Spatial organization of an enzyme cascade in a Ni-ZIF-8 framework for efficient sugar nucleotide synthesis

Abstract

Enzyme cascade reactions hold transformative potential for sugar nucleotide biosynthesis, aligning with green chemistry principles by minimizing solvent waste and purification steps. However, their potential is often compromised by low catalytic efficiency due to inefficient intermediate utilization and instability of individual enzymes. Capitalizing on the coordinatively unsaturated Ni²⁺ sites and the hydrophilic nature of Ni-doped zeolitic imidazolate framework-8 (Ni-ZIF-8), we engineered a highly active nanocomposite by incorporating a hexahistidine-tagged dual-enzyme conjugate of N-acetylhexosamine 1-kinase (BlNahK) and N-acetylglucosamine 1-phosphate uridylyltransferase (PmGlmU) (His₆-BlNahK–PmGlmU-His₆) for efficient synthesis of uridine diphosphate N-acetylglucosamine (UDP-GlcNAc)—a pivotal sugar nucleotide. The Ni-ZIF-8 scaffold acts as a sustainable nano-reactor, not only stabilizing the dual-enzyme conjugate conformation but also elevating local substrate concentrations (ATP, UTP, and GlcNAc) via synergistic electrostatic and van der Waals interactions, thereby enhancing reaction kinetics and resource efficiency. Using stimulated Raman scattering (SRS) microscopy, we directly visualized the spatial confinement and rapid consumption of the intermediate substrate GlcNAc-1-P on the nanocomposite surface, demonstrating an engineered substrate channeling-like effect, a common mechanism in native metabolon complexes that boosts cascade efficiency. The resulting nanocomposite exhibits a 4.4-fold higher activity than free enzymes and superior stability across varying temperatures and pH conditions, and retains approximately 60% of its initial activity after five reuse cycles. This work establishes a generalizable and robust strategy for constructing metal–organic framework (MOF)–enzyme complexes with broad applicability in high-value sugar nucleotide biosynthesis and other complex bioconversion processes requiring metabolic flux control.

---

Green foundation

1. We present a bio-inspired enzyme–MOF nanocomposite for the efficient and sustainable biosynthesis of sugar nucleotides. By leveraging specific metal–enzyme affinity for precise spatial organization, our system mimics natural metabolons to minimize intermediate diffusion and energy waste, aligning with the core principles of green chemistry.

2. The nanocomposite demonstrates a 4.4-fold activity enhancement over free enzymes, achieving 92% conversion under mild conditions (37 °C, pH 8.0). Its superior thermal/pH stability and reusability (60% activity retained after five cycles) significantly reduce resource consumption. The substrate enrichment and channeling-like effect maximize the usage of expensive nucleoside triphosphates (ATP/UTP), drastically improving process economics and reducing waste.

3. This platform establishes a versatile strategy for complex biomanufacturing. Future work will extend it to other sugar nucleotide and oligosaccharide biosyntheses, integrate co-factor regeneration systems to enable fully sustainable and economically viable production, and explore continuous-flow reactors for industrial applications.

Introduction

Enzyme cascade reactions, which integrate multiple biocatalytic steps into a one-pot system, represent a paradigm of green chemistry by emulating the efficiency of natural metabolic pathways.^1–3 This strategy eliminates the need for intermediate isolation and purification, significantly reducing solvent waste, energy consumption, and overall processing costs. As such, enzymatic cascades have become pivotal tools for sustainable industrial biotechnology, facilitating the environmentally benign synthesis of high-value compounds such as fine chemicals, chiral agrochemical intermediates, pharmaceuticals and biofuels.^4,5 However, the broader implementation of these enzymatic cascades is constrained by several fundamental limitations, including intrinsic instability and poor recyclability of native enzymes and inefficient intermediate utilization in homogeneous reaction systems due to mass transport limitations and unfavourable reaction equilibria that compromise atom and energy economy.^6–8

In nature, multienzyme systems achieve remarkable catalytic efficiency through sophisticated spatial organization either as tightly clustered complexes or within specialized microcompartments.^9–11 This architectural precision minimizes metabolite leakage, and enables efficient substrate channeling, thus maximizing metabolic flux and reaction yields. Inspired by these biological blueprints, researchers have developed various artificial multienzyme platforms to optimize mass transfer by achieving the channeling effect.^12–14 Among them, enzyme material hybridization has emerged as a particularly robust strategy, offering significant advantages in terms of enhanced catalytic efficiency through engineered proximity effects and improved intermediate transferring and retention.¹⁵ In our previous work, we reported that the co-assembly of poly(4-vinylpyridine) (P4VP) and arylamine N-oxygenase Cmll facilitated the electron transfer which introduced significantly enhanced multistep monooxygenation reactions.¹⁶ Beyond this, a variety of materials ranging from biological polymers (e.g., chitosan and DNA)^17–19 to advanced inorganic nanomaterials (e.g., graphene scaffolds and perovskite oxides)^20–22 have been investigated for enzyme hybridization and immobilization.

Water-stable metal–organic frameworks (MOFs) have recently gained prominence as exemplary green platforms for enzyme encapsulation and immobilization by creating robust, recyclable, heterogeneous biocatalysts, aligning perfectly with the principles of sustainable chemistry, such as energy efficiency and waste prevention.^23–26 Their tunable architectures, extraordinary surface areas, and controllable porosity underpin these advantages, facilitating mass transferring, high enzyme loading, stabilization, and reuse.^27,28 These characteristics are exemplified by the successful co-encapsulation of glucose oxidase (GOx) and horseradish peroxidase (HRP) in zeolitic imidazolate framework-8 (ZIF-8), which has shown remarkable cascade catalytic efficiency.^29–31 Advancements like Ni²⁺-doped ZIF-8 (Ni-ZIF-8), fabricated via site-specific metal anchoring, further enhance the platform's credentials by improving protein incorporation efficiency and stabilizing enzymes like pepsin, thereby extending the catalyst lifespan and reducing material usage.^24,32,33 However, critical challenges impede the full realization of MOF-based enzyme systems as ideal green technologies. Key limitations include the random orientation, leaching of enzymes and reduced enzymatic activity due to non-specific physical adsorption³⁴ and undesirable enzyme conformational changes.³⁵ Moreover, significant mass transfer limitations arise from the typical sub-2 nm pore sizes of most MOFs resulting in substantially compromised enzymatic activity for larger substrates.²³ A fundamental understanding of the MOF's chemical environment and its effect on diverse enzymatic cascades remains incomplete, highlighting the critical need for structurally tailored, multi-functional MOF platforms designed explicitly to maximize biocatalytic performance and sustainability.^27,28

Sugar nucleotides, the activated forms of sugar molecules synthesized by multiple enzyme systems in nature, are essential glycosyl donors for the biosynthesis of structurally diverse glycans and glycoconjugates.³⁶ Of particular biological and pharmaceutical significance is uridine diphosphate N-acetylglucosamine (UDP-GlcNAc), a central metabolite that functions as the key substrate for glycosaminoglycan and proteoglycan biosynthesis,³⁷ and protein N-acetylglucosaminylation.³⁸ Conventional chemical synthesis and purification of these intricate molecules are notoriously inefficient, relying on hazardous reagents and protecting groups, and generating significant waste, spurring the development of multienzyme biocatalytic platforms for their production.^39–42 For example, Chen et al. firstly utilized the cascade reaction catalyzed by Pasteurella multocida N-acetylglucosamine 1-phosphate uridylyltransferase (PmGlmU) and N-acetylhexosamine 1-kinase (NahK_ATCC55813) for the synthesis of UDP-GlcNAc and its derivatives.⁴¹ Recently, some innovative approaches such as continuous flow biocatalysis⁴³ and engineered protein scaffolds have demonstrated improved production yields.⁴⁴ However, current cascade systems remain constrained by suboptimal reaction kinetics, which is especially important due to the substantial cost and stability issues associated with nucleoside triphosphate substrates (ATP and UTP). These fundamental challenges highlight the urgent need for novel biocatalytic engineering solutions to enable green and scalable production of this biologically important molecule.

In this work, we developed a novel biocatalytic platform for highly efficient and sustainable synthesis of UDP-GlcNAc by engineering a Ni-doped ZIF-8 (Ni-ZIF-8) scaffold for the site-specific incorporation of a hexahistidine tagged dual-enzyme conjugate (His₆-BlNahK–PmGlmU-His₆, abbreviated as B–P) (Fig. 1). Histidine in the enzyme complex functions as a pivotal coordinating ligand via its intrinsic molecular structure: its imidazole moiety enabling specific chelation with metal centers (e.g., coordinatively unsaturated Ni²⁺ in Ni-ZIF-8). This chelation interaction confers strong, targeted binding by mitigating non-specific adsorption, thereby ensuring stable conjugation between the enzyme complex and the Ni-ZIF-8 framework. As expected, the designs leverage Ni²⁺–His₆ affinity interactions to create a homogeneous nanostructure with exceptional enzyme loading capacity. The resulting nanocomposite exhibited 4.4-fold higher catalytic activity over free enzymes in solution during 30-minute reactions. Mechanistic studies integrating computational modeling and experimental data revealed that the Ni-ZIF-8 matrix acts as a sustainable nano-reactor, concentrating substrates (ATP, UTP, and GlcNAc) around its surface via synergistic electrostatic and van der Waals interactions. Critically, the intermediate product (GlcNAc-1-P) was efficiently sequestered within the nanostructures, promoting a channeling-like effect. This eliminates the diffusion of reactive intermediates, minimizing side reactions and maximizing the efficiency of the tandem catalysis. The nanocomposite platform's sustainability is further underscored by its robust operational stability, withstanding varied pH and temperature conditions, and retaining around 60% initial activity over five consecutive reaction cycles. By leveraging specific metal–enzyme affinity for precise spatial organization, we present a bio-inspired enzyme–MOF nanocomposite for the efficient and sustainable biosynthesis of sugar nucleotide. This system mimics natural metabolons to minimize intermediate diffusion and energy waste, aligning with the core principles of green chemistry.

Fig. 1 Schematic illustration of BlNahK(B)–PmGlmU(P)@Ni-ZIF-8 nanocomposite fabrication (a), and its application for highly efficient UDP-GlcNAc synthesis from nucleoside triphosphates (ATP and UTP) and GlcNAc substrates (b).

Results and discussion

Fabrication of B–P@Ni-ZIF-8 nanocomposites

The N-acetylhexosamine 1-kinase from Bifidobacterium longum JCM1217 (BlNahK, B)⁴⁵ and N-acetylhexosamine 1-phosphate uridylyltransferase from Pasteurella multocida (PmGlmU, P)⁴¹ have been employed in a cascade reaction to catalyze UDP-GlcNAc biosynthesis from ATP, UTP and GlcNAc. While functional, the inefficient utilization of costly nucleoside triphosphates remains a significant challenge that has been scarcely addressed. Noting that some microorganisms co-localize NahK and GlmU homologs within a single operon,⁴⁵ we hypothesized that the spatial organization could enhance cascade efficiency. To test this, we constructed a B–P dual-enzyme conjugate using SnoopCatcher/SnoopTag technology (Fig. S1–S4).⁴⁶ Each enzyme contained a hexahistidine affinity tag (His₆ tag) at its N-terminal. The SnoopCatcher and SnoopTag motifs were fused at the C-terminal of BlNahK and PmGlmU, respectively (Fig. S1 and S2). The engineered enzymes had their activity maintained (Fig. S5). The two enzymes were coupled readily via a spontaneous isopeptide bond formation upon gentle mixing (Fig. S4). Remarkably, the B–P dual-enzyme conjugate exhibited ∼50% higher activity compared to free enzymes (Fig. S6). Notably, this enhanced effect showed an inverse correlation with enzyme concentration (Fig. S7) suggesting that the proximity of the enzymes facilitates the tandem catalysis.

Building upon these findings, we propose to use Ni-doped ZIF-8 (Ni-ZIF-8) as an advanced scaffold for optimization of the cascade reaction based on two key rationales. Firstly, surface-exposed undercoordinated metal sites enable specific coordination with His₆-tagged enzymes, through affinity interactions ensuring proper spatial organization and conformation stability of the dual-enzyme conjugates (Fig. 1).^32,47 Secondly, the intrinsic surface charge and hydrophilic character of Ni-dopped MOFs may promote substrate enrichment and intermediate retention, effectively mimicking natural metabolic channeling through localized concentration effects that accelerate cascade reactions (Fig. 1).²⁴ Through systematic optimization of the metal–ligand stoichiometry, we determined an optimal molar ratio of 2-MIM : Zn²⁺ = 16 : 2 for the synthesis of ZIF-8. For the preparation of Ni-ZIF-8, the amounts of 2-MIM (16 equivalents) and total metal ions (Zn²⁺ + Ni²⁺, 2 equivalents) were kept constant, while the Zn²⁺ : Ni²⁺ ratio was varied. We found that a Zn²⁺ : Ni²⁺ ratio of 1 : 5 simultaneously maximized nanoparticle yield and quality. This formulation produced monodisperse particles with excellent colloidal stability (Fig. S8–S10). The enzyme incorporation was achieved through mild aqueous-phase incubation at 25 °C for 0.5 hours. The Ni-ZIF-8 carrier exhibited exceptional loading capacity (3.3 ± 0.2 mg enzyme per mg carrier). This high loading capacity may be attributed to the high availability of vacant sites of Ni²⁺ or co-ordinately unsaturated Zn²⁺ on the surface of the MOF as discussed below.³²

Nanostructure characterization of B–P@Ni-ZIF-8 composites

Ni-ZIF-8 is a porous material composed of zinc ions, nickel ions, and 2-MIM which features coordinatively unsaturated zinc and nickel sites on its surface.³² The better affinity of histidine to Ni²⁺ relative to Zn²⁺ is another key to making Ni-ZIF-8 an optimal choice for MOF–enzyme composition fabrication.^24,33 Leveraging this property, the dual enzyme–MOF nanocomposite was prepared by coordinating a His₆-tagged B–P with nickel ions in Ni-ZIF-8. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were used to investigate the microstructure of B–P@Ni-ZIF-8 (Fig. 2). SEM analysis (Fig. 2a and b) revealed monodisperse spherical nanoparticles averaging ∼100 nm in diameter. The observed increase in surface roughness relative to pristine Ni-ZIF-8 suggests successful enzyme incorporation. TEM characterization (Fig. 2c and d) corroborated the spherical morphology and additionally identified a distinct surface coating (Fig. 2e), consistent with a core–shell architecture where the enzyme layer encapsulated the MOF core. The elemental composition was analysed by energy-dispersive spectroscopy (EDS) mapping (Fig. 2f–l), revealing uniform distributions of C, N, Zn and Ni (from the Ni-ZIF-8 framework) alongside O (characteristic of enzymes). Quantitative EDS analysis determined the atomic composition to be 73.66% C, 9.41% N, 5.73% O, 9.44% Zn, and 0.61% Ni, confirming the expected composite structure.

Fig. 2 Characterization of Ni-ZIF-8 and B–P@Ni-ZIF-8. SEM images of Ni-ZIF-8 (a) and B–P@Ni-ZIF-8 (b). TEM images of Ni-ZIF-8 (c) and B–P@Ni-ZIF-8 (d and e). EDX mapping of B–P@Ni-ZIF-8 (f–l). FT-IR spectra of Ni-ZIF-8, B–P and B–P@Ni-ZIF-8 (m). XPS of Ni-ZIF-8 and B–P@Ni-ZIF-8 (n and o). Size (p) and zeta potential (q) analysis of Ni-ZIF-8, B–P and B–P@Ni-ZIF-8. Error bars represent the standard deviation of the mean of triplicate samples. TGA of Ni-ZIF-8, B–P and B–P@Ni-ZIF-8 (r).

The B–P@Ni-ZIF-8 composites were further characterized by using Fourier-transform infrared spectroscopy (FT-IR) and X-ray photoelectron spectroscopy (XPS) analyses. FT-IR spectra confirmed protein incorporation through characteristic bands: O–H/N–H stretches (3000–3500 cm⁻¹),⁴⁷ carboxylate vibrations (1630 cm⁻¹ν_as COO⁻; 1410 cm⁻¹ν_s COO⁻),⁴⁸ and changed C–N stretching (1050 cm⁻¹) indicative of π–π stacking between His₆-tags and the MOF framework (Fig. 2m).⁴⁹ Additionally, XPS analysis of B–P@Ni-ZIF-8 revealed a new peak at 287.8 eV, attributed to the OC–NH bond, which is absent in pristine Ni-ZIF-8. This confirms the successful incorporation of enzymes (Fig. 2n and o).⁵⁰

Dynamic light scattering (DLS) and zeta potential measurements revealed that Ni-ZIF-8 nanocomposites exhibited a hydrodynamic diameter of 170.6 ± 3.8 nm (Fig. 2p) and a positive surface charge (+29.0 ± 0.1 mV) (Fig. 2q). Upon enzyme incorporation, the resulting B–P@Ni-ZIF-8 nanocomposites showed a significant increase in hydrodynamic diameter (210.7 ± 5.5 nm) and complete reversal of surface charge compared to pristine MOF (−23.1 ± 0.4 mV). SEM further corroborated the size increase with functionalization of His₆ tagged enzymes (Fig. S11). The larger hydrodynamic size observed by DLS, compared to EM-derived sizes, likely reflects nanoparticle aggregation in solution.

The thermogravimetric analysis (TGA) results corroborate the successful incorporation of a large amount of enzyme into the Ni-ZIF-8 framework. As shown in Fig. 2r, all samples—B–P, Ni-ZIF-8, and B–P@Ni-ZIF-8—exhibited weight loss before 200 °C, which is attributed to the evaporation of solvent molecules. In the temperature range of 200 to 500 °C, B–P@Ni-ZIF-8 showed a mass loss of approximately 40%, corresponding to the thermal decomposition of the incorporated enzyme molecules. The free B–P enzyme exhibited a significant weight loss of about 55% within this temperature interval. Above 500 °C, Ni-ZIF-8 underwent a rapid mass reduction due to structural collapse and subsequent carbonization under intense thermal conditions.^51,52

In a competitive elution assay, high NaCl concentrations (500 mM) failed to release the enzymes, whereas imidazole (20–250 mM) triggered concentration-dependent enzyme dissociation from the nanocomposites (Fig. S12). This different elution behavior confirms that enzyme incorporation occurs primarily through specific His₆–metal affinity interactions rather than nonspecific electrostatic adsorption between B–P and Ni-ZIF-8. This affinity-based incorporation minimized the enzyme leaching during the reactions.

Ni and Zn coordination in B–P@Ni-ZIF-8: Ni-dominant metal–His₆ tag interactions

ZIF-8 has been employed as a scaffold for incorporating His₆-tagged proteins.^49,53 Although effective, this incorporation has been attributed to coordinatively unsaturated Zn²⁺ sites of defective MOFs, a process that often leads to the formation of uncontrollable nanostructures. To elucidate the underlying metal–enzyme interaction mechanism within our nanocomposite, we conducted a synchrotron X-ray absorption spectroscopy study including Ni and Zn K-edge X-ray absorption near-edge structure (XANES) spectra and Fourier-transformed extended X-ray absorption fine structure (EXAFS) spectra to probe the valence state and local atomic structure of the metal centres in B–P@Ni-ZIF-8.

As shown in Fig. 3a, XANES analysis reveals pronounced alterations in the Ni electronic structure upon enzyme incorporation. Both Ni-ZIF-8 and B–P@Ni-ZIF-8 exhibit prominent pre-edge features characteristic of Ni²⁺ in distorted octahedral or square-pyramidal coordination environments. Protein loading induces a ∼0.5–1 eV positive edge shift, enhanced white-line intensity, and a dramatic amplification of the pre-edge feature (∼8331 eV, 1s → 3d transition). These changes indicate augmented d-orbital hybridization and reduced local symmetry, consistent with the coordination of protein ligands to Ni sites. In contrast, the virtually unchanged Zn K-edge spectra of Ni-ZIF-8 relative to ZIF-8 confirm its framework integrity. Only very slight edge and pre-edge shifts were observed after B–P incorporation (Fig. 3b). Consistently, the EXAFS results show that protein binding disrupts long-range ordering at Ni sites (Fig. 3c and d), while Zn exhibits remarkable stability in both k- and R-space (Fig. 3e and f). This corroborates that Ni provides most specific protein anchoring sites, whereas Zn maintains the framework structure with minimal participation in protein binding. Wavelet transform (WT) analysis further reveals a minor Zn–N scattering contributing from the imidazolate linkers at R ≈ 1.5–2.0 Å after enzyme incorporation (Fig. 3g and S13a). B–P@Ni-ZIF-8 exhibits redistributed intensity in the k–R space and a disappearance of the Ni–Ni scattering feature at R ≈ 2.5–3.0 Å that is present in the parent material (Fig. 3h and S13b).

Fig. 3 Ni (a) and Zn (b) K-edge XANES spectra (insets are the magnified images); Ni (c and d) and Zn (e and f) EXAFS in the k-space of ZIF-8, Ni-ZIF-8 and B–P@Ni-ZIF-8, and WT-EXAFS of Ni-ZIF-8 and B–P@Ni-ZIF-8 with Zn (g) and Ni (h) edges, depicting k–R intensity contours.

Overall, this spectroscopic evidence confirms that in the B–P@Ni-ZIF-8 nanocomposite, Ni serves as the primary protein anchoring site, while Zn remains largely unperturbed and contributes to MOF structural integrity. These findings align with the observed high homogeneity of the nanocomposite's nanostructure and its corresponding high protein loading capacity.

Catalytic performance of the B–P@Ni-ZIF-8 nanocomposite

The catalytic activity of the B–P@Ni-ZIF-8 was quantitatively evaluated by monitoring UDP-GlcNAc production. Reactions were conducted under the following conditions: 5 µM loaded enzyme, 2 mM GlcNAc, 2.4 mM ATP/UTP, 1 mM MgCl₂, 100 mM Tris-HCl (pH 8.0), at 37 °C. The performance was directly compared to that of the free enzyme system. UDP-GlcNAc production was quantified using high-performance liquid chromatography (HPLC), as detailed in the Methods section in the SI (Fig. S14).

As shown in Fig. 4a, the free enzymes achieved a GlcNAc conversion of only ∼21% within 30 minutes. Conversion increased to ∼40% at 60 minutes but progressed sluggishly thereafter, reaching a mere ∼48% after 90 minutes. In contrast, the B–P@Ni-ZIF-8 catalyzed reaction reached a near-quantitative conversion (∼92%) within 30 minutes. Based on the initial reaction rates, the B–P@Ni-ZIF-8 nanocomposite exhibited a 4.4-fold enhancement in catalytic activity over the free enzymes.

Fig. 4 Catalytic performance comparison of B–P@Ni-ZIF-8, B + P@Ni-ZIF-8 and free enzymes. (a) Activity comparison based on UPD-GlcNAc production; (b) temperature; (c) pH; and (d) the recyclability test of B–P@Ni-ZIF-8 and B + P@Ni-ZIF-8. Error bars represent the standard deviation of the mean of duplicate reactions.

To investigate the origin of the enhanced catalytic efficiency, kinetics studies of BlNahK and PmGlmU—individually and with MOF incorporation—were conducted to evaluate the MOF scaffold's effect on each enzyme (Table S1). The k_cat/K_M value of BlNahK@Ni-ZIF-8 decreased significantly compared to that of free BlNahK (33.84 µM⁻¹ s⁻¹vs. 220.95 µM⁻¹ s⁻¹), primarily due to an increase in the K_M value (2.35 µM vs. 0.21 µM). This suggests that MOF incorporation substantially reduces the substrate affinity and accessibility to BlNahK. In contrast, the k_cat/K_M value of PmGlmU@Ni-ZIF-8 increased approximately 2.5-fold relative to the free enzyme (106.90 µM⁻¹ s⁻¹vs. 43.23 µM⁻¹ s⁻¹). This enhancement was attributed to both improved substrate affinity (K_M value decreased from 2.03 µM to 1.37 µM) and a higher turnover rate (K_cat increased from 87.94 s⁻¹ to 156.48 s⁻¹). It is noteworthy that the kinetic parameters of each enzyme within the B–P@Ni-ZIF-8 remained similar to those of the individually immobilized enzyme (21.34 µM⁻¹ s⁻¹vs. 33.84 µM⁻¹ s⁻¹ for BlNahK and 114.30 µM⁻¹ s⁻¹vs. 106.90 µM⁻¹ s⁻¹ for PmGlmU). This indicates that the enzyme conjugation did not alter the behavior of either enzyme, while the MOF scaffold modulates the specific activity of individual enzymes distinctly. Interestingly, the overall catalytic efficiency of B–P@Ni-ZIF-8 was considerably greater than that of an individually immobilized enzyme, suggesting the presence of an additional mechanism facilitating tandem catalysis.

Furthermore, we co-immobilized the BlNahK and PmGlmU (at a 1 : 1 ratio without conjugation, abbreviated as B + P) on Ni-ZIF-8 to construct B + P@Ni-ZIF-8. This composite exhibited intermediate activity, achieving approximately 72.0% conversion from GlcNAc to UDP-GlcNAc in a 30-minute reaction. Its activity was about 2.6-fold higher than that of the free enzymes, yet remained substantially lower than the performance of B–P@Ni-ZIF-8 (Fig. 4a). These results indicate that conjugating the two enzymes prior to incorporation into Ni-ZIF-8 is critical for maximizing cascade efficiency, underscoring the importance of controlled enzyme proximity for effective tandem catalysis within the MOF environment.

The temperature and pH stability of B–P@Ni-ZIF-8 were evaluated and compared with those of B + P@Ni-ZIF-8 and the free enzymes. As shown in Fig. 4b, the optimal reaction temperature for all samples was 45 °C, at which the activity was defined as 100%. Both B–P@Ni-ZIF-8 and B + P@Ni-ZIF-8 exhibited significantly improved thermal stability compared to the free enzymes. At 70 °C, the relative activity of the free enzyme dropped below 20%, while that of B–P@Ni-ZIF-8 remained above 70%. Even at 80 °C, B–P@Ni-ZIF-8 still retained 25% of its original activity. As illustrated in Fig. 4c, B–P@Ni-ZIF-8 maintained over 95% relative activity at pH 9. In contrast, the free enzymes suffered substantial activity loss under this alkaline condition, declining to below 50%. These results demonstrate that B–P@Ni-ZIF-8 possesses significantly enhanced resistance to both high temperature and alkaline environments compared to the free enzymes.

Additionally, with a dialysis approach, we identified that B–P@Ni-ZIF-8 demonstrated improved recyclability. After five consecutive reaction cycles, the system maintained around 60% of its initial activity, which is significantly higher than B + P@Ni-ZIF-8 (Fig. 4d) and B + P (Fig. S15). This high reusability and structural integrity can be attributed to the strong coordination between the His-tagged enzymes and the Ni nodes in the ZIF-8 framework, which helps maintain enzyme stability over repeated uses.

Although significantly outperforming the free enzyme, the B–P@Ni-ZIF-8 nanocomposite's activity decreased under harsh conditions including high temperature and alkaline pH and during the recycles. The decrease in enzyme activity may be attributed to the instability of the defective MOF structure, the subtle enzymes’ conformational changes or substrate diffusion limitations due to aggregations during the reactions. Nevertheless, among the reported strategies for enzymatic sugar nucleotide production,^{41,43,44,54,55} our system exhibits distinct advantages in catalytic efficiency and conversion rate. As summarized in Table S2, whereas some state-of-the-art multienzyme systems require prolonged reaction times (16–48 h) to achieve above 95% conversion,^43,55 our system attains 92% conversion within 30 min. Furthermore, our nanocomposite demonstrates superior recyclability compared to other reusable systems: for instance, the protein-scaffolded PSK-(G4S)3-CipA retains its initial activity after 5 cycles but shows a 1.65-fold catalytic efficiency enhancement,⁴⁴ while the commonly used carrier EziG™ maintains >50% activity over 5 cycles under flow reaction conditions.⁴³ In contrast, our nanocomposite retains approximately 60% of its initial activity after 5 consecutive batch reactions, combined with its significantly higher catalytic rate, highlighting its great potential for scalable applications, particularly when integrated with flow chemistry, an area currently under investigation in our laboratory.

Computational evidence for MOF-enhanced dual-enzyme conjugate conformation and substrate localization

To elucidate the mechanistic advantages of the MOF environment in facilitating enzyme cascade reactions, we employed an integrated computational approach combining quantum chemistry-based docking and molecular dynamics (MD) simulations to analyse the dynamic behaviour of dual-enzyme conjugate and small molecular substrates in the MOF environment (Fig. 5a) compared to a solution system, as detailed in the Experimental section.

Fig. 5 Effect of MOF environment on the B–P dual-enzyme conjugate conformation and substrate localization. (a) Structural model of the B–P@Ni-ZIF-8; (b) RMSD values of the B–P conjugate of B–P@Ni-ZIF-8 and free B–P conjugate in solution. (c) Distance variation between B and P enzymes of B–P@Ni-ZIF-8 and free B–P in solution. (d–f) RDF analysis of ATP, GlcNAc, and UTP binding to the B–P conjugate of B–P@Ni-ZIF-8 and free B–P in solution. (g–i) Corresponding energy analysis of ATP, GlcNAc, and UTP substrate–biocomposite interactions.

Structural dynamics analysis revealed that the MOF environment significantly enhanced the conformational stability of the dual-enzyme conjugate. As shown in Fig. 5b, the B–P conjugate in B–P@Ni-ZIF-8 exhibited faster convergence of root-mean-square deviation (RMSD) to equilibrium, with substantially reduced fluctuations (1.548 ± 0.077 nm) compared to the free enzyme system (1.126 ± 0.269 nm) from 50 nm to 200 nm, indicating superior dual-enzyme structural maintenance and minimal conformational changes in the MOF environment. Consistently, the radius of gyration (RG) of B–P conjugate in B–P@Ni-ZIF-8 was reduced significantly relative to that in solution and remained stable between 50 and 200 ns (Fig. S16). Inter-enzyme distance analysis further revealed that the B–P conjugate in B–P@Ni-ZIF-8 maintained a tightly controlled distance of 5.35 nm (±0.29 nm), whereas the conjugate in solution exhibited a broader distance distribution ranging from 4.81 to 6.19 nm (Fig. 5c). This reduction in inter-enzyme distance of the nanocomposite suggests strong spatial confinement effects that optimally position the enzymes for efficient tandem catalysis, as discussed below.

To evaluate substrate spatial organization, radial distribution function (RDF) analysis was performed.⁵⁶ Compared to aqueous solution, significantly higher RDF intensities were observed for ATP and UTP within the MOF environment (Fig. 5d and f), demonstrating preferential accumulation of these substrates around the nanocomposite. Although less pronounced, GlcNAc also exhibited elevated RDF values in the MOF system (Fig. 5e), suggesting the involvement of hydrogen-bonding and/or van der Waals interactions facilitating its localization within the MOF architecture.

To further elucidate the molecular basis of enhanced substrates–nanocomposite interactions, we conducted a comprehensive free energy decomposition analysis (Table S3). The total binding free energy (ΔG_bind) revealed significantly stronger substrate affinity in the MOF system versus the solution phase (Fig. 5g–i). These findings were corroborated by the increased hydrogen bond formation (HBNUM) between enzymes and substrates within B–P@Ni-ZIF-8 compared to free enzymes in solution (Fig. S17). These results imply that the MOF framework promotes a local substrate-concentrating effect, thereby enhancing the frequency of enzyme–substrate interactions.

Zeta potential measurements revealed significant shifts in the surface charge of Ni-ZIF-8 upon exposure to ATP, UTP, and GlcNAc (Fig. S18), providing direct evidence of substrate adsorption onto the MOF. Collectively, these results demonstrate that the nanocomposite synergistically enhances enzymatic turnover by stabilizing B–P dual-enzyme conjugates and increasing the substrate local concentration.

Mechanistic insights into the channeling effect in B–P@Ni-ZIF-8

In a typical solution-phase cascade reaction, UDP-GlcNAc production is constrained by the rate-limiting activity of PmGlmU (Table S1), resulting in substantial accumulation of GlcNAc-1-P. We found that reactions catalyzed by B–P@Ni-ZIF-8 maintained a significantly lower concentration of GlcNAc-1-P compared to the free enzyme system (Fig. 6a). This implies a rapid turnover of the intermediate product within the nanocomposite, which cannot be fully contributed to the enhanced single-enzyme performance upon MOF incorporation, as discussed previously. We hypothesize that B–P@Ni-ZIF-8 facilitates the tandem catalysis by restricting diffusion of GlcNAc-1-P from the nanocomposite into the bulk solution, thereby promoting the cascade reaction via a channeling-like effect on its surface (Fig. 6b).

Fig. 6 Spatial and temporal analysis of the intermediate product GlcNAc-1-P during the cascade reaction and the working model. (a) HPLC quantification of GlcNAc-1-P in the solution during reactions catalysed by the free B–P dual-enzyme conjugate versus B–P@Ni-ZIF-8. Error bars represent the standard deviation of the mean of duplicate reactions. (b) Schematic of the channeling-like effect of GlcNAc-1-P. (c and d) SRS microscopy analysis of GlcNAc-1-P on the B–P@Ni-ZIF-8 surface catalysed by BlNahK. (e and f) In situ analysis of the rapid consumption of GlcNAc-1-P on the B–P@Ni-ZIF-8 surface during the cascade reaction.

To test this hypothesis, we performed MD simulations to analyse the association of GlcNAc-1-P with B–P@Ni-ZIF-8. The simulations revealed strong preferential binding of GlcNAc-1-P to the nanocomposite surface (ΔG = −31.7 kJ mol⁻¹), driven by both electrostatic (79%) and van der Waals (21%) forces (Table S3 and Fig. S19). We next employed stimulated Raman scattering (SRS) microscopy to spatially and temporally resolve the distribution of GlcNAc-1-P on B–P@Ni-ZIF-8 during reactions, leveraging its distinct spectral signature at 2750–3075 cm⁻¹ (Fig. 6c–e). When only BlNahK was active (in the presence of GlcNAc and ATP, but without UTP), the GlcNAc-1-P signal on the nanocomposite surface increased markedly (∼250% from 5 to 30 minutes reaction; Fig. 6c and d). In contrast, when both BlNahK and PmGlmU were active (in the presence of GlcNAc, ATP and UTP), the GlcNAc-1-P signal was an order of magnitude lower and showed only slight increase (∼30%) over 30 minutes of reaction (Fig. 6c and d). These results confirmed the rapid consumption of the intermediate product by the nanocomposite. Furthermore, the addition of free PmGlmU and UTP to the reaction did not significantly alter the surface-associated GlcNAc-1-P signal (Fig. S20), confirming its limited diffusion into the solution. Collectively, these results demonstrate the rapid consumption of GlcNAc-1-P on the B–P@Ni-ZIF-8 surface via a channeling-like mechanism (Fig. 6b), which enhances the cascade reaction efficiency analogous to a native metabolon.

Conclusion

In summary, this study reports a highly efficient nanocomposite for UDP-GlcNAc synthesis. It is constructed by incorporating the BlNahK–PmGlmU dual-enzyme conjugate onto Ni-ZIF-8 via His₆ tag–metal affinity interactions. The resulting nanocomposite exhibits superior enzyme loading capacity, excellent catalytic efficiency—achieving approximately 4.4-fold higher activity than the free enzyme system—good recyclability and enhanced tolerance across a broad range of temperature and pH conditions. Through comprehensive characterization, we identified Ni (rather than Zn) as the dominant metal involved in the nanocomposite formation, which underpins the high enzyme loading capacity. The MOF scaffold not only stabilizes the conformation of the B–P conjugate but also attracts and sequesters substrates, both of which contribute to the enhanced catalytic performance. Furthermore, using stimulated Raman scattering (SRS) microscopy, we directly visualized the channeling effect of the reaction intermediate GlcNAc-1-P around the nanocomposite, providing the first experimental evidence for such an effect in this context. We propose that the bio-nanocomposite mimics a native metabolon by confining substrates and minimizing intermediate leakage, thereby facilitating efficient cascade catalysis. This platform offers a versatile and scalable approach for efficient biosynthesis of sugar nucleotides, with broad applicability to other multi-enzyme systems requiring precise metabolic control. Moreover, our findings establish a facile design principle for Ni-ZIF-8-based multienzyme systems, advancing their potential in complex biocatalytic applications.

Experimental

Protein design and cloning

The genes of BlNahK and PmGlmU enzymes,^41,57 along with SnoopCatcher (GenBank accession no. KU500646),⁵⁸ were synthesized by Sangon Biotech and cloned into the pET-22b vector. The BlNahK and PmGlmU genes were inserted between the NdeI and XhoI restriction sites, while the SnoopCatcher gene was flanked by BamHI and EcoRI sites. To construct the His₆-BlNahK–SnoopCatcher fusion, BlNahK was amplified using primers containing BamHI and EcoRI sites (primer sequences are provided in Table S4) and subsequently cloned into the pET-22b-SnoopCatcher plasmid. For the His₆-PmGlmU–SnoopTag construct, PmGlmU was amplified with primers incorporating NdeI and XhoI sites (see Table S4), and the resulting fragment was inserted into the pET-22b vector. The reverse primer included an encoding sequence for a C-terminal SnoopTag (GenBank: KU356870; residues 734–745: KLGDIEFIKVNK,⁵⁸ in bold). Both constructs feature an N-terminal His₆ tag, with SnoopCatcher or SnoopTag fused C-terminally to BlNahK or PmGlmU, respectively.

Protein expression and purification

His₆-BlNahK, His₆-BlNahK–SnoopCatcher, His₆-PmGlmU, and His₆-PmGlmU–SnoopTag were overexpressed in E. coli BL21 cells. Briefly, single colonies were inoculated into 10 mL of LB liquid medium containing 100 μg mL⁻¹ ampicillin and cultured overnight at 37 °C with shaking at 220 rpm. The next day, cultures were diluted 1 : 50 into fresh LB medium and grown at 37 °C with shaking at 220 rpm until the optical density at 600 nm (OD₆₀₀) reached 0.6–0.8. Protein expression was induced by adding IPTG to a final concentration of 0.1 mM, followed by incubation at 16 °C with shaking at 180 rpm for 16 h. Cells were harvested by centrifugation at 4000 rpm for 10 min at 4 °C, and the pellet was resuspended in Buffer A (50 mM Tris-HCl, 300 mM NaCl, pH 7.5) containing 0.1% Triton X-100. The suspension was sonicated on ice for 20 min to lyse cells (the ultrasonic conditions were set to 600 W power with a 5 s on and 5 s off cycle), and cell debris was removed by centrifugation at 10 000 rpm for 30 min at 4 °C. The supernatant was subjected to SDS-PAGE analysis and Ni column affinity chromatography. Bound proteins were eluted with a gradient elution with increasing concentrations of imidazole from 10 mM to 500 mM, dialyzed overnight against Buffer A, and quantified using a BCA protein assay kit. Purified proteins were stored at −80 °C with 10% glycerol as a cryoprotectant.

Reaction of His₆-BlNahK–SnoopCatcher and His₆-PmGlmU–SnoopTag to make a dual-enzyme conjugate (B–P)

His₆-BlNahK–SnoopCatcher and His₆-PmGlmU–SnoopTag were purified from Escherichia coli using nickel-affinity resin. The dual-enzyme conjugate B–P was subsequently prepared according to a previously described method.⁵⁹ Briefly, 20 μM His₆-BlNahK–SnoopCatcher was incubated with 20 μM His₆-PmGlmU–SnoopTag in 100 mM Tris-HCl (pH 8.0) for 4 h at room temperature. The reaction mixture was subsequently subjected to Ni column affinity chromatography using a linear imidazole gradient (10–500 mM) to isolate the dual-enzyme conjugate B–P, followed by verification via SDS-PAGE.

Synthesis of ZIF-8 and Ni-ZIF-8

ZIF-8 and Ni-ZIF-8 nanoparticles were synthesized following previously reported methods with slight modifications.^32,60 For ZIF-8 preparation, 0.16 M of 2-methylimidazole (2-MIM, 16 equivalents) were dissolved in 5 mL of methanol. Separately, different amounts of zinc nitrate hexahydrate (0.1, 0.5, 1, 2, 5, and 10 equivalents, M/M) were each dissolved in 5 mL of methanol. The 2-MIM solution was then added dropwise to the zinc nitrate solutions under vigorous stirring. After stirring for 4 hours, the products were washed three times with methanol, collected by centrifugation, and dried overnight at 50 °C.

Ni-ZIF-8 samples were synthesized following the same procedure, but with zinc nitrate hexahydrate replaced by mixtures of zinc nitrate hexahydrate and nickel nitrate hexahydrate. The total metal ion concentration was maintained at 0.02 M (2 equivalents), while the concentration of 2-MIM was kept constant at 0.16 M (16 equivalents). Six different samples were prepared with varying Zn²⁺ : Ni²⁺ molar ratios of 1 : 0.5, 1 : 1, 1 : 2.5, 1 : 5, 1 : 10, and 1 : 20.

Synthesis of B–P@Ni-ZIF-8

Ni-ZIF-8 (2 mg) was dispersed in 1 mL of a B–P solution (100 μM) and stirred at room temperature for 1 hour. The precipitate was collected and washed three times with deionized water. The resulting B–P@Ni-ZIF-8 composite was resuspended in 1 mL of solvent for subsequent use. The loading efficiency of B–P was calculated based on the difference in protein concentration before (C₀) and after (C) loading, as follows:

Loading rate (%) = [(C₀ − C)/C₀]× 100 (1)

Characterization

The microstructural and morphological characteristics of the samples were investigated using field-emission scanning electron microscopy (FE-SEM, TESCAN CLARA) and transmission electron microscopy (TEM, JEM-F200, and JEOL), with high-resolution TEM (HRTEM) images acquired at magnifications of up to 1.5 million times. Elemental composition was analyzed by energy-dispersive X-ray spectroscopy (EDS) coupled with TEM and X-ray photoelectron spectroscopy (XPS, Thermo Fisher K-Alpha Plus) for surface chemical state analysis. Fourier-transform infrared (FT-IR) spectra were recorded on a Thermo Scientific Nicolet iS5 spectrometer using KBr pellets. The particle size and zeta potential were measured by dynamic light scattering (DLS) on a Zetasizer Nano-ZS, with results presented as the average of three measurements. Thermal stability was assessed via thermogravimetric analysis (TGA, NETZSCH TG 209 F3 Tarsus) under a nitrogen atmosphere from 30 to 800 °C at a heating rate of 10 °C min⁻¹.

X-ray absorption spectroscopy (XAS) measurements

XAS data at the Ni K-edge (∼8333 eV) and Zn K-edge (∼9660 eV) were collected at the Synchrotron Light Research Institute (SLRI) in Thailand, using Beamline 1.1 W in transmission and fluorescence mode. The synchrotron operated at 1.2 GeV with a beam current of ∼100–150 mA. Samples were measured at room temperature under ambient conditions. Reference spectra for standards (Ni foil, NiO, Ni(OH)₂, Zn foil, ZnO, and pure ZIF-8) were acquired simultaneously for energy calibration. Multiple scans (3 times per sample) were averaged to improve the signal-to-noise ratio.

Raw XAS data were processed using the Demeter package software, version 0.9.26. Energy calibration was performed by aligning the first inflection point of the metal foil references. Background subtraction employed the AUTOBK algorithm with R_bkg = 1.0 Å and spline range adjusted to k = 0–12 Å⁻¹. Normalization was achieved over the post-edge region (150–800 eV above the edge). EXAFS signals (χ(k)) were extracted with k³-weighting, and Fourier transforms were computed in the k-range of 2–10 Å⁻¹ (for Zn) or 2–12 Å⁻¹ (for Ni) using a Hanning window (dk = 1.0). Wavelet transforms were generated using a custom or integrated tool (hama, based on the Morlet wavelet function) to decompose signals into the k–R space, with parameters η = 6 and σ = 1 for resolution optimization. No quantitative fitting was performed in this preliminary analysis; comparisons were qualitative, focusing on shifts in edge positions, peak intensities, oscillations, and ridge features.

Enzyme activity assay

The activities of free enzymes, B–P, and B–P@Ni-ZIF-8 were determined as reported previously with slight modification.⁴¹

For the activities of His₆-BlNahK or His₆-BlNahK–SnoopCatcher, the reactions were carried out as follows: 2 mM GlcNAc, 2.4 mM ATP, 1 mM MgCl₂, and 1 μM His₆-BlNahK or His₆-BlNahK–SnoopCatcher in 100 mM Tris-HCl buffer, pH 8.0, in a 200 μL reaction mixture were incubated at 37 °C for 30 min. Ethanol (200 μL) was added to quench the reactions. The reaction mixtures were centrifuged at 4 °C, 13 000 rpm for 10 min to collect the supernatant for analysis.

For the activities of His₆-PmGlmU or His₆-PmGlmU–SnoopTag, the reactions were carried out as follows: 2 mM GlcNAc-1-P, 2.4 mM UTP, 1 mM MgCl₂, and 1 μM His₆-PmGlmU or His₆-PmGlmU–SnoopTag in 100 mM Tris-HCl buffer, pH 8.0, in a 200 μL reaction mixture were incubated at 37 °C for 30 min. Ethanol (200 μL) was added to quench the reactions.

For the activity tests of B + P, B–P, B + P@Ni-ZIF-8, and B–P@Ni-ZIF-8, the reactions were carried out as below: 2 mM GlcNAc, 2.4 mM ATP, 2.4 mM UTP, 1 mM MgCl₂, and 100 mM Tris-HCl buffer, pH 8.0, supplemented with B + P (1 μM each enzyme), B–P (1 μM dual-enzyme), B + P@Ni-ZIF-8 (1 μM each enzyme), or B–P@Ni-ZIF-8 (1 μM B–P dual-enzyme), respectively.

The reactions were incubated for 5, 15, 30, 60, 90 and 120 min and quenched. Samples were analyzed using an Agilent 1260II LC HPLC system equipped with a 5 µm Dionex CarboPac™ PA-100 BioLC™ column (4 × 250 mm, analytical). Mobile phases consisted of A (0.1% acetic acid in water) and B (0.5 M NaCl) at a flow rate of 1 mL min⁻¹. Analytes were detected at 205 nm and 245 nm using a gradient elution program: 0–15 min, 100–50% phase A; 15–25 min, 50–0% phase A; followed by a 5 min re-equilibration to initial conditions. All reactions were performed in duplicate.

Stability determination of B + P, B + P@Ni-ZIF-8 and B–P@Ni-ZIF-8

Thermal stability. The relative activities of B + P (5 μM each), B + P@Ni-ZIF-8 (5 μM each) and B–P@Ni-ZIF-8 (5 μM) were calculated after dispersion in 100 mM Tris-HCl buffer at different temperatures (35, 45, 60, 70, and 80 °C) for 30 min and quenched.

pH stability. The relative activities of B + P (5 μM each), B + P@Ni-ZIF-8 (5 μM each) and B–P@Ni-ZIF-8 (5 μM) were maintained in 100 mM Tris-HCl buffer at different pH values (7–9) for 30 min and quenched.

After the above treatment, reaction conditions were set as follows: 2 mM GlcNAc, 2.4 mM ATP, 2.4 mM UTP, 1 mM MgCl₂, 100 mM Tris-HCl buffer, pH 8.0 in a 200 μL reaction mixture.

Determination of enzymatic kinetic parameters

Enzymatic kinetic parameters—Michaelis constants (K_M), maximum reaction rates (V_max), turnover numbers (k_cat), and catalytic efficiencies (k_cat/K_M)—for the target enzyme (B, P, B@Ni-ZIF-8, P@Ni-ZIF-8, or B–P@Ni-ZIF8) were determined via HPLC using the activity assay described above. Substrate concentrations were varied as follows: ATP (0.5 to 2.5 mM) with fixed 3 mM GlcNAc for enzyme B, or UTP (0.5 to 2.5 mM) with fixed 3 mM GlcNAc-1-P for enzyme P. Reaction mixtures contained 1 mM MgCl₂, 100 mM Tris-HCl buffer (pH 8.0), 5 μM enzyme, and water to a final volume of 200 μL, incubated at 37 °C for 10 min and quenched. K_M and V_max values were calculated by fitting experimental data to the Michaelis–Menten equation:

V = V_max·[S]/(K_M + [S]) (2)

where [S] stands for the concentration of the target substrate and V stands for the reaction rate at this concentration of the substrate.

Competitive elution assay

Appropriate amounts of B + P@Ni-ZIF-8 and B–P@Ni-ZIF-8 composites were washed 3 times with PBS buffer, centrifuged to remove the supernatant, and resuspended. High-salt solution (0.5 M) and imidazole solutions (20 mM, 100 mM, 250 mM) were added, followed by incubation for 30 min. After centrifugation, the supernatant was analyzed by protein gel electrophoresis to observe the presence of B + P/B–P bands.

Enzyme recyclability assay

To evaluate the enzymatic stability of B + P, B + P@Ni-ZIF-8 and B–P@Ni-ZIF-8, reactions were conducted under the following conditions: 2 mM GlcNAc, 2.4 mM ATP, 2.4 mM UTP, 1 mM MgCl₂, and 100 mM Tris-HCl buffer (pH 8.0), to which either B + P (5 μM each), B + P@Ni-ZIF-8 (5 μM each) or B–P@Ni-ZIF-8 (5 μM of B–P) was added in a total reaction mixture volume of 1 mL. After 30 minutes, 200 μL aliquots of the reaction mixture were collected and quenched for analysis. The remaining mixture was dialyzed at 4 °C against fresh buffer under gentle agitation (300 rpm) using a dialysis membrane with a 3 kDa molecular weight cut-off (MWCO) to remove residual substrates and products. Prior to each subsequent cycle, the volume removed was replaced with an equivalent amount of fresh substrates. This procedure was repeated for a total of five cycles.

Model construction for the MD study

The structure of Ni-ZIF-8 was constructed and optimized using CP2K with the PBE method, employing the DZVP-MOLOPT-SR-GTH basis set and pseudopotentials.⁶¹ The optimized structure was subsequently used for docking studies.

The protein structure was predicted using AlphaFold2.⁶² The B–P conjugate was constructed based on the transamination mechanism of the SnoopCatcher and SnoopTag system. The sequences of His₆-BlNahK–SnoopCatcher and His₆-PmGlmU–SnoopTag are as shown in Fig. S1–S3.

Quantum chemistry-based docking

The automated Interaction Site Screening (aISS) module in the XTB quantum chemistry program⁶³ was used for efficient conformational sampling.⁶⁴ A multi-level theoretical approach was employed to systematically explore the potential energy surface of the molecular system. First, xTB-IFF was used to automate the screening of interaction sites. The aISS module systematically sampled intermolecular interaction sites using a genetic algorithm. GFN2-xTB was applied for full geometry optimization of the pre-screened structures.⁶⁵

Molecular dynamics simulations

Gromacs 2021.3 ⁶⁶ was used with the AMBER14SB force field for the protein and GAFF2 for small molecules.⁶⁷ The system, solvated in TIP3P water with ATP/GlcNAc near enzyme B and GlcNAc-1-P/UTP near enzyme P, underwent vibrational analysis (CP2K) and force field parameterization (Sobtop https://sobereva.com/soft/Sobtop). Simulations employed Verlet/CG algorithms with PME electrostatics (1.2 nm cutoff), 1.4 nm non-bonded cutoffs, and steepest descent minimization (50 000 steps). After NVT/NPT equilibration (30 ps each at 300 K/1 bar, using V-rescale/Berendsen coupling), a 200 ns production MD was run under an external electric field (LINCS constraints, 2 fs timestep). Trajectories were analyzed using RMSD (threshold = 0.2 nm) and R_g (compactness), with visualization performed in PyMOL 2.5.1 ⁶⁸ and LigPlot + 2.1.⁶⁹

SRS microscopy

For label-free imaging of nanoparticles, hyperspectral SRS imaging was performed using a commercial multimodal nonlinear optical microscopy system (UltraView, Zhendian (Suzhou) Medical Technology Co., Ltd, Suzhou, China). One of the laser beams for SRS is a tunable pump beam (680–1300 nm); the other is a Stokes beam with a fixed wavelength (1045 nm, modulated at a 2.7 MHz resonant frequency). For all experiments, the power after the galvo was 40 mW for 801 nm and 200 mW for 1045 nm. The microscope was equipped with an objective of UPLSAPO 60× water (NA = 1.2). The spectrum at 2750–3075 cm⁻¹ was scanned with a step size of 3 cm⁻¹.

LASSO (least absolute shrinkage and selection operator)

For each hyperspectral SRS image stack captured, a pixel-wise spectral unmixing was performed using the LASSO (least absolute shrinkage and selection operator). For every single pixel, the LASSO decomposes its spectrum into the combination of several pure chemicals: D = CS + E. D is the detected signal. C is the decomposed concentration. S is the measured spectral profiles of pure chemicals. E is the residual term.⁷⁰

In this experiment, the hyperspectral SRS image stack of sample B–P@Ni-ZIF-8 was D, and two spectral profiles were chosen as S including Ni-ZIF-8 and B–P; the hyperspectral SRS image stack of sample B–P@Ni-ZIF-8 + GlcNAc-1-P was D, and three spectral profiles were chosen as S including Ni-ZIF-8, B–P and GlcNAc-1-P. Other samples that need to be unmixed using the LASSO can be processed in the same way.

Author contributions

Youbo Yu: data curation, formal analysis, methodology, software, validation, writing – original draft, and writing – review & editing. Xingcheng Zhou: validation, visualization, and writing – review & editing. Lisha Yan: validation, visualization, and writing – review & editing. Li Yang: software, validation, and visualization. Jie Hong: conceptualization, formal analysis, methodology, software, validation, writing – original draft, and writing – review & editing. Krongthong Kamonsuangkasem: methodology, validation, and visualization. Peiyi Wang: methodology, supervision, validation, and visualization. Kandegama Wishwajth: methodology, validation, and visualization. Gefei Hao: methodology, supervision, validation, and visualization. Libo Zhang: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, and writing – original draft.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data that support the findings of this study are available from the corresponding authors on reasonable request. Additional data that support the findings of this study are available in the supplementary information (SI). Supplementary information: figures and tables. See DOI: https://doi.org/10.1039/d5gc06007a.

Acknowledgements

This work was supported by the National Key Research and Development Program of China (No. 2024YFE0214300), the Post-Subsidy Fund Project of Guizhou International Science and Technology Cooperation Base (Qiankehe Platform GHJD (2025) 008), the National Natural Science Foundation of China (NSFC) Excellent Young Scientists Fund (Overseas, L. Z.) and the starting grant of Guizhou University (L. Z.).

References

1. E. T. Hwang and S. Lee, Multienzymatic Cascade Reactions via Enzyme Complex by Immobilization, ACS Catal., 2019, 9(5), 4402–4425,  DOI:10.1021/acscatal.8b04921.
2. M. W. Ullah, S. Manan, M. Ul-Islam, W. A. Khattak, K. A. Khan, J. Liu, G. Yang and J. Sun, Cell-free systems for biosynthesis: towards a sustainable and economical approach, Green Chem., 2023, 25(13), 4912–4940,  10.1039/D3GC00985H.
3. D. Yi, T. Bayer, C. P. S. Badenhorst, S. Wu, M. Doerr, M. Höhne and U. T. Bornscheuer, Recent trends in biocatalysis, Chem. Soc. Rev., 2021, 50(14), 8003–8049,  10.1039/D0CS01575J.
4. H. E. Schoemaker, D. Mink and M. G. Wubbolts, Dispelling the Myths–Biocatalysis in Industrial Synthesis, Science, 2003, 299(5613), 1694–1697,  DOI:10.1126/science.1079237.
5. A. I. Benítez-Mateos, D. R. Padrosa and F. Paradisi, Multistep enzyme cascades as a route towards green and sustainable pharmaceutical syntheses, Nat. Chem., 2022, 14(5), 489–499,  DOI:10.1038/s41557-022-00931-2.
6. H. Luo, T. P. Lai, Y. Gao, W. Li, Y. Fan and L. Zhang, Engineered pH-buffering nanocomposite for robust one-pot multi-enzyme cascade synthesis of γ-aminobutyric acid, Green Chem., 2025, 27(27), 8216–8225,  10.1039/D5GC02022K.
7. M. E. Hassan, X. Zhu, E. F. de Souza, M. M. Elnashar and F. Lu, Enzyme immobilization advances: a key to unlocking renewable bioenergy potential, Green Chem., 2025, 27(37), 11289–11311,  10.1039/D5GC03388H.
8. W. Jiang, X. Guan, W. Liu, Y. Li, H. Jiang and T. Ngai, Rotating Liquid Marble Microreactors for Enhanced Enzymatic Reactions, Adv. Funct. Mater., 2025, 2504193,  DOI:10.1002/adfm.202504193.
9. J.-D. Zhuang, J.-M. Shi, A.-M. Lu, L. Liu and J. Voglmeir, ATP-independent enzymatic cascade for chitin-to-glucose bioconversion, Green Chem., 2025, 27(43), 13863–13870,  10.1039/D5GC04343C.
10. R. Xin, Y. Wang, Q. Chen, J. Yang, Y. Wang and G. Luo, Boosting multi-enzyme cascade activity for glucose biosynthesis by kinetics-oriented grouped immobilization, Green Chem., 2025, 27(37), 11380–11391,  10.1039/D5GC03123K.
11. X. Gao, G. Tang, J. Yan, S. Fang, K. Tian, F. Lu and H.-M. Qin, Programmed protein scaffold for multienzyme assembly empowering the biosynthesis of rare sugars, Chin. J. Catal., 2025, 72, 95–105,  DOI:10.1016/S1872-2067(25)64675-4.
12. G. A. Ellis, W. P. Klein, G. Lasarte-Aragonés, M. Thakur, S. A. Walper and I. L. Medintz, Artificial Multienzyme Scaffolds: Pursuing in Vitro Substrate Channeling with an Overview of Current Progress, ACS Catal., 2019, 9(12), 10812–10869,  DOI:10.1021/acscatal.9b02413.
13. Z. Wu, J. Shi, B. Zhang, Y. Jiao, X. Meng, Z. Chu, Y. Chen, Y. Cheng and Z. Jiang, Conjugated microporous polymers-scaffolded enzyme cascade systems with enhanced catalytic activity, Chin. J. Catal., 2024, 63, 213–223,  DOI:10.1016/S1872-2067(24)60088-4.
14. L. Zeng, T. Guo, J. Liu, L. Chu, X. Zhang, D. Fan and Y. Bai, Self-Assembly of P22 Virus-like Particles in Liquid–Liquid Phase-Separated Condensates for Enhanced Enzyme Catalysis, ACS Nano, 2025, 19(44), 38472–38488,  DOI:10.1021/acsnano.5c11558.
15. C. Ben, S. Zhao, Q. Wu, H. He, M. Yu, D. Liu, D. He, X.-M. Song and Z. Song, Hydrophobic Ionic Liquid Gel Microspheres as Bi-Component Carriers with a Liquid Phase to Immobilize Enzymes for Enhanced Performance, Adv. Funct. Mater., 2024, 34(46), 2407913,  DOI:10.1002/adfm.202407913.
16. L. Zhang, Y. Xu, T. M. Makris and Q. Wang, Enhanced Arylamine N-Oxygenase Activity of Polymer–Enzyme Assemblies by Facilitating Electron-Transferring Efficiency, Biomacromolecules, 2018, 19(3), 918–925,  DOI:10.1021/acs.biomac.7b01706.
17. Y. L. Nunes, F. L. de Menezes, I. G. de Sousa, A. L. G. Cavalcante, F. T. T. Cavalcante, K. da Silva Moreira, A. L. B. de Oliveira, G. F. Mota, J. E. da Silva Souza, I. R. de Aguiar Falcão, T. G. Rocha, R. B. R. Valério, P. B. A. Fechine, M. C. M. de Souza and J. C. S. dos Santos, Chemical and physical Chitosan modification for designing enzymatic industrial biocatalysts: How to choose the best strategy?, Int. J. Biol. Macromol., 2021, 181, 1124–1170,  DOI:10.1016/j.ijbiomac.2021.04.004.
18. O. I. Wilner, Y. Weizmann, R. Gill, O. Lioubashevski, R. Freeman and I. Willner, Enzyme cascades activated on topologically programmed DNA scaffolds, Nat. Nanotechnol., 2009, 4(4), 249–254,  DOI:10.1038/nnano.2009.50.
19. Y. Ren, L. Zhang, T. Sun, Y. Yin and Q. Wang, Enzyme Immobilization on a Delignified Bamboo Scaffold as a Green Hierarchical Bioreactor, ACS Sustainable Chem. Eng., 2022, 10(19), 6244–6254,  DOI:10.1021/acssuschemeng.2c00346.
20. Y. Zheng, L. Wang, H. Liu, J. Yang, R. Zhang, L. Zhang and Z.-A. Qiao, A Modular Co-assembly Strategy for Ordered Mesoporous Perovskite Oxides with Abundant Surface Active Sites, Angew. Chem., Int. Ed., 2022, 61(37), e202209038,  DOI:10.1002/anie.202209038.
21. P. Milkin, A. Antony, H. Cai, A. Debevc, C. Topal, A. Linhardt, A. Synytska and L. Ionov, Highly Sensitive Electrochemical Biosensor Based on Hairy Particles with Controllable High Enzyme Loading and Activity, Adv. Funct. Mater., 2025, 2507589,  DOI:10.1002/adfm.202507589.
22. S. Tang, L. Xu, K. Dong, Q. Wang, J. Zeng, X. Huang, H. Li, L. Xia and L. Wang, Curvature effect on graphene-based Co/Ni single-atom catalysts, Appl. Surf. Sci., 2023, 615, 156357,  DOI:10.1016/j.apsusc.2023.156357.
23. J. Liang, S. Gao, J. Liu, M. Y. B. Zulkifli, J. Xu, J. Scott, V. Chen, J. Shi, A. Rawal and K. Liang, Hierarchically Porous Biocatalytic MOF Microreactor as a Versatile Platform towards Enhanced Multienzyme and Cofactor-Dependent Biocatalysis, Angew. Chem., Int. Ed., 2021, 60(10), 5421–5428,  DOI:10.1002/anie.202014002.
24. X.-G. Yang, J.-R. Zhang, X.-K. Tian, J.-H. Qin, X.-Y. Zhang and L.-F. Ma, Enhanced Activity of Enzyme Immobilized on Hydrophobic ZIF-8 Modified by Ni2+ Ions, Angew. Chem., Int. Ed., 2023, 62(7), e202216699,  DOI:10.1002/anie.202216699.
25. K. Liang, R. Ricco, C. M. Doherty, M. J. Styles, S. Bell, N. Kirby, S. Mudie, D. Haylock, A. J. Hill, C. J. Doonan and P. Falcaro, Biomimetic mineralization of metal-organic frameworks as protective coatings for biomacromolecules, Nat. Commun., 2015, 6(1), 7240,  DOI:10.1038/ncomms8240.
26. H. An, J. Song, T. Wang, N. Xiao, Z. Zhang, P. Cheng, S. Ma, H. Huang and Y. Chen, Metal–Organic Framework Disintegrants: Enzyme Preparation Platforms with Boosted Activity, Angew. Chem., Int. Ed., 2020, 59(38), 16764–16769,  DOI:10.1002/anie.202007827.
27. Y. Du, X. Jia, L. Zhong, Y. Jiao, Z. Zhang, Z. Wang, Y. Feng, M. Bilal, J. Cui and S. Jia, Metal-organic frameworks with different dimensionalities: An ideal host platform for enzyme@MOF composites, Coord. Chem. Rev., 2022, 454, 214327,  DOI:10.1016/j.ccr.2021.214327.
28. Y. Shu, W. Liang and J. Huang, Heterogeneity in enzyme/metal–organic framework composites for CO2 transformation reactions, Green Chem., 2023, 25(11), 4196–4221,  10.1039/D2GC04682B.
29. X. Wu, J. Ge, C. Yang, M. Hou and Z. Liu, Facile synthesis of multiple enzyme-containing metal-organic frameworks in a biomolecule-friendly environment, Chem. Commun., 2015, 51(69), 13408–13411,  10.1039/c5cc05136c.
30. H. Babaei, Z. G. Nejad, S. Yaghmaei and F. Farhadi, Co-immobilization of multi-enzyme cascade system into the metal–organic frameworks for the removal of Bisphenol A, Chem. Eng. J., 2023, 461, 142050,  DOI:10.1016/j.cej.2023.142050.
31. T. Man, C. Xu, X.-Y. Liu, D. Li, C.-K. Tsung, H. Pei, Y. Wan and L. Li, Hierarchically encapsulating enzymes with multi-shelled metal-organic frameworks for tandem biocatalytic reactions, Nat. Commun., 2022, 13(1), 305,  DOI:10.1038/s41467-022-27983-9.
32. C. Chen, M. R. Alalouni, X. Dong, Z. Cao, Q. Cheng, L. Zheng, L. Meng, C. Guan, L. Liu, E. Abou-Hamad, J. Wang, Z. Shi, K.-W. Huang, L. Cavallo and Y. Han, Highly Active Heterogeneous Catalyst for Ethylene Dimerization Prepared by Selectively Doping Ni on the Surface of a Zeolitic Imidazolate Framework, J. Am. Chem. Soc., 2021, 143(18), 7144–7153,  DOI:10.1021/jacs.1c02272.
33. L. L. Zulfa, A. R. P. Hidayat, W. P. Utomo, R. Subagyo, E. N. Kusumawati, Y. Kusumawati, D. Hartanto, W. Widyastuti and R. Ediati, Facile synthesis of Ni-ZIF-8 with improved photodegradation performance for methylene blue, Case Stud. Chem. Environ. Eng., 2024, 10, 100828,  DOI:10.1016/j.cscee.2024.100828.
34. Y. Pan, H. Li, J. Farmakes, F. Xiao, B. Chen, S. Ma and Z. Yang, How Do Enzymes Orient When Trapped on Metal–Organic Framework (MOF) Surfaces?, J. Am. Chem. Soc., 2018, 140(47), 16032–16036,  DOI:10.1021/jacs.8b09257.
35. Q. Li, Z. Armstrong, A. MacRae, M. Lenertz, L. Feng and Z. Yang, On the interface of enzyme and spatial confinement: The impacts of confinement rigidity, shape, and surface properties on the interplay of enzyme structure, dynamics, and function, Chem. Phys. Rev., 2023, 4(4), 041302,  DOI:10.1063/5.0167117.
36. C. Li and L. X. Wang, Chemoenzymatic Methods for the Synthesis of Glycoproteins, Chem. Rev., 2018, 118(17), 8359–8413,  DOI:10.1021/acs.chemrev.8b00238.
37. B. M. Zimmer, J. J. Barycki and M. A. Simpson, Mechanisms of coordinating hyaluronan and glycosaminoglycan production by nucleotide sugars, Am. J. Physiol.: Cell Physiol., 2022, 322(6), C1201–C1213,  DOI:10.1152/ajpcell.00130.2022.
38. D. Kang, J. Lee, G. Yook, S. Jeong, J. Shin, M.-S. Kim, Y.-J. Kim, H. Jung, J. Ahn, T. W. Kim, M. J. Chang, C. B. Chang, S.-B. Kang, W. H. Yang, Y.-h. Lee, J. W. Cho, E. C. Yi, C. Kang and J.-H. Kim, Regulation of senescence-associated secretory phenotypes in osteoarthritis by cytosolic UDP-GlcNAc retention and O-GlcNAcylation, Nat. Commun., 2025, 16(1), 1094,  DOI:10.1038/s41467-024-55085-1.
39. L. Zhang, H. Yu, Y. Bai, B. Mishra, X. Yang, J. Wang, E. B. Yu, R. Li and X. Chen, A Neoglycoprotein-Immobilized Fluorescent Magnetic Bead Suspension Multiplex Array for Galectin-Binding Studies, Molecules, 2021, 26(20), 6194,  DOI:10.3390/molecules26206194.
40. L. Zhang, Y. Li, R. Li, X. Yang, Z. Zheng, J. Fu, H. Yu and X. Chen, Glycoprotein In Vitro N-Glycan Processing Using Enzymes Expressed in E. coli, Molecules, 2023, 28(6), 2753,  DOI:10.3390/molecules28062753.
41. Y. Chen, V. Thon, Y. Li, H. Yu, L. Ding, K. Lau, J. Qu, L. Hie and X. Chen, One-pot three-enzyme synthesis of UDP-GlcNAc derivatives, Chem. Commun., 2011, 47(38), 10815–10817,  10.1039/C1CC14034E.
42. S. Li, S. Wang, Y. Wang, J. Qu, X.-w. Liu, P. G. Wang and J. Fang, Gram-scale production of sugar nucleotides and their derivatives, Green Chem., 2021, 23(7), 2628–2633,  10.1039/D1GC00711D.
43. T. L. Roberts, J. P. Dolan, G. J. Miller, M. A. D. Lima and S. C. Cosgrove, A modular, reusable biocatalytic flow system for UDP-GlcNAc production, React. Chem. Eng., 2025, 10(6), 1221–1226,  10.1039/D5RE00127G.
44. J. Li, Y. Liu, L. Hu, R. Xu, W. Zhang, S. Hu, Y. Wang, G. Du and Z. Kang, Construction of immobilized enzyme cascades for the biosynthesis of nucleotide sugars UDP-N-acetylglucosamine and UDP-glucuronic acid, Syst. Microbiol. Biomanuf., 2024, 4((3), 895–905,  DOI:10.1007/s43393-024-00238-y.
45. M. Nishimoto and M. Kitaoka, Identification of N-Acetylhexosamine 1-Kinase in the Complete Lacto-N-Biose I/Galacto-N-Biose Metabolic Pathway in Bifidobacterium longum, Appl. Environ. Microbiol., 2007, 73(20), 6444–6449,  DOI:10.1128/AEM.01425-07.
46. W. Abdallah, V. Chirino, I. Wheeldon and S. Banta, Catalysis of Thermostable Alcohol Dehydrogenase Improved by Engineering the Microenvironment through Fusion with Supercharged Proteins, ChemBioChem, 2019, 20(14), 1827–1837,  DOI:10.1002/cbic.201900066.
47. X. Niu, S. Yan, L. Wang, J. Chen, R. Zhao, H. Li, J. Liu and K. Wang, Induction of chiral polymers from metal-organic framework for stereoselective recognition, Anal. Chim. Acta, 2022, 1196, 339546,  DOI:10.1016/j.aca.2022.339546.
48. P.-H. Chang, C.-Y. Chen, R. Mukhopadhyay, W. Chen, Y.-M. Tzou and B. Sarkar, Novel MOF-808 metal–organic framework as highly efficient adsorbent of perfluorooctane sulfonate in water, J. Colloid Interface Sci., 2022, 623, 627–636,  DOI:10.1016/j.jcis.2022.05.050.
49. Y. Ma, Y. Xiang, X. Li, D. Zhang and Q. Chen, Recombinant streptococcal protein G-modified metal-organic framework ZIF-8 for the highly selective purification of immunoglobulin G from human serum, Anal. Chim. Acta, 2024, 1288, 342175,  DOI:10.1016/j.aca.2023.342175.
50. C. Li, W. Yang, X. Zhang, Y. Han, W. Tang, T. Yue and Z. Li, A 3D hierarchical dual-metal–organic framework heterostructure up-regulating the pre-concentration effect for ultrasensitive fluorescence detection of tetracycline antibiotics, J. Mater. Chem. C, 2020, 8(6), 2054–2064,  10.1039/C9TC05941E.
51. D. M. Aziz and D. K. Hama, Advanced Ni(II)-Doped -ZIF-8 Nanoscale crystals: A breakthrough in efficient dihydropyrimidine synthesis and catalyst reusability with comprehensive structural characterization, Microporous Mesoporous Mater., 2025, 387, 113533,  DOI:10.1016/j.micromeso.2025.113533.
52. Y. Weng, X. Xu, P. Yan, J. You, X. Chen, H. Song and C.-X. Zhao, Enzyme encapsulation in metal-organic frameworks using spray drying for enhanced stability and controlled release: A case study of phytase, Food Chem., 2024, 452, 139533,  DOI:10.1016/j.foodchem.2024.139533.
53. S. Puangpathumanond, H. L. Chee, C. Sevencan, X. Yang, O. S. Lau and T. T. S. Lew, Stomata-targeted nanocarriers enhance plant defense against pathogen colonization, Nat. Commun., 2025, 16(1), 4816,  DOI:10.1038/s41467-025-60112-w.
54. Y. Zhai, M. Liang, J. Fang, X. Wang, W. Guan, X.-w. Liu, P. Wang and F. Wang, NahK/GlmU fusion enzyme: characterization and one-step enzymatic synthesis of UDP-N-acetylglucosamine, Biotechnol. Lett., 2012, 34(7), 1321–1326,  DOI:10.1007/s10529-012-0910-y.
55. J. Zheng, H. Xu, B. Li, M. Sohail, J. Bi, F. Zhang, R. J. Linhardt, H. Huang and X. Zhang, Spatially Segregated MOF Bioreactor Enables Versatile Modular Glycoenzyme Assembly for Hierarchical Glycan Library Construction, ACS Appl. Mater. Interfaces, 2023, 15(16), 19807–19816,  DOI:10.1021/acsami.2c22094.
56. E. K. Goharshadi, A review on the radial distribution function: Insights into molecular structure, intermolecular interactions, and thermodynamic properties, J. Mol. Liq., 2025, 433, 127900,  DOI:10.1016/j.molliq.2025.127900.
57. Y. Li, H. Yu, Y. Chen, K. Lau, L. Cai, H. Cao, V. K. Tiwari, J. Qu, V. Thon, P. G. Wang and X. Chen, Substrate promiscuity of N-acetylhexosamine 1-kinases, Molecules, 2011, 16(8), 6396–6407,  DOI:10.3390/molecules16086396.
58. G. Veggiani, T. Nakamura, M. D. Brenner, R. V. Gayet, J. Yan, C. V. Robinson and M. Howarth, Programmable polyproteams built using twin peptide superglues, Proc. Natl. Acad. Sci. U. S. A., 2016, 113(5), 1202–1207,  DOI:10.1073/pnas.1519214113.
59. K. D. Brune, C. M. Buldun, Y. Li, I. J. Taylor, F. Brod, S. Biswas and M. Howarth, Dual Plug-and-Display Synthetic Assembly Using Orthogonal Reactive Proteins for Twin Antigen Immunization, Bioconjugate Chem., 2017, 28(5), 1544–1551,  DOI:10.1021/acs.bioconjchem.7b00174.
60. Y. B. Yu, Y. Liu, S. Li, X. Y. Yang and Z. Guo, The pH-responsive zeolitic imidazolate framework nanoparticle as a promising immune-enhancing adjuvant for anti-caries vaccine, J. Dent., 2023, 130, 104413,  DOI:10.1016/j.jdent.2023.104413.
61. T. D. Kühne, M. Iannuzzi, M. Del Ben, V. V. Rybkin, P. Seewald, F. Stein, T. Laino, R. Z. Khaliullin, O. Schütt, F. Schiffmann, D. Golze, J. Wilhelm, S. Chulkov, M. H. Bani-Hashemian, V. Weber, U. Borštnik, M. Taillefumier, A. S. Jakobovits, A. Lazzaro, H. Pabst, T. Müller, R. Schade, M. Guidon, S. Andermatt, N. Holmberg, G. K. Schenter, A. Hehn, A. Bussy, F. Belleflamme, G. Tabacchi, A. Glöß, M. Lass, I. Bethune, C. J. Mundy, C. Plessl, M. Watkins, J. VandeVondele, M. Krack and J. Hutter, CP2K: An electronic structure and molecular dynamics software package - Quickstep: Efficient and accurate electronic structure calculations, J. Chem. Phys., 2020, 152(19), 194103,  DOI:10.1063/5.0007045.
62. J. Jumper, R. Evans, A. Pritzel, T. Green, M. Figurnov, O. Ronneberger, K. Tunyasuvunakool, R. Bates, A. Žídek, A. Potapenko, A. Bridgland, C. Meyer, S. A. A. Kohl, A. J. Ballard, A. Cowie, B. Romera-Paredes, S. Nikolov, R. Jain, J. Adler, T. Back, S. Petersen, D. Reiman, E. Clancy, M. Zielinski, M. Steinegger, M. Pacholska, T. Berghammer, S. Bodenstein, D. Silver, O. Vinyals, A. W. Senior, K. Kavukcuoglu, P. Kohli and D. Hassabis, Highly accurate protein structure prediction with AlphaFold, Nature, 2021, 596(7873), 583–589,  DOI:10.1038/s41586-021-03819-2.
63. C. Bannwarth, E. Caldeweyher, S. Ehlert, A. Hansen, P. Pracht, J. Seibert, S. Spicher and S. Grimme, Extended tight-binding quantum chemistry methods, Wiley Interdiscip. Rev.:Comput. Mol. Sci., 2021, 11(2), e1493,  DOI:10.1002/wcms.1493.
64. C. Plett and S. Grimme, Automated and Efficient Generation of General Molecular Aggregate Structures, Angew. Chem., Int. Ed., 2023, 62(4), e202214477,  DOI:10.1002/anie.202214477.
65. C. Bannwarth, S. Ehlert and S. Grimme, GFN2-xTB—An Accurate and Broadly Parametrized Self-Consistent Tight-Binding Quantum Chemical Method with Multipole Electrostatics and Density-Dependent Dispersion Contributions, J. Chem. Theory Comput., 2019, 15(3), 1652–1671,  DOI:10.1021/acs.jctc.8b01176.
66. D. Van Der Spoel, E. Lindahl, B. Hess, G. Groenhof, A. E. Mark and H. J. C. Berendsen, GROMACS: Fast, flexible, and free, J. Comput. Chem., 2005, 26(16), 1701–1718,  DOI:10.1002/jcc.20291.
67. X. He, V. H. Man, W. Yang, T. S. Lee and J. Wang, A fast and high-quality charge model for the next generation general AMBER force field, J. Chem. Phys., 2020, 153(11), 114502,  DOI:10.1063/5.0019056.
68. B. H. M. Mooers, Shortcuts for faster image creation in PyMOL, Protein Sci., 2020, 29(1), 268–276,  DOI:10.1002/pro.3781.
69. R. A. Laskowski and M. B. Swindells, LigPlot+: multiple ligand-protein interaction diagrams for drug discovery, J. Chem. Inf. Model., 2011, 51(10), 2778–2786,  DOI:10.1021/ci200227u.
70. J. Zhang, J. Shin, N. Tague, H. Lin, M. Zhang, X. Ge, W. Wong, M. J. Dunlop and J. X. Cheng, Visualization of a Limonene Synthesis Metabolon Inside Living Bacteria by Hyperspectral SRS Microscopy, Adv. Sci., 2022, 9(32), e2203887,  DOI:10.1002/advs.202203887.

---