 1550 digitizer with Clampex software (Molecular Devices). To induce an I/R injury model, the isolated heart was initially perfused with Tyrode’s solution for 1 h. Ischemic damage was simulated by stopping the perfusion buffer flow for 30 min, during which the hearts were kept in Tyrode’s solution at 37 °C, followed by a 45-min reperfusion period. To protect the heart tissue from I/R injury, at the beginning of ischemia, the heart was placed into the Tegaderm chamber (2.2 cm × 2.2 cm × 3.6 cm) as previously described in H₂ diffusion modeling with the MEA–hydrogel device underneath the chamber. The cathodic side was facing up to supply H₂ to the heart. The MEA device was charged with a 5-mA current for 6 min. For each different groups (sham, I/R and I/R+H₂), five rat hearts were used.

Multichannel ECG isochronal maps

Fabrication of a multichannel ECG electrode was adopted from the previously reported method⁴³. Recordings from the 16-channel multielectrode array on the ex vivo rat heart were analyzed by customized Python scripts. Gaussian interpolation was used to enhance readability.

Assessment of infarct size

Following 30 min of ischemia and 45 min of reperfusion in the Lan- gendorff apparatus, myocardial infarct size in the isolated rat heart was assessed using TTC staining. The heart was frozen in a −80 °C refrigerator and subsequently sectioned into 1-mm-thick transverse slices. The slices were incubated in 1% TTC solution at 37 °C for 30 min. After staining, images of the heart slices were captured immediately for analysis.

In vivo skin ischemia and reperfusion model

Ten C57BL/6 mice (8–10 weeks) mice were randomly classified into two groups (control group and H₂ group). They were anesthetized with isoflurane. The dorsal hair was shaved and the area was cleaned with 70% ethanol. The skin was gently pulled up and placed between two round ceramic magnetic plates (12 mm in diameter and 5 mm thick) with an average weight of 2.4 g and 1,000-G magnetic force (Magnetic Source) for 6 h and then removed. The resultant “pinch” procedure was designed to leave a 0.25-cm skin bridge between the two magnets. After removing the ceramic magnetic plates, for the H₂ group, the H₂ generation device, with the Tegaderm film, MEA–hydrogel, PI film and PCB, were placed onto one skin wound. The PCB was programmed to charge the MEA–hydrogel device with a 5-mA current for 6 min every 12 h for a total of three times. Mice were only anesthetized during the 6-min electrochemical H₂-charging period. For the control group, after removing the ceramic magnetic plates, the Tegaderm film is placed on the skin wound. Digital images were captured on day 0, day 2, day 4, day 6 and day 7. On day 7, a thermal camera was used to capture the temperature distribution of the mouse ulcers. Skin tissue was collected for histological analysis and immunohistochemical analysis.

Histology and immunohistochemistry

For tissue histological analysis, H&E, CD4, CD31 and NQO1 (1:1,000; Proteintech 67240-1-Ig) histological tissue sections were prepared by the Human Tissue Resource Center at the University of Chicago. Tissue sections of the mice dorsal skin were prepared and stained with NQO1, CD4, CD31 and H&E. All tissue histological results were analyzed using the CaseViewer software (3DHISTECH). The CD4 statistic result was calculated from sampling in a random 400 μm × 400 μm square under the wound. The CD31 statistic result was calculated from sampling in a random 800 μm × 800 μm square under the wound.

Data processing and statistics

Data analysis was performed with Python scripts using the CV2, NumPy, Matplotlib, SciPy and Pandas libraries. Plotting was performed with GraphPad Prism 9, Origin and Python Matplotlib. Statistics were calculated using GraphPad Prism 9, unless otherwise clarified. Images were processed using ImageJ software. Before hypothesis testing, the Shapiro–Wilk test was used to assess the normality of all datasets involving biological replicates (n = 4–5). Datasets that passed the normality test (P > 0.05) were analyzed using parametric tests, including two-sample two-tailed t-tests and one-way analysis of variance (ANOVA). A value of P < 0.05 was considered statistically significant; ns indicates no statistical significance. All error bars indicate the standard deviation, unless otherwise stated. No statistical methods were used to predetermine the sample sizes. For all animal studies, animals were randomly assigned to various experimental groups. After data collection, all data were analyzed quantitatively without the knowledge of group allocation, after which the samples were matched to their group for data visualization and statistical analysis. No data were excluded from the analysis.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary Material

The online version contains supplementary material available at https://doi.org/10.1038/s44286-025-00259-x.

Data availability

The research findings presented in this study are supported by data included in the Article and Supplementary Information. Source data are provided with this paper and are publicly available via GitHub at https://github.com/wenli-web/Hydrogen-evolution-in-hydrogel-electrochemical-cell/tree/main.

Code availability

Scripts used for data analysis and Bluetooth user interface in this study are available via GitHub at https://github.com/wenli-web/Hydrogen-evolution-in-hydrogel-electrochemical-cell/tree/main.