All-in-One photo-biocatalysis system for sustainable chemical transformations

Abstract

Enzyme-photo-coupled systems (EPCSs) offer a green strategy for producing high-value chemicals, but significant gaps remain in comparison to natural photosynthetic systems. Herein, we report an all-in-one EPCS (AiO-EPCS) by integrating photocatalytic (photocatalyst, P; electron mediator, M; electron donor, D) and biocatalytic (enzyme and cofactor) units in a hydrogen-bonded organic framework (HOF) for efficient NAD(P)H regeneration and light-driven enzymatic transformations, functioning like a chloroplast. The photocatalytic system assembled by incorporating P and M onto poly(allylamine hydrochloride) (PAH), PAH_PM, exhibits an 11-fold enhanced photocatalytic NAD(P)H regeneration yield over the free mixture. Subsequently, two different AiO-EPCSs are fabricated by trapping two biocatalytic systems into HOF with the assistance of PAH_PM. A double-layer architecture forms within the framework, partially separating the photo- and bio-catalytic modules while keeping all catalytic components in proximity. Moreover, adsorption of D onto HOF enables the system to be recycled by a single D addition, aligning AiO-EPCSs with natural photosynthesis in economic sustainability.

Enzyme-photo-coupled systems (EPCSs) offer a green strategy for producing high-value chemicals. In this study, the authors report an all-in-one EPCS (AiO-EPCS) by integrating photocatalytic and biocatalytic (enzyme and cofactor) units into a hydrogen bonded organic framework, for NAD(P)H regeneration and light-driven enzymatic transformations.

Introduction

Artificial enzyme-photo-coupled systems (EPCSs) have been developed to mimic natural photosynthetic systems, offering more economically viable and sustainable alternatives to traditional biocatalysis methods, especially those depending on oxidoreductases^1–7. However, there is still a significant gap between artificial EPCSs and natural photosynthetic systems due to the inherent complexity of the latter⁸. The primary challenge lies in integrating all functional modules as in natural photosynthetic systems, which benefits efficient photochemical conversion and synergistic process intensification⁸. In recent decades, considerable efforts have been devoted to constructing integrated EPCSs^2,3,9–12. For example, Sun et al. developed a capsule structure inspired by thylakoid membranes, co-anchoring cadmium sulfide quantum dots (photocatalyst) with a rhodium (Rh)-based organometallic complex (electron mediator) and alcohol dehydrogenase on the inner and outer surfaces of a protamine-induced titanium dioxide membrane, respectively, which could promote the mass transfer and thus enhance the product yield¹³. However, the co-immobilization process demands precise control and tedious steps, which complicate large-scale production. Chen et al. designed a Rh-modified metal–organic framework (Rh-NU-1006) that functioned simultaneously as a photocatalyst, an electron mediator, and a support for enzyme immobilization¹⁴. This strategy enabled the integration of three catalytic components in a simple and efficient manner with scalability. While these reported artificial EPCSs exhibited improved photocatalytic efficiency, their performance and cost-effectiveness remain a huge gap with natural photosynthetic systems. Nicotinamide (phosphate) cofactors (NAD(P)) typically serve as essential intermediaries that bridge the photocatalytic and biocatalytic units^10,15–17. Co-immobilizing NAD(P) with other modules in EPCSs is essential due to their high cost and pivotal role in electron transfer. Furthermore, though the electron donor (water) in natural photosynthetic systems¹⁵ is not strictly confined to the chloroplast, it does not undergo long-distance transfer during photocatalytic processes. Therefore, integrating a photocatalytic system (including the photocatalyst, electron mediator, and electron donor), a cofactor, and enzymes could offer a highly integrated approach that closely mimics natural photosynthetic systems to a maximum degree in composition, economic feasibility, and energy conversion, but such an all-in-one system has not been reported so far.

Hydrogen-bonded organic frameworks (HOFs) have emerged as a versatile platform for diverse applications through the incorporation of inorganic and organic guest species, owing to their facile synthesis and exceptional biocompatibility^18–27. Additionally, HOFs can facilitate proton transfer²⁸, benefiting from the hydrogen transfer in the photocatalytic regeneration pathway of NAD(P). Despite these advantages, most reported HOF systems have been limited to the encapsulation of homogeneous guest molecules via coprecipitation, whereas integrating multiple functional components to construct efficient EPCSs remains particularly challenging due to their intricate composition.

Herein, we create an all-in-one EPCS (AiO-EPCS) based on a HOF platform for NAD(P)H regeneration and enzymatic transformations, which exhibits the merit of reusability for all photocatalytic and biocatalytic units, functioning as artificial chloroplast organelles (Fig. 1). The integrated photoreaction unit is constructed through attaching porphyrin-based photocatalyst (P) and Rh-based electron mediator (M) onto poly(allylamine hydrochloride) (PAH) in the molecular-scale (Fig. 1a). The resulting PAH-coupled P&M (PAH_PM) displays an 11-fold enhancement in NAD(P)H regeneration over the free P and M system, likely due to the increased electron transfer in the integrated structure. Afterwards, cofactor and enzyme are in situ co-encapsulated within HOF with the assistance of PAH_PM, which benefits in enhancing guest loading and enzyme activity recovery²⁹. Moreover, the distinct binding affinities of PAH_PM-BioHOF-1 toward the enzyme and cofactor result in a double-layer configuration (Fig. 1b), providing partial spatial isolation of the photo- and biocatalytic modules without compromising their interfacial proximity. Two different AiO-EPCSs with NAD and NADP, respectively, are constructed, and they exhibit excellent chiral amine and ester yield under light. These AiO-EPCSs exhibit excellent recyclability (including electron donor (D)) due to the adsorption of D to HOF apart from the protection of the HOF shell. Overall, this work reports the possibility of constructing chloroplast-like organelles by rationally integrating all functional components in an EPCS, which surpasses the criterion for the state-of-the-art artificial EPCSs^{1–3,13,14,30}.

Fig. 1. Schematic illustration for the synthesis process of AiO-EPCSs.

a Preparation process of integrating P and M onto PAH. b Hierarchical assembly of photocatalytic system, NAD(P)H, and enzyme within BioHOF-1. Enzyme, E; NAD(P)H, N(P); monomer 1, m1; monomer 2, m2.

Results

Preparation and characterization of PAH_PM

In this study, meso-tetra(4-carboxyphenyl)porphine and [Cp*Rh-(bpy)H₂O]²⁺ were chosen as the model photocatalyst and electron mediator, respectively. BioHOF-1, assembled from tetraamidinium (m1) and tetracarboxylate (m2)³¹, was selected as the platform for fabricating AiO-EPCSs due to its stable architecture and good biocompatibility^29,31–33. P and M were integrated onto PAH one by one via the amidation reaction and/or coordination interaction (Fig. 1a and Supplementary S1). Notably, PAH was covalently modified with 4,4’-dicarboxylic acid-2,2’-bipyridine (DCBPy) prior to anchoring M (Fig. 1a). The successful synthesis of PAH_P, PAH_P-DCBPy, and PAH_PM was confirmed by proton nuclear magnetic resonance (¹H NMR), Fourier transform infrared (FT-IR) spectroscopy, and/or X-ray photoelectron spectroscopy (XPS) through the characteristic hydrogen peaks, amide bond and H₂O¹ (M) characteristic peaks, as well as atomic content changes and Rh 3 d₅ ⁄₂ and Rh 3 d₃  ⁄₂ peaks, respectively (Fig. 2a, b, Supplementary Figs. 2–5 and Table 1, Supplementary Discussion). TCPP loadings in PAH_P and PAH_PM were 274 and 124 μmol g^-1, respectively (Supplementary Figs. 6–9, Supplementary Discussion), while M loading in PAH_PM was 392.3 μmol g^-1 (Supplementary Table 2). Subsequently, PAH_PM was packed within BioHOF-1 in situ according to the approach recently reported by our group²⁹ to probe the feasibility of its encapsulation. As shown in Fig. 2c, m2 was added drop by drop into the mixture of m1 and PAH_PM, resulting in brown precipitate (denoted as PAH_PM-BioHOF-1). The morphology of PAH_PM-BioHOF-1 maintained the rod-like structure and similar dimensions (270 - 310 nm in average) of BioHOF-1 (Supplementary Figs. 10–11). The powder X-ray diffraction (PXRD) pattern of PAH_PM-BioHOF-1 aligned closely with BioHOF-1 (Fig. 2d). These results imply that the encapsulation of PAH_PM did not influence the architecture and crystallinity of BioHOF-1. The XPS spectrum of PAH_PM-BioHOF-1 confirmed the PAH_PM component in BioHOF-1 according to the existence of Rh 3 d peak (Fig. 2b). The energy-dispersive X-ray spectra (EDS) mapping further determined the presence of PAH_PM in BioHOF-1 by detecting Rh, a characteristic element of M (Fig. 2e). Visualization by confocal laser scanning microscopy (CLSM) demonstrates that PAH_PM was predominantly embedded within the internal framework of BioHOF-1 in the PAH_PM-BioHOF-1 composite (Supplementary Figs. 12–16, Supplementary Discussion).

Fig. 2. Characteristics of the integrated photocatalytic units.

a FT-IR spectra of PAH, TCPP, PAH_P, and PAH_PM. b XPS spectra of PAH, PAH_P, PAH_PM, and PAH_PM-BioHOF-1. c Schematic of synthesizing PAH_PM-BioHOF-1. d PXRD patterns of BioHOF-1 (1), pristine PAH_PM-BioHOF-1 (2), and PAH_PM-BioHOF-1 after photocatalytic regenerating NADH (3) and NADPH (4). e TEM images and EDS elemental mapping of C, N, O, and Rh for PAH_PM-BioHOF-1.

The optical absorption properties of PAH_PM and PAH_PM-BioHOF-1 were analyzed via solid-state UV−Visible diffuse reflectance spectroscopy (UV−Vis DRS), with TCPP and PAH_P included for comparison. As shown in Fig. 3a, Supplementary Figs. 17–19, and Supplementary Discussion, a noticeable red-shift in absorption bands of PAH_P, PAH_PM, and PAH_PM-BioHOF-1 was observed compared to TCPP, suggesting an increase in solar utilization capability. Tauc plots derived from the Kubelka−Munk function³⁴ were utilized to estimate the optical gaps of these four samples. As shown in Fig. 3b, compared to TCPP, the energy band gaps of PAH_P, PAH_PM, and PAH_PM-BioHOF-1 reduced, implying more efficient transfer of their photogenerated electrons. The cause is likely due to the presence of a large number of nitrogen atoms in their structures, which typically facilitates electron transfer³⁰. The attachment of M to PAH_P made a further decrease in its bandgap, resulting from the integration of P and M, which is beneficial in promoting the transfer efficiency of excited electrons. Further assembly of PAH_PM within BioHOF-1 did not affect its optical band gap, maintaining its excellent photoelectronic structure. The XPS valence band (VB) spectroscopy was then performed on the four samples (Fig. 3c and Supplementary Fig. 20). Based on the VB and optical band gap values, their conduction band (CB) potentials were calculated (Fig. 3c). All CB values of these four photocatalysts were lower than those of NAD⁺/NADH^2,10, while all VB values, except that of TCPP, were higher than the oxidation potential of the electron donor triethanolamine (TEOA; TEOA_OX/TEOA^2,10) (Fig. 3c). The energy-favorable electron band structures of these integrated photocatalysts make them feasible to regenerate NAD(P)H cofactors under the illumination of visible light.

Fig. 3. Photoelectronic performance of integrated photocatalytic units.

a UV–vis DRS spectra, b Tauc plot spectra, c schematic band structure, d PL spectra, e PL lifetime, and f transient photocurrent curves of TCPP, PAH_P, PAH_PM, and PAH_PM-BioHOF-1.

The photogenerated electron transfer and utilization processes of the above photocatalysts were examined through photoelectrochemical analysis. As shown in Fig. 3d, Supplementary Fig. 21, and Supplementary Discussion, the steady-state photoluminescence (PL) spectra of PAH_PM and PAH_PM-BioHOF-1 displayed a relatively low PL intensity compared to TCPP and PAH_P, suggesting that the integration of P and M could significantly suppress the recombination of photoexcited electron-hole pairs, which is favorable to the electron transfer and utilization^1–3. The transient fluorescence lifetime measurement reveals that PAH_PM-BioHOF-1 exhibited the shortest excited state lifetime (0.86 ns), whereas PAH_PM displayed the longest charge-carrier lifetime (3.4 ns, Fig. 3e). This difference may arise from the hydrogen bond networks within PAH_PM-BioHOF-1 (Supplementary Fig. 19), which could facilitate the interfacial charge transfer (Supplementary Fig. 22, Supplementary Discussion) and thus promote the conversion of the excited energy into nonradiative decay pathways. The prolonged fluorescence lifetime of PAH_PM over TCPP and PAH_P may be due to the reduced aggregation-induced quenching and increased rigidity of microenvironment due to the introduction of the M unit, which suppress nonradiative relaxation, consequently leading to an increased lifetime^35,36. Furthermore, the transient photocurrent responses of the four materials were investigated under intermittent visible light irradiation in 20-s cycles. As shown in Fig. 3f, all the four photocatalysts held the ability for photoinduced generation of electrons, with the photocurrent signals of PAH_PM > PAH_PM-BioHOF-1 > PAH_P > TCPP. The comparable photocurrent response of PAH_PM-BioHOF-1 with PAH_PM was due to the formation of beneficial charge-transfer pathways within PAH_PM-BioHOF-1 (Supplementary Figs. 19 and 22), despite the inherently poor electrical conductivity of BioHOF-1.

Photocatalytic NAD(P)H regeneration performance

Given the excellent photoelectronic performance of PAH_PM and PAH_PM-BioHOF-1, we examined the photocatalytic regeneration of NAD(P)H of these two catalysts. Photocatalytic NADH regeneration of them was first explored and compared with PAH_P and TCPP, with the successful production of bioactive NADH further verified by ¹H NMR spectroscopy (Supplementary Figs. 23-24, Supplementary Discussion). As depicted in Fig. 4a, PAH_PM displayed the highest NADH yield among the four materials during the 60-min reaction processes, followed by PAH_PM-BioHOF-1. This excellent performance was more remarkable in the initial reaction period (≤5 min, Fig. 4b). Specifically, PAH_PM and PAH_PM-BioHOF-1 both demonstrated ultrafast catalytic regeneration rates within the first 5 min, approximately 3 – 5 folds higher than TCPP and PAH_P. These improvements suggest that the integration of P and M could efficiently facilitate the electron dynamics, thus enhancing the NADH regeneration yield. Despite a decline in catalytic efficiency after 5 min, the NADH regeneration yields of PAH_PM and PAH_PM-BioHOF-1 remained substantially higher than those of TCPP and PAH_P after a 60 min-reaction (Fig. 4b). This observation highlights that the advantages introduced by the integration persisted in a long reaction period, beneficial to the efficiency and sustainability of photocatalysis. Subsequently, the photocatalytic efficiencies of the four materials in NADPH regeneration were investigated (Supplementary Figs. 24–25). As illustrated in Fig. 4c, d, PAH_PM exhibited the maximum NADPH regeneration efficiency over the entire reaction duration (180 min), further confirming that the integration effect is capable of significantly enhancing photocatalytic performance. Notably, the NADPH yield of PAH_PM-BioHOF-1 was comparable to that of PAH_P and TCPP, in contrast to its superior performance in NADH regeneration (Fig. 4a, b). This difference may arise from the relatively larger molecular size of NADP⁺ compared to NAD⁺ (Supplementary Fig. 26), which could hinder its efficient interaction with M within BioHOF-1. Despite this, PAH_PM-BioHOF-1 demonstrated a stable and sustainable photocatalytic regeneration rate over extended periods, and outperformed most recently reported photocatalytic systems (Supplementary Table 3), providing a robust basis for subsequent enzyme-photo-coupled catalytic applications.

Fig. 4. Photocatalytic NAD(P)H regeneration performance.

a Yield of NADH as a function of reaction time and b the corresponding NADH regeneration yield at 5 and 60 min with TCPP, PAH_P, PAH_PM, and PAH_PM-BioHOF-1. Statistical analysis was performed using two-way ANOVA followed by Šídák’s multiple comparisons test (two-sided). Adjusted P values are as follows: at 5 min, TCPP vs PAH_PM-BioHOF-1, p < 0.0001; PAH_P vs PAH_PM-BioHOF-1, p < 0.0001; PAH_PM vs PAH_PM-BioHOF-1, p = 0.0002. At 60 min, TCPP vs PAH_PM-BioHOF-1, p < 0.0001; PAH_P vs PAH_PM-BioHOF-1, p = 0.0002; PAH_PM vs PAH_PM-BioHOF-1, p < 0.0001. c Yield of NADPH as a function of reaction time and d the corresponding NADPH regeneration yield at 5 and 180 min with TCPP, PAH_P, PAH_PM, and PAH_PM-BioHOF-1. Statistical analysis was performed using two-way ANOVA followed by Šídák’s multiple comparisons test (two-sided). Adjusted P values are as follows: at 5 min, TCPP vs PAH_PM, p < 0.0001; PAH_P vs PAH_PM, p < 0.0001; PAH_PM vs PAH_PM-BioHOF-1, p < 0.0001. At 180 min, TCPP vs PAH_PM, p < 0.0001; PAH_P vs PAH_PM, p < 0.0001; PAH_PM vs PAH_PM-BioHOF-1, p < 0.0001. *** p < 0.001 and **** p < 0.0001. Recyclability of photocatalytic (e) NADH and f NADPH regeneration with PAH_PM and PAH_PM-BioHOF-1. Data represent mean ± s.d. from three independent experiments (n = 3 independent biological replicates).

Subsequently, we evaluated the recyclability of these integrated complexes in NAD(P)H regeneration. After 8 cycles, PAH_PM-BioHOF-1 retained 77% and 85% of initial activity for regenerating NADH and NADPH, respectively (Fig. 4e, f). In contrast, PAH_PM exhibited a significant loss in catalytic activity during the recycling and reuse, retaining only 28% and 24% of residual activity in regenerating NADH and NADPH, respectively, after 8 cycles (Fig. 4e, f). The substantial decline in the photocatalytic performance of PAH_PM may be attributed to the leaching of Rh in the absence of the BioHOF-1 protective shell (Supplementary Table 2), a phenomenon previously reported¹. Moreover, the difficulty in recovering PAH_PM also resulted in a decrease in its activity in repeated use (Supplementary Table 4). Consequently, while PAH_PM demonstrated somewhat superior photocatalytic NAD(P)H regeneration performance, PAH_PM-BioHOF-1 exhibits much greater potential for practical applications due to its excellent stability and reusability. To further assess the stability of PAH_PM-BioHOF-1, its structural integrity was characterized following photocatalytic reactions. Compared to the pristine PAH_PM-BioHOF-1, its surface became rough with emerging nanoparticles, regardless of whether it was used for regenerating NADH or NADPH (Supplementary Figs. 27 and 28). It is considered that the high ionic strength in the regeneration reaction system affected the hydrogen bonding network on the surface of PAH_PM-BioHOF-1, making its surface become loose and leading to the formation of the nano-sized aggregates²⁹. Nevertheless, PXRD analysis revealed that the post-reaction PAH_PM-BioHOF-1 samples retained their pristine crystalline structure well (Fig. 2d). Furthermore, EDS mapping determined M still presented in the main body of PAH_PM-BioHOF-1 after the photocatalytic NAD(P)H regeneration reactions (Supplementary Figs. 29 and 30). These results indicate that PAH_PM-BioHOF-1 displays efficient and sustainable catalysis performance in regenerating NAD(P)H.

AiO-EPCS for chiral amine synthesis with NAD recycling

An NADH-dependent amine dehydrogenase (AmDH) (Supplementary Fig. 31) was utilized as a model enzyme to construct the first AiO-EPCS. This enzymatic reaction can convert precursor ketones into chiral amines (Supplementary Fig. 32), demonstrating significant potential as valuable intermediates for pharmaceutical applications³⁷. To enhance guest loading and protect enzyme activity recovery, we tend to co-encapsulate AmDH and NADH in PAH_PM-BioHOF-1 with the assistance of the polyelectrolyte (PAH_PM)²⁹. Therefore, the AmDH&NADH/PAH_PM complex was first constructed through electrostatic interactions between positive-charged PAH_PM and negative-charged AmDH and NADH (Supplementary Figs. 33 and 34 and Fig. 1b). Notably, AmDH and NADH primarily interacted electrostatically with the remaining unreacted amino groups in PAH_PM, rather than the P and M units that are both negatively charged (Supplementary Fig. 33). An integrated system comprising P, M, AmDH, and NADH, i.e., Am&N@PAH_PM-BioHOF-1, was then fabricated by hydrogen bonding- and electrostatic interactions-mediated self-assembly of PAH_PM-BioHOF-1 (Fig. 1b). The morphology and crystallinity of Am&N@PAH_PM-BioHOF-1 kept well with the pristine BioHOF-1 (Supplementary Fig. 35 and Fig. 5a). FT-IR, EDS mapping, and adsorption experiments demonstrate that AmDH and NADH were successfully incorporated into PAH_PM-BioHOF-1, which was according to the characteristic absorptions of enzyme (β-sheet structures and C = C stretching vibrations from aromatic residues), the characteristic elements (AmDH: S; NADH: P), and little adsorption of NADH on the HOF surface, respectively (Fig. 5b, c, Supplementary Figs. 36 and 37, Supplementary Discussion). CLSM was utilized to visualize the specific distributions of AmDH and NADH within PAH_PM-BioHOF-1, in which AmDH was labeled with fluorescein isothiocyanate (FITC), while NADH cofactor has the autofluorescence property. As illustrated in Supplementary Figs. 38 and 40 and Supplementary Discussion, intense green (Am) and blue (N) fluorescence were observed throughout the PAH_PM-BioHOF-1 structure. Moreover, the CLSM images of cross-sections (Z-axis) showed the internal luminescence (green and blue) inside Am-FITC&N@PAH_PM-BioHOF-1 (Fig. 5d). These results revealed that AmDH and NADH were indeed immobilized inside PAH_PM-BioHOF-1 rather than the material surface. The 3D CLSM images of Am-FITC&N@PAH_PM-BioHOF-1 displayed a distinct double-layer distribution for AmDH and NADH, with AmDH localized in the inner layer (green) and NADH (blue) in the outer layer (Fig. 5e, f and Supplementary Fig. 41). This may be due to the binding affinity difference between AmDH/NADH and PAH_PM-BioHOF-1, thus leading to their competitive encapsulation during the charge-driven HOF assembly process. To verify this speculation, their binding capability was investigated and compared through a combination of experiments and calculations (Supplementary Figs. 42–54 and Tables 5 and 6, Supplementary Discussion). The results demonstrate that AmDH with stronger binding energy was encased into PAH_PM-BioHOF-1 prior to NADH, thus resulting in the double-layer distribution. As AmDH and NADH primarily interacted with the unmodified amino groups on PAH_PM, the double-layer architecture might make the partial compartmental distribution between photocatalytic and biocatalytic components through the NADH barrier, resembling the light and dark reactions in the chloroplast. Despite this, the charge-transfer property of PAH_PM-BioHOF-1 was nearly unchanged upon co-encapsulating AmDH and NADH (Supplementary Figs. 22 and 55, Supplementary Discussion).

Fig. 5. Co-encapsulation of AmDH and NADH within PAH_PM-BioHOF-1.

a PXRD patterns of BioHOF-1 (1), PAH_PM-BioHOF-1 (2), pristine (3) and post-reaction (4) Am&N@PAH_PM-BioHOF-1. b FT-IR spectra of pristine and post-reaction Am&N@PAH_PM-BioHOF-1, as well as PAH_PM-BioHOF-1. c TEM images and EDS elemental mapping of Rh, P, and S for Am&N@PAH_PM-BioHOF-1. d CLSM images of Am-FITC&N@PAH_PM-BioHOF-1 at a series of focal planes measured every 0.1 μm along the Z-axis. e The 3D top and f side view of CLSM images of Am-FITC&N@PAH_PM-BioHOF-1.

The all-in-one Am&N@PAH_PM-BioHOF-1 system (hereafter referred to as APS/Am&N) as an artificial chloroplast reactor was then evaluated for synthesizing a chiral amine (Fig. 6a). The loadings of M, AmDH, and NADH in PAH_PM-BioHOF-1 were 34 μmol g^-1, 576, and 71 mg g^-1, respectively (Supplementary Table 2 and Fig. 6b). 2-Hexanone was chosen as the model substrate, which can be converted to (R)-2-aminohexane by the catalysis of AmDH with the assistance of NADH (Fig. 6a). The photocatalytic conversion of (R)-2-aminohexane was carried out under the illumination of visible light with APS/Am&N. Notably, the reductive amination reaction would be triggered immediately upon the addition of 2-hexanone due to the immobilization of NADH in PAH_PM-BioHOF-1. To evaluate the advantage of integrating photocatalytic and biocatalytic systems, the NADH-dependent AmDH systems at different forms were prepared for comparison (Supplementary Fig. 56), including immobilized AmDH and free NADH system (Am@P_PM-HOF + N), a hierarchical system with an outer PAH_PM-BioHOF-1 shell and an inner Am&N@PAH-BioHOF-1 core (H-P_PM/Am&N), and free PAH_PM, AmDH, and NADH system (P_PM+Am+N). As illustrated in Fig. 6c, the product conversion of APS/Am&N (20.0%) was significantly higher than that of Am@P_PM-HOF + N (7.8%) and P_PM + Am+N (11.2%) at the 30-min reaction. The excellent catalytic efficiency of APS/Am&N may be attributed to the proximity effect of PAH_PM, AmDH, and NADH within BioHOF-1, which facilitated the electron and mass transfer. The conversion rate of H-P_PM/Am&N (19.4%) at 30 min was comparable to that of APS/Am&N (Fig. 6c), likely due to the co-immobilization of NADH and AmDH, which enabled AmDH to efficiently utilize the preloaded NADH for catalytic reactions during the early stage. As the depletion of preloaded NADH, the enzymatic conversion of AmDH gradually depended on the regenerated NADH by PAH_PM. The faster cofactor transfer rate in APS/Am&N would facilitate the whole catalytic efficiency. Therefore, the product yield of APS/Am&N gradually surpassed that of H-P_PM/Am&N after 30 min (Fig. 6d). Finally, APS/Am&N achieved a 92% product yield with over 99% enantiomeric excess (ee) at 11 h, whereas H-P_PM/Am&N reached only 77% conversion (Fig. 6d and Supplementary Fig. 57 and Supplementary Table 7). These results further demonstrate the benefits of integrating a photocatalytic system, NAD cofactors, and a biocatalytic system over separating them hierarchically. Notably, despite integration of all the catalytic components in APS/Am&N, electrostatic repulsion between the photo- and biocatalytic units prevented close contact between them, as discussed above. This separation therefore would mitigate the direct exposure of enzyme molecules to reactive oxygen species (ROS), primarily singlet oxygen, generated during photocatalysis (Supplementary Figs. 58–61 and Supplementary Discussion). Moreover, the double-layer distribution of AmDH and NADH in BioHOF-1 consequently led to a spatial partitioning of their interacting photocatalytic system (PAH_PM), with part of PAH_PM located in the outer layer, making ROS generated in outer layer separate from the enzyme. However, ROS generated in the inner layer could still reach AmDH and damage the enzyme to some extent. Therefore, APS/Am&N exhibited a decrease in enzymatic activity upon prolonged illumination (Supplementary Fig. 62). Despite this, APS/Am&N retained significantly higher activity than that of the free counterpart, which exhibited little activity after 12 h of light irradiation (Supplementary Fig. 62). It is worth noting that the hierarchical distribution of AmDH and NADH did not affect the biocatalytic conversion in APS/Am&N. Cofactors electrostatically immobilized on the support undergo reversible adsorption-desorption dynamics, which confer high mobility and efficient enzyme accessibility^38,39. In addition, the hierarchical porosity of BioHOF-1 (average pore size: ~7.1 nm) arising from interparticle voids and minor localized structural heterogeneity²⁹ could allow the free transfer of bulky molecular NAD(P) within the framework (Supplementary Fig. 26). Collectively, the electrostatically anchored NAD in APS/Am&N could continually shuttle between PAH_PM and AmDH. This dynamic behavior of NAD ensured both close spatial proximity and effective interfacial contact between the photo- and biocatalytic systems. Therefore, while H-P_PM/Am&N may ensure more effective enzyme-photocatalyst separation through the HOF shell, APS/Am&N achieved a higher overall product yield by virtue of the closer spatial proximity of its functional units (Fig. 6d). Collectively, the specific configuration of APS/Am&N partially mitigated photoreaction-induced enzyme damage while preserving close spatial proximity among the functional modules, thereby enabling excellent photoenzymatic conversion efficiency.

Fig. 6. Enzyme-photo-coupled catalytic performance of APS/Am&N.

a Schematic showing the production of (R)-2-aminohexane with Am&N@PAH_PM-BioHOF-1 under light. b Immobilization results of AmDH and NADH in APS/Am&N. c Enzymatic conversion of (R)-2-aminohexane under light irradiation with different biocatalytic systems at initial 30 min. Entry 1 - 4 represented Am@P_PM-HOF + N, APS/Am&N, H-P_PM/Am&N, and P_PM + Am+N, respectively. Statistical analysis was performed using two-way ANOVA followed by Šídák’s multiple comparisons test (two-sided). Adjusted P values are as follows: 1 vs 2, p = 0.0001; 2 vs 4, p = 0.0007. ***p < 0.001. d Product yield of (R)-2-aminohexane with different biocatalytic systems as a function of time under light. e Reusability and accumulated NAD TTNs of APS/Am&N in 8 recycles. Note that TEOA (400 mM) was added only in the first batch of reaction. Data represent mean ± s.d. from three independent experiments (n = 3 independent biological replicates).

It is worth noting that the whole catalytic efficiencies of Am@P_PM-HOF + N and P_PM + Am+N were comparable to that of APS/Am&N after 2 h (Fig. 6d). The cause may be the partitioning and storage of free NADH from bulk solution to HOF for Am@P_PM-HOF + N and the electrostatic attraction of PAH_PM to AmDH and NADH for P_PM + Am+N. These two phenomena made NADH close to the photocatalytic and biocatalytic units, so the conversions of Am@P_PM-HOF + N and P_PM + Am+N approached to that of APS/Am&N at the later stage of reaction (Fig. 6d). However, only APS/Am&N could recycle all functional components, exhibiting even over 100% residual activity after 8 cycles of reuse (Supplementary Fig. 63). The slight increase in activity in the 8th cycle was due to the adsorption of TEOA (D) onto PAH_PM-BioHOF-1 (Supplementary Figs. 64–68, Supplementary Discussion), which led to a gradual rise in TEOA concentration over 8 continuous additions of fresh TEOA, thus enhancing the overall catalytic efficiency of APS/Am&N. Due to the D adsorption, the whole catalytic activity of APS/Am&N was almost maintained over multiple cycles without D addition in repeated uses (Fig. 6e). Such an all-in-one integration, combining the photocatalyst, electron mediator, electron donor, cofactor, and enzymes, allowed all functional modules to be efficiently recycled, which has not been reported for any previous EPCSs (Supplementary Table 3, Supplementary Discussion).

The morphology, crystal structure, and elemental composition of APS/Am&N kept well after an 11-h photo-enzymatic catalysis reaction or 8 recycling cycles (Fig. 5a–c and Supplementary Figs. 35 and 36 and 69–73). Collectively, these results demonstrate that APS/Am&N possessed excellent operational stability and photosynthesis performance. Furthermore, the 11-h product yield of APS/Am&N (92% with 0.13 mM immobilized NADH) was comparable to that of Am@P_PM-HOF + N⁺ (97% with 1 mM free NAD⁺) (Fig. 6d). This highlights the high cofactor recycling efficiency of APS/Am&N, enabling an excellent product conversion with minimal cofactor input. The total turnover numbers (TTNs) of NADH accumulated to 296 after the 8 cycles of reuse (Fig. 6e).

AiO-EPCS for ester synthesis with NADP recycling

To evaluate the versatility of AiO-EPCS, an NADPH-dependent phenylacetone monooxygenase (PAMO) was activated (Supplementary Figs. 74 and 75) in the PAH_PM-BioHOF-1 nanoreactor (Fig. 1b). The synthesized PAMO&NADPH@PAH_PM-BioHOF-1 (abbreviated as APS/PA&NP) can catalyze the transformation of bicyclo[3.2.0]hept-2-en-6-one into 3-oxabicyclo[3.3.0]oct-6-en-2-one under visible light illumination, with efficient NADP recycling (Fig. 7a). TEM, PXRD, FT-IR, EDS, and CLSM data demonstrated the successful assembly of APS/PA&NP and its double-layer distribution structure (Fig. 7b, c and Supplementary Figs. 76–81). Loadings of M, PAMO, and NADPH in APS/PA&NP were 49.4 µmol g^-1, 103 mg g^-1 and 8.56 µmol g^-1, respectively (Supplementary Table 2 and Fig. 82). Figure 7d and Supplementary Fig. 83, along with Supplementary Table 7 illustrate the formation of 3-oxabicyclo[3.3.0]oct-6-en-2-one catalyzed by APS/PA&NP under light illumination. APS/PA&NP exhibited a product yield of 89% upon a 7-h reaction (Fig. 7d). By comparison to other three systems, including immobilized PAMO and free NADPH system (PA@P_PM-HOF + NP), free PAH_PM, PAMO, and NADPH system (P_PM + PA + NP), and a hierarchical system with an outer PAH_PM-BioHOF-1 shell and an inner PA&NP@PAH-BioHOF-1 core (H-P_PM/PA&NP), APS/PA&NP revealed a similar performance character with APS/Am&N described above. Specifically, the conversion of APS/PA&NP (12.6%) was significantly higher that of PA@P_PM-HOF + NP (7.0%) at the initial reaction period (30 min) due to the proximity effect in APS/PA&NP (Fig. 7d). In the later reaction period (1 - 7 h), the catalytic efficiency difference between APS/PA&NP and PA@P_PM-HOF + NP / P_PM + PA + NP gradually decreased or became negligible (Fig. 7d), possibly due to the proximity of catalytic modules driven by non-bonding interactions in the latter two systems (PA@P_PM-HOF---NP and P_PM---NP). In contrast, H-P_PM/PA&NP displayed poor conversion compared to APS/PA&NP at 1 – 7 h (Fig. 7d), most likely owing to the mass transfer resistance on NADPH between the two HOF layers. Furthermore, the product yield of APS/PA&NP gradually surpassed that of the free P, M, PAMO, and NADP⁺/NADPH mixtures (P + M + PA + NP⁺ and P + M + PA + NP) after 1.5 h (Fig. 7e), further indicating the benefits of proximity effect and operational stability in the APS/PA&NP system. Notably, the catalytic performance of APS/PA&NP was slightly lower than that of P + M + PA + NP⁺ and P + M + PA + NP at 0 – 1 h (Fig. 7e), which was likely caused by the partial light shielding by the HOF shell over the internal PAH_PM. TEM, PXRD, FT-IR, and EDS data demonstrated the stable topology structure and the well retained catalytic components of APS/PA&NP after a 7-h photo-biocatalysis reaction (Supplementary Figs. 76–78 and 84–85). APS/PA&NP maintained nearly 100% of initial activity after 8 cycles of PAMO-photo-coupled reaction, achieving accumulated TTNs of 1225 for NADP (Fig. 7f). The morphology, crystallinity, and chemical composition of APS/PA&NP were still retained after 8 recycles (Supplementary Figs. 76–78 and 86–87). These results proved the potential of APS/PA&NP in economical and sustainable synthesis through a photo-chemical energy conversion mode.

Fig. 7. Enzyme-photo-coupled catalytic performance of APS/PA&NP.

a Scheme of enzyme-photo-coupled reaction catalyzed by APS/PA&NP with NADP recycling. b TEM images and EDS elemental mapping of Rh, P, and S for APS/PA&NP. c CLSM images for the distributions of PAMO-FITC and NADPH in PAH_PM-BioHOF-1. BF, bright field. d Investigating mass transfer with photo-PAMO coupling catalytic systems in different forms. e Product yield of 2-oxabicyclo[3.3.0]oct-6-en-3-one with PAMO at different forms as a function of time under light. f Reusability and accumulated NADP TTNs of APS/PA&NP in 8 recycles. Data represent mean ± s.d. from three independent experiments (n = 3 independent biological replicates).

Discussion

A biomimetic chloroplast-like nanoreactor is designed and constructed by an all-in-one strategy, where all functional modules, including P, M, D, cofactor, and enzyme, are integrated in PAH-BioHOF-1. Specifically, P and M are anchored onto PAH through covalent bonding reactions. This molecular-scale integration enables the photocatalytic system to present excellent light adsorption and electron transfer efficiency, resulting in an 11-fold enhancement in NAD(P)H photocatalytic regeneration yield over the corresponding free mixtures. Then, the photocatalytic module (PAH_PM), and biocatalytic components (cofactor and enzyme) are confined within BioHOF-1 in a nanoscale compartmentalized distribution via hydrogen bonding- and electrostatic interaction-induced self-assembly. Two different enzymatic systems, including NADH-dependent AmDH and NADPH-dependent PAMO, are validated in PAH_PM-BioHOF-1, featuring a double-layer distribution. This hierarchical organization is attributed to the differentiated binding affinities of PAH_PM-BioHOF-1 (primarily the free amino groups of PAH_PM) toward enzymes and cofactors, which in turn lead to distinct nucleation rates of these components within the framework. As a result, the enzyme and cofactor are spatially segregated during framework growth, giving rise to a well-defined hierarchical architecture. Such a spatial organization enables partial isolation of the enzymatic components from the photocatalytic system while preserving interfacial proximity, thus balancing enzyme protection and efficient photo-enzymatic coupling. Building on the mechanistic insights revealed in this work, future efforts may focus on further refining the spatial organization to mitigate photoreaction-induced enzyme deactivation. For example, this could be achieved by strengthening interfacial barriers between the photocatalytic and biocatalytic modules via tuning the interactions between PAH_PM and enzymes/cofactors, or by introducing additional elements capable of buffering reactive oxygen species. Collectively, these rational design strategies are expected to enhance enzyme stability under prolonged illumination while maintaining efficient photo–biocatalytic coupling.

Both APS/Am&N and APS/PA&NP demonstrate excellent product yields (92% and 89%, respectively) under visible light illumination, due to the nanoscale proximity among P, M, cofactor, and enzyme. Moreover, these two integrated systems display excellent operational stability and recyclability caused by the structural tightness within PAH_PM-BioHOF-1. The adsorption of PAH-BioHOF-1 for TEOA (D) endows the photo-biocatalysis systems to sustain without the addition of TEOA for at least 8 batch reactions, further promoting the synergistical intensification and process sustainability of the AiO-EPCS design. Overall, AiO-EPCS presented in this study has great potential in fabricating different photo-biocatalysis systems for various applications.

Methods

Synthesis of PAH_PM

The synthesis of PAH_PM was achieved by sequentially modifying PAH with P and M. Thus, PAH_P was first synthesized. Specifically, P (70.48 mg), 3-[(ethylimino)methylidene]amino-N, N-dimethylpropan-1-amine (EDC, 537.2 mg), and N-hydroxysuccinimide (NHS, 124.9 mg) were added to the 5 mL of phosphate buffer (50 mM, pH 5.5) and then incubated at low temperature and in the dark for 30 min to allow activation. Afterwards, the above mixture was added to the HEPES buffer (50 mM, pH 8.0) containing 10 mg mL^-1 PAH to react overnight under dark condition. The unreacted small chemicals were removed through dialysis (MWCO: 8 kDa) after the reaction. Fianlly, the as-prepared PAH_P was collected by freeze drying.

DCBPy (6.93 mg), EDC (48.1 mg), and NHS (11.2 mg) were added to the 5 mL of phosphate buffer (50 mM, pH 5.5) and then incubated at low temperature and in the dark for 30 min to allow activation. Afterwards, the above mixture was added to the HEPES buffer (50 mM, pH 8.0) containing the as-obtained PAH_P to react overnight under dark condition. The as-synthesized PAH_P-DCBPy and [Cp*RhCl₂]₂ were dispersed in the deionized water under dark and stirring conditions for 12 h. The separation and purification procedures were the same as above.

Synthesis of PAH_PM-BioHOF-1

Briefly, the 1 mg mL^-1 of PAH_PM in 0.5 mL of methanol was mixed with an 8 mg mL^-1 m1 (tetraamidinium monomer) aqueous solution (0.25 mL) to form mixture I. Tetracarboxylate monomer 2 (1.5 mg) was dissolved in 0.5 mL of deionized water containing 1% v/v aqueous ammonia (0.05 mL) to form solution II. Subsequently, solution II was added dropwise to mixture I while stirring in the dark at room temperature for 1 h. PAH_PM-BioHOF-1 (brown precipitate) was then formed, recovered by centrifugation (12000 × g, 10 min), and washed three times with deionized water.

Constructing AiO-EPCSs

For the synthesis of AmDH&NADH@PAH_PM-BioHOF-1 (APS/Am&N), 8 mg mL^-1 of PAH_PM was suspended in methanol, and then it was mixed with Tris-HCl buffer (50 mM, pH 8.0) containing 5.2 mg mL^-1 AmDH and 2 mg mL^-1 NADH. The mixture was gently stirred at 4 °C for 1  h. Subsequently, 8 mg mL^-1 of m1 was added to the former mixture to form mixture III. Monomer 2 solution (3 mg/mL) was added dropwise to mixture III while stirring in the dark at 4 °C for 1  h.

For the preparation of PAMO&NADPH@PAH_PM-BioHOF-1 (APS/PA&NP), 8 mg mL^-1 of PAH_PM suspended in methanol was mixed with Tris-HCl buffer (50 mM, pH 7.5) containing 1 mg mL^-1 PAMO and 0.2 mg mL^-1NADPH. The mixture was gently stirred at 4 °C for 1  h. Afterwards, 8 mg mL^-1 m1 aqueous solution was added to it to form mixture IV. Monomer 2 solution (3 mg/mL) was added dropwise to mixture IV while stirring in the dark at 4 °C for 1  h.

The above resulted brown precipitates were collected by centrifugation at 12000 g for 10 min and washed three times with 50 mM Tris-HCl buffer (pH 8.0 for APS/Am&N and pH 7.5 for APS/PA&NP).

Photocatalytic regeneration of NAD(P)H

A quartz cuvette (3.1 mL) was used as the reactor for the photocatalytic regeneration of NAD(P)H, with a 100 W LED lamp (420 nm) serving as the light source to simulate visible sunlight. The distance between the light source and the quartz reactor was maintained at 8 - 10 cm. The photocatalytic regeneration reactions were carried out in the 3 mL of 2 M NH₄Cl buffer (pH 8.7) containing 42.16 µM TCPP, 0.13 mM M, 1 mM NAD(P)⁺, and 400 mM TEOA. The reaction systems were equilibrated in the dark for 15 minutes before being exposed to the LED lamp. The regeneration yield of NAD(P)H was determined by measuring the absorbance at 340 nm of the reaction system using a UV-Vis spectrophotometer (UV-2600i, Shimadzu, Japan).

Enzyme-photo-coupled catalytic reaction

The enzyme-photo-coupled catalytic reactions of AmDH were carried out in 2 M NH₄Cl buffer (pH 8.7) containing 0.82 mg mL^-1 AmDH, 0.13 mM NADH (or 1 mM NAD⁺), 42.16 µM TCPP, 0.13 mM, 400 mM TEOA, and 5 mM 2-hexanone. The enzyme-photo-coupled catalytic reactions of PAMO were performed with 0.16 mg mL^-1 PAMO, 9.29 µM NADP, 42.16 µM TCPP, 0.13 mM M, 400 mM TEOA, 10 μM FAD, and 5 mM bicycle[3.2.0]hept-2-en-6-one in 2 M NH₄Cl buffer (pH 8.7). The addition of M, enzyme, and NAD(P)H in different systems were adjusted according to their loadings in APS/Am&N or APS/PA&NP to ensure equivalent mass participation in all AmDH- or PAMO-relative reactions. All reactions were conducted under the illumination of an LED lamp.

Statistics and reproducibility

All experiments were independently repeated at least three times with similar results unless otherwise stated. For experiments involving quantitative analysis, data are presented as mean ± s.d. from n = 3 independent biological replicates. Micrographs and other representative images shown in the figures are representative of three independent experiments with similar results.

Other experiment procedures see Supplementary Information.

Reporting summary

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

Author contributions

S.L. performed most of the experiments. S.Y.Z. performed the enzyme-photo-coupled catalytic experiments. S.L. and Y.S. analysed the data and wrote the original draft of the manuscript. Y.S. conceived and supervised the project. All authors discussed the results and revised the manuscript.

Peer review

Peer review information

Nature Communications thanks Guosheng Chen and the other anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information files, and from the corresponding author(s) upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-72050-2.