Creation of anaerobic microenvironments by photosystem I in porous glass nanopores enables photoinduced H₂ evolution under aerobic conditions

Abstract

Under aerobic conditions, artificial photosynthesis devices must maintain a steady electron flow toward H₂ evolution. Dissolved O₂ competes for photogenerated electrons, diverting them into oxygen reduction pathways instead of hydrogen production. In this study, we immobilized photosystem I (PSI) in a porous glass plate (PGP) with 50 nm diameter through-thickness nanopores and analyzed electron transfer using transient absorption kinetics. A kinetic model resolves three processes following charge separation: charge recombination with rate constant k₁, electron loss from the terminal iron–sulfur cluster F_B⁻ in the reduced state to external oxidants, including O₂, with rate constant k₂, and re-reduction of the oxidized reaction center chlorophyll P700 by an external electron donor with rate constant k₃. Remarkably, illumination of a device in which PSI is immobilized in the PGP caused the rate constant k₂ to decrease from 6.5 s⁻¹ at 0 min to 1.2–1.3 s⁻¹ after 7–10 min. This shows that photochemical O₂ consumption by PSI decreases the intrapore O₂ concentration, suppressing electron transfer from F_B⁻ to oxidants. Combined with diffusion-limited O₂ delivery, a local low-O₂ microenvironment is formed within the nanopores over time. Extending this concept to a PSI–Pt nanoparticle assembly (PSI–PtNP), we observed sustained light-driven H₂ evolution under aerobic conditions in PGP-immobilized PSI–PtNP, whereas the bulk solution showed much lower activity under aerobic conditions. The system achieved a turnover number of 400 ± 100 mol H₂ per mol PSI and an O₂ tolerance of 44 ± 15%. These findings establish a framework for reaction field engineering in nanopores and guide the design of oxygen-tolerant systems.

---

Introduction

Artificial photosynthesis has attracted remarkable attention as a strategy for producing solar fuels, particularly H₂, from water using sunlight.^1,2 A key challenge for practical deployment lies not only in achieving high activity under inert atmospheres but also in selectively directing photogenerated high-energy electrons toward the desired reaction while suppressing deactivation and competing pathways under aerobic conditions (Fig. 1a).^3–5 In natural photosynthesis, the light reactions proceed through the coordinated action of photosystem II (PSII) and photosystem I (PSI) (Fig. 1b), both embedded in the thylakoid membrane.^6–8 PSII oxidizes water to generate electrons, which are relayed via quinones and the cytochrome b₆f complex to plastocyanin (PC) or cytochrome c₆ (cyt c₆), and subsequently to PSI. PSI catalyzes light-driven electron transfer from PC/cyt c₆ on the lumenal side to ferredoxin on the stromal side, thereby supplying reducing equivalents for downstream metabolism. Upon photoexcitation of the reaction center chlorophyll pair P700, an electron is transferred sequentially to A₀, A₁, and the [4Fe–4S] clusters (F_X, F_A, and F_B), generating the charge-separated state P700⁺/F_B⁻ (Fig. 1c).^9–11 Because charge recombination between P700⁺ and the reduced forms of the intermediate cofactors is approximately 1000 times slower than the forward electron transfer, the photochemical quantum yield for formation of P700⁺/F_B⁻ from the ground state is essentially unity.¹²

Fig. 1 (a) Illustration of oxygen inhibition in photocatalytic H₂ evolution by O₂. (b) Structure of PSI (PDB ID 1JB0).¹⁰ (c) Electron transfer chains in the reaction center. (d) Photographs of the surface and cross-section of the PGP, and a schematic illustration of the nanopore distribution in the PGP.

Thus, PSI combines efficient light harvesting with highly efficient charge separation and has been widely employed in biohybrid and semi-artificial photochemical systems.^13,14 On this basis, PSI conjugates have been extensively investigated for light-driven H₂ production by coupling PSI with proton reduction catalysts, including hydrogenases,^15,16 metal complexes,^17,18 and Pt-based^19–22 catalysts. In these systems, cyt c₆ (or PC) typically serves as the electron donor. Although high activities have been demonstrated, many studies have been carried out under anaerobic conditions to avoid oxygen inhibition. Two strategies have been proposed to mitigate oxygen inhibition. One approach is to endow the hydrogen evolution catalyst itself with oxygen tolerance. Recently, Rumbaugh et al. reported that a PSI–hydrogenase complex, in which the O₂-tolerant [FeFe] hydrogenase was fused to the stromal-side PSI subunit PsaE, retained 2.3% of its H₂ evolution activity under aerobic conditions relative to that under anaerobic conditions (i.e., oxygen tolerance).²³ Another approach is to engineer the reaction environment to create a localized low-oxygen microenvironment. In recent microreactor and cell coating systems using living algal cells, low-oxygen regions can form near cells or within aggregates even under air, enabling light-driven H₂ production by activating hydrogenase and suppressing its inactivation, thereby sustaining and enhancing H₂ generation.^24,25

Creating localized low-oxygen microenvironments under artificial (cell-free) conditions is advantageous because it can be combined with catalyst-level oxygen tolerance. We have previously shown that immobilizing photosensitizers/hydrogenase (or formate dehydrogenase) or PSI–Pt nanoparticle conjugates (PSI–PtNP) within the porous glass plates (PGPs) enables light-driven H₂ or formate production under aerobic conditions.^26–28 PGPs^26–29 are optically transparent in the visible region and contain through-thickness nanopores with high internal surface area, allowing immobilization of photochemical components within a light-transmitting solid support (Fig. 1d). To advance strategies that mitigate oxygen inhibition by forming localized anaerobic microenvironments, quantitative analysis of PGP-based photoreaction systems is essential. Here, we combine time-resolved spectroscopy with biohybrid device construction in a nanoporous platform to elucidate the mechanism of microenvironment formation. PSI is immobilized within the nanopores of PGPs, and light-induced electron transfer branching is quantified by transient absorption spectroscopy (TAS) and analyzed using a kinetic framework that resolves charge recombination,²² electron transfer to external acceptors including O₂,³⁰ and reduction of oxidized P700 by external donors.³¹ By comparing aerobic and anaerobic conditions and tracking illumination-dependent changes under aerobic conditions, we clarify how confinement alters the external acceptor pathway in a nanoporous environment. We then extend this approach to a PSI–PtNP system immobilized in a PGP to evaluate light-driven H₂ evolution, directly comparing immobilized and solution-phase systems while optimizing immobilization for uniform distribution of PSI–PtNP and cyt c₆ within the porous matrix.

Experimental section

Purification of PSI

PSI was isolated from the cyanobacterial membranes of T. vulcanus.³² Purified PSI was solubilized in 40 mmol per L 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)–NaOH (pH 7.8) containing 50 mmol per L NaCl and 0.03% (w/v) n-dodecyl-β-D-maltoside (DM). The PSI concentration was determined via UV-vis absorption spectroscopy using an extinction coefficient (ε) of 16.4 L µmol⁻¹ cm⁻¹ (PSI monomer equivalent)³³ with a double-beam spectrophotometer (UV-1800 PC, SHIMADZU, Japan).

Preparation of a PGP

A PGP, 1 mm × 4 cm × 4 cm, with an average pore diameter of 50 nm was prepared by acid leaching of phase-separated borosilicate glass, as previously described.^34,35 The parent glass (62.5 SiO₂, 28.3 B₂O₃, and 9.2 Na₂O; wt%) was melted in a platinum crucible and heat-treated at 610 °C for 32 h to induce phase separation. The resulting glass was then leached in 1 N H₂SO₄ at 90 °C for 2 days. The pore diameter, pore volume, and specific surface area of the PGP were determined via mercury intrusion porosimetry, as described previously.²⁶

Preparation and characterization of PtNPs

PtNPs were prepared according to previously reported methods.³⁶ The particle size of the PtNPs was determined by dynamic light scattering (DLS) using an ELSZ-2000 ZS instrument (Otsuka Electronics, Japan) and by transmission electron microscopy (TEM) using a JEM-1010 microscope (JEOL, Japan).

Preparation of cyt c₆

cyt c₆ from Synechocystis sp. PCC 6803 was cloned and heterologously expressed in E. coli according to previous reports and purified by Ni²⁺ affinity chromatography.^37–39

Preparation of PSI–PtNP

PSI–PtNP was prepared according to previous reports.²⁰ PSI (1 µmol L⁻¹) and PtNPs (2 µmol L⁻¹) were mixed in buffer (40 mmol per L HEPES–NaOH, pH 7.8, 50 mmol per L NaCl, 0.03% [w/v] DM) and stirred at 4 °C for 2 h in the dark. Unbound PtNPs were removed using an Amicon Ultra-50 kDa filter (Merck).

Negative-stain TEM imaging and two-dimensional (2D) classification of PSI and PSI–PtNP

Negative-stain TEM was used to examine the morphology of PSI and PSI–PtNP. Samples were applied to TEM grids, negatively stained with 2% (w/v) aqueous ammonium molybdate for 30 s, and then imaged. For PSI, single-particle analysis was performed using Scipion,⁴⁰ with particle picking and extraction carried out in Xmipp.⁴¹ The extracted particles were subjected to 2D classification in RELION⁴² using only particle images with Z-scores of <3.

Transient absorption spectroscopy

TAS measurements of PSI were conducted in a measurement buffer (40 mmol per L HEPES–NaOH, pH 7.8, containing 0.03% (w/v) DM, 10 mmol per L ascorbic acid and 0.1 mmol per L 2,6-dichloroindophenol [DCIP]) using a 1 cm pathlength quartz cuvette. For PGP measurements, the plate was incubated at 4 °C for 1 h in the same buffer containing 5.8 µmol per L PSI. It was rinsed and placed in a 2 mm pathlength quartz cell filled with the measurement buffer. Anaerobic samples were prepared by keeping the samples overnight in the dark in an N₂-filled glovebox. TAS was performed with a 700 nm probe from a halogen lamp and excitation using a xenon flash lamp (1 µs pulse) filtered to 400–500 nm at 0.2 Hz. The transmitted probe intensity was detected with a photomultiplier tube through a 700 nm filter. Signals were recorded with a digital oscilloscope at 10 µs time resolution up to 4 s, averaged over 128 shots, and fitted to multiexponential functions (OriginLab).

Electron transfer kinetics

The redox state of the P700/F_B pair is represented as [P700–F_B]. After charge separation, PSI is generated in the state [P700⁺–F_B⁻].

Here, [P700⁺–F_B⁻] and [P700⁺–F_B] are time-dependent concentrations. This state relaxes through two competing pathways. First, the electron on F_B⁻ can return to P700⁺, leading to charge recombination; the corresponding rate constant is defined as k₁ (s⁻¹). Second, the electron on F_B⁻ can be transferred to an external electron acceptor; the rate constant for this process is defined as k₂ (s⁻¹). Reported external acceptors for PSI include dissolved oxygen and oxidized ascorbic acid.^30,43 Electron transfer to the external acceptor converts [P700⁺–F_B⁻] to [P700⁺–F_B]. Finally, P700⁺ is re-reduced by ascorbate through DCIP-mediated electron transfer,³¹ with a rate constant k₃ (s⁻¹), returning the system to [P700–F_B].

In principle, [P700–F_B⁻] can be formed when P700⁺ is reduced by the donor-side pathway before F_B⁻ is oxidized. Under the present conditions, this process is assigned to the slow kinetic component with a rate constant of approximately 0.5 s⁻¹ (Table S1). In contrast, electron transfer from F_B⁻ to external electron acceptors proceeds with a rate constant of approximately 6–7 s⁻¹. Thus, donor-side reduction of P700⁺ is more than one order of magnitude slower than electron transfer from F_B⁻ to external acceptors. Therefore, the simplified kinetic model focuses on the main decay pathways of [P700⁺–F_B⁻], namely charge recombination and electron transfer to external acceptors.

Based on this scheme, we formulated rate equations for the relevant PSI redox states, including the P700⁺-containing species [P700⁺–F_B⁻] and [P700⁺–F_B], as well as [P700–F_B⁻]:

The total concentration of PSI in the P700⁺ state is given by:

[P700⁺] = [P700⁺–F_B⁻] + [P700⁺–F_B] (4)

Substituting the analytical solutions of [P700⁺–F_B⁻] and [P700⁺–F_B] (derived in the SI, eqn (S1)–(S8)) into eqn (4) yields:

where [P700⁺–F_B⁻]₀ indicates the initial concentration after charge separation, and [P700⁺–F_B⁻], [P700⁺–F_B], and [P700⁺] are time-dependent concentrations. Eqn (5) shows that the decay of the PSI population with P700⁺ after charge separation is described by a bi-exponential function. To experimentally monitor this time dependence, we performed TAS by detecting the absorbance change and converting it to [P700⁺] using the Beer–Lambert law. Absorbance changes at 700 nm provide a clear P700/P700⁺ difference signal and enable monitoring of the P700⁺ population in PSI. P700⁺–P700 difference spectra of PSI have been reported previously, and changes around 700 nm are widely used to follow the P700/P700⁺ redox transition.⁴⁴

Light-driven hydrogen evolution

The light source for the light-driven H₂ evolution experiments was a solar simulator (HAL-320, Asahi Spectra). After calibration, an RG610 filter (HOYA), a heat-cut filter (HA50, HOYA), and a neutral-density filter (ND30, HOYA) were installed in the light path. The emission spectrum of the light source was measured using a fiber-optic spectrometer (SILVER-NOVA, StellarNet). Under these conditions, the photon flux density in the 400 to 700 nm range was 230 µmol m⁻² s⁻¹ and the irradiance was 44 W m⁻². For the solution system, 3 mL of a PSI–PtNP suspension (30 nmol per L PSI–PtNP, 100 mmol per L ascorbic acid, 0.03% (w/v) DM, 4 µmol per L cyt c₆, 20 mmol per L 2-(N-morpholino)ethanesulfonic acid (MES)–NaOH, pH 6.2) was placed in a 5 mL vial, sealed with a butyl septum, and stirred. Anaerobic samples were prepared by bubbling N₂ for 45 min. For immobilized experiments, a PGP was incubated in a PSI–PtNP solution (2.9 µmol per L PSI–PtNP in 20 mmol per L MES–NaOH, pH 6.2, containing 0.03% (w/v) DM) at 4 °C for 1 h in the dark, rinsed, and then incubated overnight in a cyt c₆ solution (100 µmol per L cyt c₆, 100 mmol per L ascorbic acid, 20 mmol per L MES–NaOH, pH 6.2, 50 mmol per L NaCl). After rinsing, the plate was immersed in 3 mL of reaction solution (100 mmol per L ascorbic acid, 4 µmol per L cyt c₆, 20 mmol per L MES–NaOH, pH 6.2) in an 8 mL vial and sealed. Anaerobic samples were prepared by bubbling N₂ for 45 min. H₂ was quantified by withdrawing 100 µL of headspace gas with a gastight syringe and analyzing it by GC (GC-2010, Shimadzu). Turnover number (TON) was calculated from the amount of H₂ produced after 10 h of illumination. TON and O₂ tolerance were calculated using eqn (S9) and (S10).

Results and discussion

TAS measurements of PGP-immobilized PSI

From the UV-vis absorption spectrum of PGP-immobilized PSI (Fig. 2a), the amount of immobilized PSI was estimated, giving an intrapore PSI concentration of 0.57 µmol L⁻¹, which corresponds to 0.20 µmol PSI per m² (Fig. S1). In this study, PSI was used for TAS measurements to simplify the interpretation of P700⁺ decay. In PSI–PtNP, electron transfer from F_B⁻ to PtNPs and possible back electron transfer from PtNPs to PSI can occur in addition to electron transfer from F_B⁻ to dissolved O₂. These processes overlap in the 700 nm P700⁺ transient absorption signal and cannot be separated by the present analysis. Therefore, PSI was used as a simplified model to evaluate the effect of the PGP nanopore environment on the external electron-acceptor pathway.

Fig. 2 (a) Photographs of the PGP sample after PSI immobilization, showing the surface and side views. (b) Kinetic scheme describing light-induced charge separation and subsequent branching processes in PSI: charge recombination of the [P700⁺–F_B⁻] state (rate constant k₁), electron transfer from F_B⁻ to external electron acceptors (k₂), and reduction of oxidized P700⁺ by an external electron donor (k₃).

Next, for analyzing the electron transfer pathways from the terminal acceptor F_B following charge separation between the primary donor P700 and F_B in the PSI reaction center, we adopted the kinetic scheme shown in Fig. 2b. Based on eqn (5), the transient absorption decay time courses monitored at 700 nm were fitted to obtain the rate constants k₁, k₂, and k₃, enabling a comparison between PSI in solution and PSI immobilized in a PGP. Furthermore, to evaluate the effect of light irradiation on the kinetics of PSI in solution and PGP-immobilized PSI, respectively, under aerobic conditions, samples were pre-irradiated for 0–10 min prior to TAS measurements (Fig. S2 and S3). The resulting rate constants are summarized in Table 1.

Table 1 Rate constants k₁ and k₂ obtained by fitting the transient absorption time courses of PSI in solution and PGP-immobilized PSI. Rate constant k₃ is shown in Table S1

Condition	Pre-irradiation time (min)	Rate constant (s^−1)
		k1		k2
		Solution	PGP	Solution	PGP
Aerobic	0	6.6	3.6	6.2	6.5
	1	6.8	3.0	6.3	4.8
	2	—	3.6	—	2.4
	3	6.5	3.6	6.2	1.7
	5	6.7	3.6	6.3	1.3
	7	—	3.6	—	1.3
	10	6.7	3.8	6.2	1.2
Anaerobic		6.4	3.8	1.0	2.3

---

For PSI in solution under aerobic conditions (Fig. 3a), k₁ = 6.6 s⁻¹ and k₂ = 6.2 s⁻¹ were obtained (Table 1), and k₃ = 0.49 s⁻¹ (Table S1). Previous work has shown that the charge-separated state P700⁺–F_B⁻ undergoes charge recombination with a time constant of 65 ms.²² This corresponds to the rate constant of approximately 15 s⁻¹, and the k₁ values obtained in this study are of the same order of magnitude. For PSI in solution, the TAS kinetic traces were unchanged across the tested pre-irradiation times for 0–10 min (Fig. 3a). The fitted rate constants remained constant, with k₁ = 6.4–6.8 s⁻¹, k₂ = 6.2–6.3 s⁻¹, and k₃ = 0.49–0.51 s⁻¹ over 0–10 min (Fig. 3c, d, Tables 1, and S1). Thus, under these conditions, pre-irradiation does not affect electron transfer kinetics of PSI in solution. This negligible time dependence in solution can be explained quantitatively by the larger O₂/PSI molar ratio and mechanistically by rapid O₂ resupply under homogeneous conditions. For the solution system, the PSI concentration was 180 nmol L⁻¹. Using an air-saturated O₂ concentration of 253 µmol L⁻¹ at 25 °C, the initial O₂/PSI molar ratio in solution was approximately 1.4 × 10³. In addition, O₂ consumed by PSI in solution can be rapidly replenished from the surrounding solution and gas phase. Consistently, in a separate O₂ sensor measurement under aerobic illumination, the dissolved O₂ concentration in PSI solution decreased by only 15–18 µmol L⁻¹ and then reached a plateau (Fig. S4). This decrease corresponds to only 6–7% of the initial dissolved O₂ concentration, indicating that sufficient O₂ remained in solution and that the local O₂ concentration around PSI did not decrease enough to change k₂ detectably. In contrast, PGP-immobilized PSI exhibited a pronounced illumination history dependence (Fig. 3b), with a progressive decrease in k₂ (Fig. 3c). Specifically, k₂ decreased from 6.5 s⁻¹ at 0 min to 4.8 s⁻¹ at 1 min, 2.4 s⁻¹ at 2 min, and 1.7 s⁻¹ at 3 min, followed by a slower decline to 1.3 s⁻¹ at 5 min and a near plateau of 1.2–1.3 s⁻¹ at 7–10 min (Fig. 3c and Table 1). By comparison, k₁ remained within 3.0–3.8 s⁻¹ over the same interval (Fig. 3d and Table 1), indicating that light pre-irradiation mainly perturbs the pathway represented by k₂ in the PGP system. Consistent with this trend, the branching fraction, k₂/(k₁ + k₂) × 100, decreased from 64% at 0 min to ∼25% at 5–10 min for PGP-immobilized PSI, whereas it remained essentially constant at 48–49% for PSI in solution. These results indicate that the contribution of external acceptor pathways diminishes with illumination time only in the immobilized system. Because k₂ reflects electron transfer to external acceptors, the progressive decrease in k₂ for PGP-immobilized PSI suggests that the external acceptor branch becomes less accessible.

Fig. 3 Pre-irradiation time dependence of transient absorption decay in PSI. (a) Transient absorption decay time courses of PSI in solution recorded after 0, 3, and 10 min of pre-irradiation. (b) Corresponding time courses for PGP-immobilized PSI recorded after 0, 2, 3, and 10 min of light pre-irradiation. Gray lines show experimental results. Colored lines represent the best–fit curves obtained using eqn (5). Fitting parameters are shown in Tables 1 and S1. (c) Time dependence of k₂. The red line represents the best–fit curve obtained using eqn (7) and (8). Fitting parameters are shown in Table 2. (d) Time dependence of k₁.

Next, transient absorption kinetics were measured under anaerobic conditions. Under anaerobic conditions, both PSI in solution and PGP-immobilized PSI exhibited decay profiles similar to those observed for PGP-immobilized PSI after pre-irradiation under aerobic conditions (Fig. 4). Fitting the time courses measured under anaerobic conditions yielded smaller k₂ values than those under aerobic conditions (Table 1), indicating that dissolved O₂ is the primary external electron acceptor under aerobic conditions. Even under anaerobic conditions, k₂ was not 0, suggesting that other electron acceptors, such as oxidized ascorbic acid, may contribute.⁴³ Because the light intensity used for pre-irradiation was sufficiently high and not rate-limiting for the photoreaction, we assumed that PSI effectively operated in an excited-state regime. Under this approximation, the intrapore dissolved oxygen concentration, [O₂]_pore, is expected to follow first-order decay (eqn (6) and (7)):

where [PSI] is the PSI concentration (0.57 µmol L⁻¹ in this study) and [O₂]_bulk is the dissolved oxygen concentration in bulk solution (253 µmol L⁻¹ at 25 °C),⁴⁵ k_i is an apparent first-order rate constant for oxygen transport from a bulk solution into PGP nanopores, and k_O2 is an effective second-order rate constant for PSI-driven O₂ reduction (L µmol⁻¹ s⁻¹). Assuming that the oxygen concentration in the PGP nanopores before pre-irradiation is the same as that in the bulk solution, eqn (6) gives the following expression for the nanopore oxygen concentration as a function of illumination time.

Fig. 4 Transient absorption time courses of PSI under anaerobic conditions. (a) PSI in solution. (b) PGP-immobilized PSI. Gray lines show experimental results. Black lines represent the best–fit curves using eqn (5). Fitting parameters are shown in Tables 1 and S1.

The rate constant k₂ for electron loss from PSI to external electron acceptors can be decomposed into electron loss to O₂ and nonoxygen external acceptors, X:

k₂ = k_O2[O₂] + k_x[X] (8)

Here, [X] and k_x denote the concentration of the external electron acceptor and the rate constant for electron transfer to X, respectively; because X is present in large excess in our experiments, both can be treated as time-independent constants. By substituting [O₂]_pore from eqn (7) into [O₂] in eqn (8) and fitting the illumination time dependence of k₂ measured for PGP-immobilized PSI (Fig. 3c) with the constraint k_i ≥ 0, we obtained k_x[X] = 1.4 s⁻¹, which falls within the range of k₂ = 1–2.3 s⁻¹ under anaerobic conditions (Table 1). In this fit, k_O2[PSI] = 0.012 s⁻¹ and k_i = 0 s⁻¹ (Table 2). Because k_i cannot be assumed to be 0, this result suggests that k_O2[PSI] is significantly faster than k_i.

Table 2 Fitting parameters of the k₂ time course of PGP-immobilized PSI

Parameter	Value [unit]
kO2	0.021 [L (µmol)^−1 s^−1]
ki	0 [s^−1]
kx[X]	1.4 [s^−1]

---

Therefore, the observation that k₂ slowed upon pre-illumination only in PGP-immobilized PSI indicates that the rate of PSI-driven photoreduction of oxygen is faster than the rate of oxygen supply into the PGP nanopores. The estimated intrapore PSI concentration in PGP was 0.57 µmol L⁻¹, giving an initial O₂/PSI molar ratio of approximately 4.4 × 10². Thus, even before illumination, the O₂/PSI molar ratio in the PGP nanopores was approximately threefold lower than that in homogeneous solution. Furthermore, in the homogeneous solution system, O₂ supply from the gas phase to the liquid phase can largely compensate for PSI-driven O₂ consumption, resulting in only a small apparent change in the dissolved O₂ concentration. By contrast, O₂ supply from the bulk liquid phase into the liquid confined within the nanopores is sufficiently restricted relative to the O₂ consumption process. Therefore, immobilization of the photochemical reaction system in PGP can substantially suppress O₂ supply from the O₂-rich external environment to the local environment surrounding the O₂-sensitive photochemical components.

Photoreduction of O₂ by PSI may generate reactive oxygen species, such as superoxide and H₂O₂.³ These ROS may contribute to deactivation of PSI, cyt c₆, or PtNPs during prolonged irradiation. However, ROS were not directly quantified in this study, and their concentration and specific effects inside the PGP nanopores remain unclear. Direct quantification of ROS within the nanopore environment will be an important subject for future work. Previous studies have reported that the lifetime of reduced methyl viologen, a highly oxygen-sensitive species, is markedly prolonged upon immobilization within PGP nanopores and reaches on the order of hours compared with that in solution, which suggests that oxygen transport from the bulk solution into the nanopores is diffusion-limited.²⁶ Thus, intrapore oxygen consumption during illumination may outpace oxygen supply from the bulk solution, creating a transport bottleneck and causing k₂ to decrease. Conversely, in bulk solution, any local O₂ consumption would be rapidly compensated by diffusion and gas–liquid equilibration, consistent with the negligible time dependence of k₂.

Light-driven H₂ evolution

PtNPs with an average diameter of approximately 2 nm were characterized using TEM and DLS, and negative-stain TEM supported their binding to PSI trimers (Fig. S5–S10). UV-vis spectral overlay analysis after removal of unbound PtNPs supported a PtNP/PSI ratio of 1 : 1 on a PSI monomer-equivalent basis, corresponding to approximately three PtNPs per PSI trimer (Fig. S11). The PSI–PtNP interaction is considered to be mainly electrostatic because the PtNP surface is expected to be negatively charged owing to thiomalate-derived carboxylate groups. This interpretation is consistent with previous PSI–PtNP studies showing that carboxylate-covered PtNPs bind to the stromal-side electron-acceptor region of PSI.^20,21 Therefore, the TEM images are used as qualitative evidence supporting PSI–PtNP binding, whereas the PtNP/PSI ratio was estimated from UV-vis spectral overlay analysis. Absorption spectral analysis showed that the immobilization amounts of PSI–PtNP and cyt c₆ in PGP were 17 ± 0.5 µmol PSI–PtNP per m² and 0.31 ± 0.04 mmol cyt c₆ per m², respectively (Fig. S12), which correspond to an intrapore PSI–PtNP concentration of 0.24 ± 0.01 µmol L⁻¹. The aerobic H₂-evolution experiments were performed in sealed vials with an air headspace. The O₂ amount in the headspace was approximately 1.9 × 10⁵ equivalents relative to PSI–PtNP, indicating that O₂ was present in large excess under the sealed aerobic conditions. Under aerobic conditions, photoinduced H₂ evolution from PGP-immobilized PSI–PtNP continued for 30 h, whereas in the solution system, the amount of evolved H₂ was below the detection limit (Table 3 and Fig. S13). Even when stirring was applied to increase the reaction rate in the solution system, photoinduced H₂ evolution ceased after 10 h (Fig. S13). Under aerobic conditions, TON during 10 h of light irradiation was 400 ± 100 mol H₂ per mol PSI for PGP-immobilized PSI–PtNP and 170 ± 20 mol H₂ per mol PSI for the stirred solution system (Table 3 and Fig. S13). These values, relative to the TON under anaerobic conditions, were 44 ± 15% for PGP-immobilized PSI–PtNP and 5.0 ± 0.8% for the stirred solution system (Table 3 and Fig. S13). This result indicates that the reaction system in the PGP nanopores exhibited ninefold higher oxygen tolerance than the solution system (Table 3). Under anaerobic conditions, the TON = 900 ± 200 mol H₂ per mol PSI for PGP-immobilized PSI–PtNP was comparable to 800 ± 200 mol H₂ per mol PSI of the unstirred solution system, suggesting that immobilization in the porous glass plate does not impose a substantial additional kinetic limitation under quiescent conditions. In contrast, stirring markedly increased the activity in bulk solution, indicating that H₂ evolution under unstirred conditions is limited by mass transport and/or diffusional encounters among PSI–PtNP, cyt c₆, and redox components (Table 3 and Fig. S13).

Table 3 TON during 10 h of light irradiation and O₂ tolerance for light-driven H₂ evolution by PSI–PtNP in solution and in the PGP under anaerobic and aerobic conditions

System	Condition	TON (mol H2 per mol PSI)
		Anaerobic	Aerobic	O2 tolerance (%)
a N.D., not detected.b N.A., not applicable.
PGP	Without stirring	900 ± 200	400 ± 100	44 ± 15
Solution	Without stirring	800 ± 200	N.D.^a	N.A.^b
	With stirring	3400 ± 400	170 ± 20	5.0 ± 0.8

---

Compared with the previously reported PGP50 system²⁸ in which PSI–PtNP and cyt c₆ were co-immobilized, the PGP system developed here showed 2.4-fold higher activity under anaerobic conditions and 6.7-fold higher activity under aerobic conditions. O₂ tolerance also increased from 16% to 44 ± 15%, corresponding to a 2.8-fold improvement. In conventional co-immobilization based on simultaneous adsorption, cyt c₆ tends to localize deeper in the nanopores, whereas PSI–PtNP accumulates near the PGP surface;²⁸ conversely, sequential immobilization of PSI–PtNP followed by cyt c₆ resulted in a more homogeneous intrapore distribution (Fig. S14), which suggests more frequent PSI–cyt c₆ encounters.

In this study, we integrated the TAS of PSI immobilized in PGPs under aerobic conditions with the light-driven H₂ evolution results to discuss the mechanism by which the PSI–PtNP/PGP system retains activity under aerobic conditions (Fig. 5). Initially, O₂ is present within the nanopores. During photoirradiation, photogenerated reducing equivalents are consumed by O₂ reduction in the pore space, whereas O₂ influx from the bulk is diffusion-limited within the porous matrix. Consequently, the intrapore O₂ concentration is expected to decrease during irradiation, establishing an effectively anaerobic microenvironment within the pores. Under such intrapore conditions, competitive electron loss to O₂ is mitigated, allowing productive electron transfer to proceed from cyt c₆ to PSI–PtNP, thereby enabling H₂ evolution even under aerobic conditions.

Fig. 5 Proposed mechanism for oxygen-tolerant light-driven H₂ evolution in PGP-immobilized PSI–PtNP.

Previous studies in which photochemical reaction systems were introduced into PGPs have shown that reducing the pore diameter to approximately 20 nm limits the immobilization of photosynthetic membrane protein complexes, whereas reducing the plate thickness to approximately 0.2 mm decreases the amount of immobilized molecules.^26,29 Therefore, based on these previous studies, we used a PGP with a thickness of 1 mm and a pore diameter of 50 nm in this study. These dimensions were appropriate because the 50 nm pores were sufficiently large to accommodate PSI–PtNP and cyt c₆ without severely limiting immobilization, while being sufficiently small to restrict O₂ supply from the bulk solution. Thus, this PGP design enabled the formation of a local low-O₂ environment around the photochemical reaction system, leading to the observed oxygen-shielding effect.

However, we do not claim that this effect is unique to the PGP. Rather, the PGP served as a useful model platform for elucidating the mechanism by which light-driven H₂ evolution can be enabled under aerobic conditions by lowering the local O₂ concentration around the photochemical reaction system. In fact, recent microreactor and cell-coating systems using living algal cells have shown that low-O₂ regions can form near cells or within aggregates, thereby enabling light-driven H₂ production.^24,25 A key requirement common to our study and these previous studies is that the platform and the photochemical reaction system should be designed so that local O₂ consumption exceeds O₂ resupply from the O₂-rich external environment. In this respect, the PGP is a suitable porous material because it simultaneously satisfies the requirements of visible-light transparency, an appropriate pore diameter that is neither too large nor too small, a large internal surface area, protein compatibility, and an appropriate plate thickness.

Conclusions

In this study, PSI was immobilized within the 50 nm nanopores of a porous glass plate and analyzed via TAS. Light-driven H₂ evolution by PSI–PtNP was also compared under aerobic and anaerobic conditions. The transient absorption kinetics were analyzed using a scheme comprising charge recombination of the charge-separated state P700⁺–F_B⁻ with rate constant k₁, electron transfer from F_B⁻ to external electron acceptors with k₂, and reduction of P700⁺ by an external electron donor with k₃. Under aerobic conditions with pre-irradiation, the PSI solution system kinetics demonstrated negligible time dependence because the rate of O₂ supply from the gas phase to the liquid phase competes with the rate of PSI-driven O₂ consumption. Whereas the immobilized system showed a clear minute-scale decrease in k₂ from 6.5 s⁻¹ at 0 min to 1.2–1.3 s⁻¹ after 7–10 min, indicating a reduction in electron loss rate to external electron acceptors (primarily O₂), as reflected by a decrease in k₂/(k₁ + k₂) × 100 from 64% to ∼25%. These results suggest that the rate of O₂ consumption greatly exceeded the rate of O₂ supply from the bulk aqueous phase to the liquid phase inside the nanopores, resulting in a gradual decrease in the intrapore O₂ concentration and, consequently, the formation of a local anaerobic microenvironment within the nanopores. For light-driven H₂ evolution, the solution system exhibited a loss of activity under aerobic conditions, whereas the PSI–PtNP immobilized PGP system maintained H₂ evolution even under aerobic conditions, achieving TON = 400 ± 100 mol H₂ per mol PSI. The O₂ tolerance was 44 ± 15%, corresponding to a ninefold improvement over the stirred solution system.

Hence, these results provide quantitative kinetic evidence that PSI immobilized in a PGP progressively decreases the intrapore O₂ concentration during illumination, thereby creating locally anaerobic microenvironments within PGP nanopores. We further reveal that immobilization of PSI–PtNP enables light-driven H₂ evolution activity under aerobic conditions. Overall, these findings provide practical design guidelines for controlling intrapore oxygen and electron transfer pathways, facilitating PSI-driven hydrogen evolution under aerobic conditions.

Author contributions

All authors contributed equally to this work. M. H. and T. N. conducted most of the experiments and analyzed the electron transfer kinetics. M. H. drafted the manuscript. T. N. conceived the idea for this work and modified the manuscript. Y. A. supervised the project. N. K. and K. K. provided suggestions for the project. T. J. prepared the PGP. K. K. assisted with PSI preparation. Y. A. finalized the manuscript. All authors discussed the results and provided comments on the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The authors confirm that the data supporting the findings of this manuscript are available within the article and its supplementary information (SI). Supplementary information: UV-vis spectra, transient absorption time courses, time course of dissolved O₂ concentration, TEM and DLS data, 2D classification results, kinetic derivations (eqn (S1)–(S8)), definitions of TON and O₂ tolerance (eqn (S9) and (S10)), immobilization analysis, hydrogen-evolution data, and additional figures (PDF). See DOI: https://doi.org/10.1039/d6se00450d.

Acknowledgements

This work was supported by JSPS KAKENHI (N. K.) Grant No. 17H03645 and Grant No. 17H06434 in Scientific Research on Innovative Areas “Innovations for Light-Energy Conversion (I⁴LEC).” T. N. thanks the Takahashi Industrial and Economic Research Foundation, the Koyanagi Foundation, the Iwatani Naoji Foundation, the Japan Prize Foundation, and the Strategic Research Grant-in-Aid for Young Scientists at Osaka City University for funding. We thank Associate Professor Yoshihiro Yamaguchi (Graduate School of Science, Osaka Metropolitan University, Japan) for access to an electroporation device. We thank Professor Yasuyuki Tsuboi (Graduate School of Science, Osaka Metropolitan University, Japan) for access to DLS facilities. We thank Professor Makoto Miyata and Postdoctoral Researcher Yu-hei Tahara (Graduate School of Science, Osaka Metropolitan University, Japan) for assistance with TEM observations of platinum nanoparticles and PSI–Pt nanoparticle complexes. We thank Professor Hirozo Oh-oka (Graduate School of Science, Osaka University, Japan) for assistance with transient absorption measurements under anaerobic conditions and for providing cyt c₆ samples. We thank Assistant Professor Masaharu Kondo and Professor Takehisa Dewa (Nagoya Institute of Technology, Japan) for helpful discussions and for providing cyt c₆ samples.

References

1. A. Fujishima and K. Honda, Nature, 1972, 238, 37–38.
2. Q. Wang, T. Hisatomi, Q. Jia, H. Tokudome, M. Zhong, C. Wang, Z. Pan, T. Takata, M. Nakabayashi, N. Shibata, Y. Li, I. D. Sharp, A. Kudo, T. Yamada and K. Domen, Nat. Mater., 2016, 15, 611–615.
3. D. W. Wakerley and E. Reisner, Energy Environ. Sci., 2015, 8, 2283–2295.
4. M. G. Allan, M. J. McKee, F. Marken and M. F. Kuehnel, Energy Environ. Sci., 2021, 14, 5523–5529.
5. B. Mondal and A. Dey, Chem. Commun., 2017, 53, 7707–7715.
6. N. Nelson and A. Ben-Shem, Nat. Rev. Mol. Cell Biol., 2004, 5, 971–982.
7. N. Nelson and C. F. Yocum, Annu. Rev. Plant Biol., 2006, 57, 521–565.
8. F. R. Whatley, K. Tagawa and D. I. Arnon, Proc. Natl. Acad. Sci. U. S. A., 1963, 49, 266–270.
9. U. Schreiber, Photosynth. Res., 2017, 134, 343–360.
10. P. Jordan, P. Fromme, H. T. Witt, O. Klukas, W. Saenger and N. Krauß, Nature, 2001, 411, 909–917.
11. K. Brettel, Biochim. Biophys. Acta, Bioenerg., 1997, 1318, 322–373.
12. S. Itoh, M. Iwaki and I. Ikegami, Biochim. Biophys. Acta, Bioenerg., 2001, 1507, 115–138.
13. L. M. Utschig and K. L. Mulfort, Chem. Commun., 2024, 60, 10642–10654.
14. P. N. Ciesielski, F. M. Hijazi, A. M. Scott, C. J. Faulkner, L. Beard, K. Emmett, S. J. Rosenthal, D. Cliffel and G. K. Jennings, Bioresour. Technol., 2010, 101, 3047–3053.
15. C. E. Lubner, P. Knörzer, P. J. N. Silva, K. A. Vincent, T. Happe, D. A. Bryant and J. H. Golbeck, Biochemistry, 2010, 49, 10264–10266.
16. M. Ihara, H. Nishihara, K.-S. Yoon, O. Lenz, B. Friedrich, H. Nakamoto, K. Kojima, D. Honma, T. Kamachi and I. Okura, Photochem. Photobiol., 2006, 82, 676–682.
17. L. M. Utschig, N. M. Dimitrijevic, O. G. Poluektov, S. D. Chemerisov, K. L. Mulfort and D. M. Tiede, J. Am. Chem. Soc., 2011, 133, 16334–16337.
18. S. C. Silver, J. Niklas, P. Du, O. G. Poluektov, D. M. Tiede and L. M. Utschig, J. Am. Chem. Soc., 2013, 135, 13246–13249.
19. N. S. Ponomarenko, N. J. Zaluzec, X. Zuo, O. J. Borkiewicz, J. M. Hoffman, G. Kwon, A. B. F. Martinson, L. M. Utschig and D. M. Tiede, ACS Nano, 2025, 19, 4170–4185.
20. L. M. Utschig, N. M. Dimitrijevic, O. G. Poluektov, S. D. Chemerisov, K. L. Mulfort and D. M. Tiede, J. Phys. Chem. Lett., 2011, 2, 236–241.
21. C. J. Gisriel, T. Malavath, T. Qiu, J. P. Menzel, V. S. Batista, G. W. Brudvig and L. M. Utschig, Nat. Commun., 2024, 15, 9519.
22. M. Gorka, J. Schartner, A. Van Der Est, M. Rögner and J. H. Golbeck, Biochemistry, 2014, 53, 2295–2306.
23. T. D. Rumbaugh, M. J. Gorka, C. S. Baker, J. H. Golbeck and A. Silakov, Proc. Natl. Acad. Sci. U. S. A., 2024, 121, e2400267121.
24. Z. Xu, S. Wang, C. Zhao, S. Li, X. Liu, L. Wang, M. Li, X. Huang and S. Mann, Nat. Commun., 2020, 11, 5985.
25. Z. Xu, J. Qi, S. Wang, X. Liu, M. Li, S. Mann and X. Huang, Nat. Commun., 2023, 14, 1872.
26. T. Noji, M. Kondo, T. Jin, T. Yazawa, H. Osuka, Y. Higuchi, M. Nango, S. Itoh and T. Dewa, J. Phys. Chem. Lett., 2014, 5, 2402–2407.
27. T. Noji, T. Jin, M. Nango, N. Kamiya and Y. Amao, ACS Appl. Mater. Interfaces, 2017, 9, 3260–3265.
28. T. Noji, T. Suzuki, M. Kondo, T. Jin, K. Kawakami, T. Mizuno, H. Oh-oka, M. Ikeuchi, M. Nango, Y. Amao, N. Kamiya and T. Dewa, Res. Chem. Intermed., 2016, 42, 7731–7742.
29. T. Noji, K. Kawakami, J.-R. Shen, T. Dewa, M. Nango, N. Kamiya, S. Itoh and T. Jin, Langmuir, 2016, 32, 7796–7805.
30. A. Goyal, S. Szewczyk, G. Burdziński, M. Abram, J. Kargul and K. Gibasiewicz, Photochem. Photobiol. Sci., 2022, 21, 319–336.
31. A. A. Petrova, G. E. Milanovsky, I. A. Volkhin, M. A. Kozuleva, D. A. Cherepanov and A. Y. Semenov, Photosynth. Res., 2025, 164, 1.
32. C. Kamidaki, T. Kondo, T. Noji, T. Itoh, A. Yamaguchi and S. Itoh, J. Phys. Chem. B, 2013, 117, 9785–9792.
33. F. Müh and A. Zouni, Biochim. Biophys. Acta, Bioenerg., 2005, 1708, 219–228.
34. T. Yazawa, H. Tanaka, K. Eguchi and S. Yokoyama, J. Mater. Sci., 1994, 29, 3433–3440.
35. H. Tanaka, T. Yazawa, K. Eguchi, H. Nagasawa, N. Matsuda and T. Einishi, J. Non-Cryst. Solids, 1984, 65, 301–309.
36. S. Chen and K. Kimura, J. Phys. Chem. B, 2001, 105, 5397–5403.
37. B. Miroux and J. E. Walker, J. Mol. Biol., 1996, 260, 289–298.
38. A. Diaz, F. Navarro, M. Hervas, J. A. Navarro, S. Chavez, F. J. Florencio and M. A. De la Rosa, FEBS Lett., 1994, 347, 173–177.
39. E. Arslan, H. Schulz, R. Zufferey, P. Künzler and L. Thöny-Meyer, Biochem. Biophys. Res. Commun., 1998, 251, 744–747.
40. J. Gómez-Blanco, J. M. de la Rosa-Trevín, R. Marabini, L. Del Cano, A. Jiménez, M. Martínez, R. Melero, T. Majtner, D. Maluenda, J. Mota, Y. Rancel, E. Ramírez-Aportela, J. L. Vilas, M. Carroni, S. Fleischmann, E. Lindahl, A. W. Ashton, M. Basham, D. K. Clare, K. Savage, C. A. Siebert, G. G. Sharov, C. O. S. Sorzano, P. Conesa and J. M. Carazo, J. Struct. Biol., 2018, 204, 457–463.
41. C. O. S. Sorzano, R. Marabini, J. Velázquez-Muriel, J. R. Bilbao-Castro, S. H. W. Scheres, J. M. Carazo and A. Pascual-Montano, J. Struct. Biol., 2004, 148, 194–204.
42. R. Fernandez-Leiro and S. H. W. Scheres, Acta Crystallogr., Sect. D: Struct. Biol., 2017, 73, 496–502.
43. B. V. Trubitsin, M. D. Mamedov, A. Y. Semenov and A. N. Tikhonov, Photosynth. Res., 2014, 122, 215–231.
44. L. O. Pålsson, C. Flemming, B. Gobets, R. van Grondelle, J. P. Dekker and E. Schlodder, Biophys. J., 1998, 74, 2611–2622.
45. G. A. Truesdale and A. L. Downing, Nature, 1954, 173, 1236.

---