Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 May 24;22(39):e73922. doi: 10.1002/smll.73922

Redesign of the Ferritin Ferroxidase Center for Universal Molecular Binding or Specific Recognition

Wenming Wang 1, Hui Yao 1, Wenjun Gong 1, Danyang Ma 1, Jiaqi Qiao 1, Ying Zhang 1, Lintao Wu 2, Chengpeng Fan 3, Yaqin Zhao 1, Zhijun Wang 2, Zihan Jia 4, Yu Guo 4, Nan Zhang 4,✉, Guanghua Zhao 5,✉, Yang Yun 6,✉, Hongfei Wang 1,✉
PMCID: PMC13360382  PMID: 42178779

ABSTRACT

Although ferritin, as a versatile nanocarrier, has been engineered to improve cargo loading efficiency for various functions, including therapeutic applications, a universal design strategy enabling tunable molecular binding remains an unmet challenge. This study reports an AI‐aided structure‐guided engineering approach targeting the ferroxidase center of recombinant human heavy‐chain ferritin (rHuHF), aiming to achieve either universal molecular binding or high‐affinity specific recognition. Through site‐directed mutagenesis of key residues within and flanking the ferroxidase center, two rHuHF variants (rHuHF‐C1 and rHuHF‐C2) were generated. X‐ray crystallographic analysis revealed that the engineered pocket within rHuHF‐C2 can accommodate a broad range of hydrophobic molecules (e.g., Curcumin, CUR) via hydrophobic interactions, thus validating their universal molecular binding capability. On the other hand, leveraging AI‐assisted rational design, a variant (rHuHF‐71) was subsequently engineered to specifically bind CUR with enhanced affinity, facilitated by the formation of hydrogen bonds and optimized hydrophobic contacts. This work establishes a generalizable strategy, designated as “Excavation, Rebuilding, and Validation”, for engineering ferritin nanocages with tunable binding specificities, which holds great promise for advancing the development of protein‐based drug delivery systems and the design of small‐molecule binding proteins.

Keywords: ferritin, LigandMPNN, molecule binding, protein design, protein scaffold

This study redesigned the ferroxidase center of rHuHF into a universal pocket for broad‐spectrum hydrophobic molecule binding (validated by seven complex crystal structures). Further rational engineering enabled specific curcumin binding with enhanced affinity, and a universal design strategy was proposed ‐ offering a novel scaffold for ferritin‐based drug delivery and small‐molecule binding protein design.

graphic file with name SMLL-22-e73922-g002.jpg

1. Introduction

Proteins are one of the major classes of naturally occurring macromolecules in all living organisms, playing crucial roles in diverse biological processes [1]. Proteins are biological polymers composed of amino acids and exhibit diverse structures, ranging from simple linear chains to complex three‐dimensional conformations [2]. Protein‐based drug delivery has emerged as an innovative approach, leveraging proteins as carriers to enhance therapeutic efficacy. This strategy is gaining traction in the pharmaceutical field as a promising candidate for efficient drug and gene delivery, owing to its potential to improve bioavailability, target specificity, renewable resource abundance, cellular uptake efficiency, and reduce systemic side effects [3]. Additionally, proteins offer advantages of uniform size, atomically resolved structures, broad chemical functionality, and inherent biodegradability [4]. Numerous proteins have been applied, such as human serum albumin (HSA) [5, 6, 7], transferrin [8, 9], Rnase [10], lysozyme [11, 12, 13], virus‐capsid [14, 15], ferritin [16, 17], silk protein [18], and designed protein [4]. Among them, protein nanocages, particularly ferritins, have garnered significant attention as potential nanocarriers due to their unique structural and functional properties [16].

The de novo creation of selective small‐molecule binding proteins is a major current challenge in protein design and has seen exciting recent progress. Polizzi and DeGrado developed vdM to help design ligand‐binding proteins and designed several kinds of de novo proteins that bind the drug apixaban [19]. Recently, they developed a computational procedure to design a protein that recognizes a common pharmacophore in a series of poly(ADP‐ribose) polymerase–1 inhibitors [20]. Baker et al. also described an approach for designing high‐affinity small‐molecule binding proteins for downstream sensing [21]. Dang et al. presented a strategy to engineer de novo designed proteins for the effective clearance of drugs [22]. The light‐responsive phenylalanine‐4’‐azobenzene (AzoF)‐binding proteins were successfully designed by Cao et al. Light regulates AzoF state, modulating interactions between these two proteins and enabling photo‐regulated assembly and dissociation of their complex [23]. Notably, in these influential studies, the scaffolds utilized for small‐molecule binding protein design are predominantly de novo designed. It is therefore of considerable scientific interest to explore whether a natural protein could be engineered via rational design and modification to achieve either universal or specific binding to small molecules. Building on the aforementioned hypothesis, if ferritin served as a template for the design of small‐molecule binding proteins, with the capacity to realize either universal or specific molecule binding following rational engineering, it would hold profound implications for advancing the field of small‐molecule binding protein design.

Ferritins are ubiquitous iron storage proteins in cells and are key in iron metabolism, which are typically composed of 24 subunits self‐assembling into a hollow protein shell with an outer diameter of 12 nm and an interior cavity 8 nm in diameter. Human ferritin is composed of two structurally similar, but functionally different subunits, human light‐chain ferritin (HuLF) and human heavy‐chain ferritin (HuHF) [24, 25, 26]. Human transferrin receptor 1 (TfR1) guarantees iron supply by endocytosis upon binding of iron‐loaded transferrin and ferritin, and this receptor is often overexpressed on various cancer cell types, endowing recombinant HuHF (rHuHF) with tumor‐targeting ability [27]. Recently, rHuHF has now emerged as an attractive and promising vehicle for encapsulation and delivery of drugs and for tumor imaging. A variety of molecules, including imaging agents, therapeutic drugs, nucleic acids, metal nanoparticles, and enzymes, have been loaded into the interior cavity of ferritin nanocages for a broad range of applications [28, 29, 30, 31, 32].

Another advantage of ferritin nanocages is the ease of introducing additional functionalities through genetic or chemical manipulation [33]. Numerous ferritin variants have been rationally designed in recent years to enable novel self‐assembly forms and enhance drug encapsulation efficiency. For instance, the key subunit interface redesign (KSIR) strategy has yielded 16, 48, and 8‐mer ferritins [34, 35, 36, 37]. Subunit interface redesign can modulate the stability of ferritin nanocages, either increasing or decreasing it, thereby facilitating drug loading under specific conditions [38, 39]. Additionally, Jung et al. and Wang et al. prepared four‐fold channel‐nicked ferritin, which enabled active and efficient doxorubicin loading [40, 41]. Dmochowski et al. and Bonamore et al. replaced many inner surface accessible hydrophilic residues with hydrophobic amino acids, obtaining larger protein cavities with novel properties [42, 43]. Beck proposed a cysteine‐maleimide conjugation strategy for doxorubicin encapsulation, enabling specific loading into ferritin variants with tuned surface charges [44]. Lu et al. and Ueno et al. synthesized gold nanoclusters with a four‐fold channel or an inner surface redesigned ferritin [45, 46]. Furthermore, fusing antigens to the N‐terminus or conjugating them to the outer surface of ferritin has enabled the preparation of nanovaccines [47, 48, 49, 50]. Peptides or antibodies have also been fused to enhance tumor targeting, therapeutic efficacy, and immunotherapeutic outcomes [51, 52, 53]. In summary, the inner surface, outer surface, key subunit interfaces, and threefold/fourfold channels represent the primary targets for ferritin engineering, encompassing most of the nanocage structure except for the ferroxidase center. Of greater significance, the aforementioned engineering strategies for ferritin are often tailored to specific drug molecules, relying on empirical selection of modification sites and methodologies. This approach inherently lacks generality; therefore, it is crucial to identify universal or selective binding sites for small molecules in ferritin nanocage and develop general design methods.

The vast majority of ubiquitous H‐type ferritin subunits contain a ferroxidase center buried within the four‐helix bundle, which catalyzes the rapid oxidation of ferrous iron by O2 or H2O2 [54]. Specifically, residues E27 and Y34 from helix A, E62 and H65 from helix B, E107 from helix C, Y137 and Q141 from helix D constitute the ferroxidase center of rHuHF [55]. While numerous variants have been engineered to investigate the molecular mechanism of Fe(II) oxidation or synthesize quantum dots, no significant correlation has been established between the ferroxidase center and molecular binding/drug delivery by ferritin [56, 57].

Herein, we propose an “Excavation, Rebuilding, and Validation” strategy for the rational redesign of ferritin nanocages via site‐directed mutagenesis of the ferroxidase center and its adjacent regions, thereby affording a novel pocket for molecular binding. This three‐step protocol comprises: (i) Site‐directed substitution of residues within and flanking the ferroxidase center with alanine to expand the internal pocket, with subsequent structural validation of molecular binding capacity, (ii) employing an AI‐assisted protein design approach to rebuild and modify the pocket based on the properties of target molecules, and (iii) determination of the ferritin‐molecule complex structure to characterize intermolecular interactions at atomic resolution. Consequently, two newly engineered ferritin nanocages were prepared, which are able to bind to Curcumin (CUR) either non‐specifically or specifically, a finding validated by atomic‐resolution x‐ray crystal structure analysis. In all, we engineered a ferritin mutant harboring a hydrophobic pocket that enables promiscuous binding to diverse hydrophobic molecules. Through the integration of AI‐assisted rational design, this pocket can be further optimized for high‐affinity and specific recognition of target molecules.

2. Results and Discussion

2.1. Excavating of Ferroxidase Center in rHuHF

The subunit of eukaryotic H‐type ferritin contains a ferroxidase center where Fe(II) binds and is subsequently oxidized by O2 or H2O2 [55]. In rHuHF, residues E27, E62, H65, E107, and Q141 are pivotal for iron oxidation, while nearby Y34 and Y137 are highly conserved and facilitate electron transfer for molecular oxygen reduction to water during this process [58]. Based on the published rHuHF crystal structure (PDB: 2FHA) [59], the pocket volume was measured at about 241 Å3 (Figure 1a). Notably, these residues are hydrophilic with long side chains; thus, simultaneous alanine substitution of these residues was hypothesized to generate a larger hydrophobic pocket within the four‐helix bundle of the rHuHF subunit. To validate this hypothesis, the gene encoding variant rHuHF‐E27A‐Y34A‐E62A‐H65A‐E107A‐Y137A‐Q141A (rHuHF‐C1) was synthesized, and the protein was expressed and purified. Amino acid sequence alignment between rHuHF and rHuHF‐C1 is shown in Figure S1. rHuHF‐C1 crystallized under conditions identical to wild‐type rHuHF, suggesting that inner‐surface mutations do not affect global protein properties. X‐Ray diffraction data were collected, and the crystal structure of rHuHF‐C1 was resolved. Structural analysis revealed that the pocket dimensions in rHuHF‐C1 increased to 11 Å (upward) and 17 Å (downward), with a fourfold volume increase to 1037 Å3 (Figure 1b). Notably, mutagenesis of these key residues did not disrupt the overall structure, as evidenced by an all‐atom RMSD of 0.784 Å between rHuHF and rHuHF‐C1 (Figure 1c). Previous studies have shown that subunit interface mutations may affect ferritin assembly [34, 35, 36]. To assess tertiary structure integrity, rHuHF‐C1 morphology was visualized by transmission electron microscopy (TEM), confirming its spherical shape with an exterior diameter of ∼12 nm, nearly identical to wild‐type rHuHF (Figure 1d), which might be due to the fact that the mutated residues in rHuHF‐C1 are distal to these interfaces.

FIGURE 1.

FIGURE 1

Open in a new tab

Design of rHuHF‐C1. (a) Ferroxidase center pocket of wild‐type rHuHF. The pocket in rHuHF is visualized as a green surface, with key pocket‐forming residues shown as stick models. (b) Engineered cavity of rHuHF‐C1 (PDB: 9VO7). The rHuHF‐C1 cavity is depicted as a blue surface. (c) Structure alignment of rHuHF and rHuHF‐C1. (d) TEM of rHuHF‐C1 stained with 1% (w/w) uranyl acetate. Scale bar is 100 nm.

Given that the hydrophobic pocket in rHuHF‐C1 exhibits a narrow top and a wide bottom, hydrophobic molecules face steric hindrance during binding. Thus, expanding the pocket entrance was deemed essential. Hence, nine amino acids (Y54, H57, Q58, E61, K68, H136, E140, K143, and E147) surrounding the rHuHF‐C1 pocket were also mutated to alanine, generating variant rHuHF‐C2 (Figure 2a). Amino acid sequence alignment of rHuHF, rHuHF‐C1, and rHuHF‐C2 subunits is shown in Figure S1. Notably, this engineering introduced 16 alanine substitutions per rHuHF subunit (384 residues in the 24‐mer ferritin), raising concerns about protein expression and folding. Despite this, the rHuHF‐C2 gene was successfully synthesized, overexpressed, and purified. Surprisingly, these mutations did not compromise global protein properties; rHuHF‐C2 crystallized under conditions identical to wild‐type rHuHF (Figure 2b). Structural analysis revealed a significantly enlarged hydrophobic pocket within the four‐helix bundle, with dimensions of 23, 14, and 12 Å, and a volume of 1,244 Å3, a 200 Å3 increase over rHuHF‐C1 and 1,000 Å3 over wild‐type rHuHF. Strikingly, mutagenesis of these 16 key residues did not disrupt the overall fold, as evidenced by an all‐atom RMSD of 0.376 Å between rHuHF and rHuHF‐C2 (Figure 2c). TEM confirmed that rHuHF‐C2 self‐assembles into a 24‐mer nanocage with a 12 nm exterior diameter, consistent with crystallographic observations (Figure 2d).

FIGURE 2.

FIGURE 2

Open in a new tab

Design of rHuHF‐C2. (a) Overall structure of rHuHF‐C2 (PDB: 9VOC). Mutated residues are depicted as stick models. (b) Engineered pocket of rHuHF‐C2, visualized as a magenta surface. (c) Structure alignment of rHuHF and rHuHF‐C2. (d) TEM of rHuHF‐C2 stained with 1% (w/w) uranyl acetate. Scale bar is 100 nm.

2.2. The Loading Capacity of rHuHF‐C2

As described above, hydrophobic cavities were engineered in rHuHF‐C1 and rHuHF‐C2 subunits, with their formation significantly increasing the internal volume of the ferritin nanocage. Here, we calculated the inner volume for rHuHF, rHuHF‐C1, and rHuHF‐C2 by program MoloVol [60]. Notably, rHuHF‐C2 exhibited the largest volume at 353 947 Å3, a 12.8% increase (40 200 Å3) compared to wild‐type rHuHF following its ferroxidase center redesign. Structural analyses revealed that these modifications altered the inner cavity morphology, namely, numerous bulges were observed on the inner surfaces of rHuHF‐C1 and rHuHF‐C2, corresponding to the engineered binding cavities for hydrophobic molecules (Figure 3a).

FIGURE 3.

FIGURE 3

Open in a new tab

Encapsulation of CUR by ferritin nanocage. (a) The surface maps and volumes of the inner cavity for rHuHF, rHuHF‐C1, and rHuHF‐C2 are calculated by the program MoloVol [60]. (b) UV–vis absorption spectra of rHuHF and rHuHF‐C2. (c,d) UV–vis absorption spectra of CUR encapsulated in rHuHF and rHuHF‐C2 using pH‐based and temperature‐based loading methods.

CUR, a polyphenolic compound, targets multiple signaling molecules and exhibits potent antioxidant and anti‐inflammatory activities [61, 62, 63, 64]. As a representative of poorly soluble bioactive compounds, CUR is often encapsulated in ferritin nanocages to improve its solubility and bioavailability [35, 65, 66, 67]. Therefore, CUR was selected to evaluate the loading capacity of rHuHF‐C2 using two distinct strategies, with wild‐type rHuHF as a control. Ferritin nanocages disassemble into subunits under alkaline conditions (pH 12.0) and reassemble upon neutralization. Leveraging this property, CUR was successfully encapsulated into rHuHF and rHuHF‐C2 (Figure S2a). A second loading method exploited the thermo‐responsive disassembly‐reassembly behavior of ferritin, where subunits dissociate at 60°C and reform nanocages upon cooling (Figure S2b). To quantify CUR loading, we established a measurement system capitalizing on its solubility in alkaline solution (pH 12.0) and characteristic absorbance at 470 nm. A standard curve was generated for free CUR at 470 nm under alkaline conditions, while absorbance at 280 nm showed a linear relationship with CUR concentration (Figure S2c). Thus, the concentration of CUR encapsulated in ferritin can be directly calculated from absorbance values at 470 nm (for CUR) and 280 nm (for rHuHF and variants). The loading capacity is defined as the molar ratio of CUR to ferritin nanocage. Using the first method, we adjusted the rHuHF solution to pH 12.0 with sodium hydroxide, mixed it with CUR, and stirred for several minutes. The solution was then rapidly neutralized, enabling the ferritin nanocage to assemble around CUR. Parallel experiments were performed using rHuHF‐C2 for CUR encapsulation. Following quantitative analysis of CUR loading in rHuHF and rHuHF‐C2 using UV–vis spectroscopy, the loading efficiency (number of CUR molecules per protein nanocage) was determined: 86 of CUR molecules encapsulated in the cavity per rHuHF nanocage vs. 115 in the cavity and pockets per rHuHF‐C2. (Figure 3b,c) For the second method, rHuHF/rHuHF‐C2 was co‐incubated with CUR at 60°C for 20 min, cooled to 4°C for 2 h, dialyzed, and then subjected to UV–vis spectral analysis (Figure 3b,d). Results showed loading efficiencies of 20 and 40 for rHuHF and rHuHF‐C2, respectively. Notably, the first method yielded higher loading efficiency than the second. Engineering the ferroxidase center pocket significantly enhanced loading efficiency, as CUR molecules preferentially occupy the designed hydrophobic cavities in rHuHF‐C2. However, the binding model and detailed inter‐molecular interactions between rHuHF‐C2 cavities and CUR remain to be elucidated.

2.3. Crystal Structure of rHuHF‐C2 With CUR Binding in the Redesigned Pocket

For clarity, we designated rHuHF‐C2 encapsulating CUR via pH adjustment as rHuHF‐C2‐CUR‐1, and the complex prepared by the second method as rHuHF‐C2‐CUR‐2. Yellow crystals of the complex were obtained under conditions identical to wild‐type rHuHF (Figure 4a). Crystal structures revealed that both complexes reassembled into intact 24‐mer ferritin‐like nanocages with no subunit loss. Notably, a CUR molecule was observed bound within the redesigned pocket of each rHuHF‐C2 subunit (Figure 4b). CUR conformations in rHuHF‐C2‐CUR‐1 and rHuHF‐C2‐CUR‐2 were nearly identical (RMSD = 0.304 Å) (Figure 4c). 2Fo‐Fc electron density maps showed higher quality for rHuHF‐C2‐CUR‐2, fully encompassing CUR, consistent with its higher loading efficiency (Figure 4d). Thus, rHuHF‐C2‐CUR‐2 was selected for interaction analysis (Figure 4e). The redesigned pocket is highly hydrophobic, featuring Pi‐Sigma and Pi‐Alkyl hydrophobic interactions with CUR, involving residues A27, A54, A58, A65, V110, A144, A147, and L148. A hydrogen bond between Q23 and CUR was also identified. Although Pi‐Sigma and Pi‐Alkyl hydrophobic interactions are weak, they dominate the binding. The loading capacity results showed that rHuHF‐C2 exhibited a stronger encapsulation efficiency for CUR than rHuHF, with approximately 20 more CUR molecules per nanocage. This result is highly consistent with the findings from crystal structure studies: rHuHF‐C2 has 24 additional engineered pockets compared to rHuHF, and each pocket can bind one CUR molecule, which exactly accounts for the ability to bind more than 20 additional CUR molecules with respect to rHuHF.

FIGURE 4.

FIGURE 4

Open in a new tab

The interaction between CUR and rHuHF‐C2. (a) Crystal morphology of the rHuHF‐C2‐CUR complex. (b) Overall structures of rHuHF‐C2‐CUR‐1 (PDB: 9VOD, green) and rHuHF‐C2‐CUR‐2 (PDB: 9VOE, magenta). CUR molecules in the cavities are depicted as stick models. (c) Structural alignment of CUR molecules in rHuHF‐C2‐CUR‐1 and rHuHF‐C2‐CUR‐2 cavities. (d) 2Fo‐Fc electron density maps of CUR (blue, contoured at 1.0 σ). (e) Interaction between CUR and the rHuHF‐C2 hydrophobic pocket. Pink dashed lines denote Pi‐Sigma and Pi‐Alkyl hydrophobic interactions, yellow dashed lines indicate hydrogen bonds, and relevant residues are labeled with stick models.

2.4. The Pocket in rHuHF‐C2 Enables Universal Hydrophobic Molecule Binding

The interaction between CUR and the rHuHF‐C2 pocket is non‐specific, as evidenced by minimal hydrogen bonding in the crystal structure. This is not surprising because the pocket design does not incorporate target molecule structural features; instead, we performed systematic alanine substitution of residues within and flanking the ferroxidase center. Thus, we infer that the rHuHF‐C2 pocket may be capable of binding diverse hydrophobic small molecules, i.e., it exhibits broad‐spectrum binding capacity for this class of compounds. To validate this hypothesis, we selected six hydrophobic molecules from our laboratory's existing compound library, including Alizarin Yellow GG (ALY), 3‐Aminoazobenzene (AZB), Nitrosyl Iron Sulfur complex (NIS), Dacarbazine (DAC), Semustine (SEM), and Carmustine (CAR), for subsequent experiments. All molecules were encapsulated into rHuHF‐C2 via a heat‐cool protocol. First, we tested ALY, a water‐insoluble monoazo dye (Figure S3a). The rHuHF‐C2‐ALY complex was successfully prepared and crystallized, revealing the binding of two ALY molecules per pocket (Figure 5a). High‐quality 2Fo—Fc electron density maps confirmed stable binding (Figure S3a), mediated by Pi‐Sigma and Pai‐Alkyl hydrophobic interactions with residues A27, A30, A34, A54, A58, A61, A62, A65, M100, A103, V110, L148, A140, and A144, as well as a hydrogen bond between Q23 and ALY2 (Figure 5a).

FIGURE 5.

FIGURE 5

Open in a new tab

Complex structures of rHuHF‐C2 with ALY, AZB, NIS, SEM, DAC, and CAR. (a) Overall structure of the rHuHF‐C2‐ALY complex (PDB: 9VOF) and interaction analysis between ALY and the rHuHF‐C2 hydrophobic pocket. rHuHF‐C2 is displayed as a salmon cartoon, with ALY molecules in the pocket shown as stick models. Key residues participating in interactions with ALY are represented as green sticks. Pink dashed lines denote Pi‐Sigma and Pai‐Alkyl interactions, green dashed lines indicate hydrogen bonds, and relevant residues are labeled. (b) Overall structure of the rHuHF‐C2‐AZB complex (PDB: 9VOM) and corresponding interaction analysis. (c) Overall structure of the rHuHF‐C2‐NIS complex (PDB: 21HO) and corresponding interaction analysis. (d) Overall structure of the rHuHF‐C2‐SEM complex (PDB: 21KW) and corresponding interaction analysis. (e) Overall structure of the rHuHF‐C2‐DAC complex (PDB: 21KV) and corresponding interaction analysis. (f) Overall structure of the rHuHF‐C2‐CAR complex (PDB: 21LE) and corresponding interaction analysis. (g) Structural alignments of rHuHF‐C2‐CUR, rHuHF‐C2‐ALY, rHuHF‐C2‐AZB, rHuHF‐C2‐NIS, rHuHF‐C2‐SEM, rHuHF‐C2‐DAC, and rHuHF‐C2‐CAR. Small molecules are displayed as stick models in different colors, while rHuHF‐C2 is shown in salmon surface. (h) Frequency distribution chart of amino acid occurrence in interactions.

Next, we investigated AZB, which is also insoluble in cold water (Figure S3b). The rHuHF‐C2‐AZB complex exhibited a crystal structure analogous to that of rHuHF‐C2‐ALY, with two AZB molecules bound per pocket (Figure 5b). The 2Fo‐Fc electron density maps of AZB molecules (Figure S3b) also exhibit high quality. Hydrophobic interactions, primarily Pi‐Sigma and Pi‐Alkyl interactions, dominate the binding between rHuHF‐C2 and AZB, with no hydrogen bonds detected in this complex (Figure 5b). NIS is widely used as a carrier for nitric oxide (NO) radicals and chelatable iron pools, playing a critical role in chemical and biomedical applications (Figure S3c) [68]. In the resolved rHuHF‐C2‐NIS complex structure, only one NIS molecule was found bound within the pocket. The primary forces mediating the rHuHF‐C2‐NIS interaction remain Pi‐Sigma and Pai‐Alkyl interactions; however, unlike the previous two molecules, a small number of hydrogen bonds are present (Figure 5c). Moreover, DAC, SEM, and CAR are three distinct antitumor drugs with extensive clinical applications (Figure S3d–f) [69, 70]. Among the three resolved complex structures (rHuHF‐C2‐DAC, rHuHF‐C2‐SEM, and rHuHF‐C2‐CAR), the rHuHF‐C2 pocket binds one molecule of DAC or CAR per pocket, respectively, whereas two SEM molecules are accommodated per pocket (Figure 5d–f). Similar to AZB, ALY, and NIS, hydrophobic interactions between these antitumor drugs and rHuHF‐C2 remain the fundamental driver of their stable binding (Figure 5d–f). High‐quality 2Fo‐Fc electron density maps for DAC, SEM, and CAR further validate the reliability of these results (Figure S3d–f).

Upon structural alignment of the seven resolved rHuHF‐C2‐small molecule complex structures, we observed that these small molecules occupy distinct positions within the rHuHF‐C2 pocket. Furthermore, nearly every region of the pocket is occupied by one or more of the small molecules, and this phenomenon provides additional evidence that the rHuHF‐C2 pocket exhibits inherent versatility for small‐molecule binding, enabling small molecules to bind to the pocket interior based on their intrinsic physicochemical properties (Figure 5g). Furthermore, we analyzed the occurrence frequency of key amino acids within the rHuHF‐C2 pocket that are involved in interactions with the seven small molecules (Figure 5h). Results indicated that L106, V110, and L148 exhibited the highest occurrence frequencies; these amino acids are intrinsic to wild‐type rHuHF and localized at the central region of the pocket. Among the 11 amino acids with an occurrence frequency of 4 or more, 8 were engineered sites. Of the 23 total amino acids involved in these interactions, 14 were engineered sites, accounting for 87.5% of all 16 engineered mutation sites. These results demonstrate that, while preserving the fundamental structural properties of rHuHF, the targeted mutations successfully constructed a pocket with broad‐spectrum small molecule binding capacity at the former ferroxidase center of rHuHF.

Collectively, we successfully resolved the complex structures of rHuHF‐C2 with seven small molecules. This achievement demonstrates that rHuHF‐C2 exhibits broad‐spectrum binding activity toward hydrophobic molecules. Coupled with ferritin's inherent advantages, such as robust structural stability and facile crystallization, rHuHF‐C2 emerges as an excellent and versatile design template for small‐molecule binding proteins, which is rationally engineered from a wild‐type protein scaffold.

2.5. The Pocket in rHuHF‐C2 Can be Re‐Engineered to Enable High‐Affinity Specific Binding of Target Molecules

As demonstrated above, seven hydrophobic molecules (CUR, ALY, AZB, NIS, SEM, DAC, and CAR) bind within the rHuHF‐C2 pocket primarily via weak and non‐specific hydrophobic interactions. To enhance binding affinity and confer specificity, the rHuHF‐C2 pocket must be re‐engineered for target molecules. In the rHuHF‐C2‐CUR complex structure, CUR is surrounded by alanine residues with relatively long distances, providing ample space for substituting these residues with longer side‐chain amino acids. Here, we leverage CUR as a model to further design the ferritin pocket, with the aim of enabling specific CUR binding in the newly engineered pocket from a wild‐type protein scaffold.

To design and screen ferritins with enhanced CUR binding affinity, we performed sequence optimization based on the rHuHF‐C2‐CUR‐1 complex structure. Fourteen alanine residues (positions 27, 30, 54, 58, 62, 65, 103, 107, 136, 137, 140, 141, 144, and 147) surrounding CUR were selected for mutagenesis. Using LigandMPNN [71], 200 sequences were generated, and 95 unique sequences remained after deduplication. The top 10 sequences (sequences 1, 10, 19, 23, 29, 36, 69, 77, 83, and 91) with the highest ligand_confidence scores were shortlisted (Figure 6a). Structures of these 10 variants were predicted by AlphaFold3 [72], yielding pTM scores of 0.95 for all, indicating high structural reliability (Figure 6b). When superposed with rHuHF‐C2, all‐atom RMSD values were < 0.3 Å (Figure 6c). To evaluate specific binding potential, CUR was docked into each mutant's pocket using the program Cdocker. For each variant, the CUR conformation with the highest ‐Cdocker_interaction_energy was selected (Figure 6d). RMSD calculations were performed by aligning these docked CUR structures with CUR in the rHuHF‐C2‐CUR‐1 complex (Figure 6e,f and Figure S4). Notably, CUR in variants 1, 19, and 71 exhibited conformations similar to the native complex. Variant rHuHF‐71 was chosen for experimental validation, featuring the highest ligand_confidence score (0.4242) and lowest RMSD value (1.4545 Å).

FIGURE 6.

FIGURE 6

Open in a new tab

Re‐engineering of the rHuHF‐C2 pocket for specific CUR binding. (a) Ligand_confidence scores of sequences designed by LigandMPNN. The top 10 sequences are highlighted in pink. (b) pTM scores of these sequences were assessed by AlphaFold3. (c) Structural alignment of newly designed rHuHF variants with rHuHF‐C2. (d) ‐Cdocker_interaction_energy values for CUR docking into these structures. Conformational alignment (e) and RMSD value (f) of CUR in rHuHF‐C2 and newly designed rHuHF variants. The CUR in rHuHF‐C2‐CUR‐1 is colored in salmon. (g) UV–vis absorption spectra of rHuHF‐71 and CUR encapsulated in rHuHF‐71 using pH‐based loading methods. (h) The pocket in rHuHF‐71 is depicted as a blue surface. (i) Overall structure of rHuHF‐71‐CUR complex (PDB: 9VON). The surface electrostatic potential of rHuHF‐71 is visualized, with red denoting negatively charged regions and blue indicating positively charged regions. (j) 2Fo‐Fc electron density maps of CUR in rHuHF‐71‐CUR complex (blue, contoured at 1.0 σ). (k) Structure alignment of rHuHF‐C2‐CUR and rHuHF‐71‐CUR. (l) Interaction between CUR and the rHuHF‐71 pocket. Pink dashed lines denote Pi‐Sigma and Pi‐Alkyl hydrophobic interactions, green dashed lines indicate hydrogen bonds. (m) Amino acid conservation analysis of the top ten sequences utilized for experimental characterization using WebLogo [73].

In subsequent experiments, we expressed, purified rHuHF‐71, and loaded CUR into its pocket via pH adjustment, as performed for rHuHF‐C2. As expected, rHuHF‐71 exhibited CUR loading efficiency comparable to rHuHF‐C2, reaching 103 CUR molecules per ferritin nanocage (Figure 6g). Subsequently, we obtained high‐quality rHuHF‐71‐CUR crystals and resolved their complex structure. The structure revealed that rHuHF‐71 possesses a significantly smaller four‐helix bundle pocket (volume: 799 Å3), representing a 30% shrinkage relative to rHuHF‐C2 (Figure 6h). Concomitantly, the pocket entrance size decreased markedly due to the emergence of R140 (Figure 6i). A CUR molecule was observed embedded in the new pocket, with its 2Fo‐Fc electron density map shown in Figure 6j. Despite the rHuHF‐71‐CUR diffraction data having a lower resolution (2.5 Å) than rHuHF‐C2‐CUR complexes, the map quality was high enough to fully resolve CUR, indicating enhanced CUR affinity and occupancy. When aligning rHuHF‐71‐CUR with rHuHF‐C2‐CUR‐2, CUR conformations showed remarkable similarity (RMSD = 1.2951 Å) (Figure 6k).

In rHuHF‐71‐CUR, CUR‐protein interactions shifted from predominantly hydrophobic to include numerous hydrogen bonds. Specifically, A27, A58, A65, V110, and L148 maintained Pi‐Sigma and Pi‐Alkyl hydrophobic interactions (as in rHuHF‐C2‐CUR‐2), while Q23, C137, S62, N144, and H151 (via a water mediator) formed hydrogen bonds with CUR (Figure 6l). Conservation analysis of the top 10 sequences showed complete conservation of S62, C137, N144, and S141, with the first three directly participating in CUR hydrogen bonding, underscoring their role in specific recognition (Figure 6m). Binding energy and dissociation constant (Kd) calculations based on crystal structures revealed values of −10.50 kcal/mol (19.7 nm) for rHuHF‐C2 and −11.16 kcal/mol (6.47 nm) for rHuHF‐71, a 3‐fold affinity enhancement for rHuHF‐71. Thus, the rHuHF‐C2 pocket serves not only as an exceptional universal platform for hydrophobic molecule binding but also, upon redesign, enables high‐affinity specific binding of target molecules.

2.6. A Rational and Universal Ferritin‐Based Design Strategy for Small‐Molecule‐Specific Binding Proteins

The above work demonstrates that the ferroxidase center in rHuHF can be redesigned for non‐specific or specific molecular binding. However, not all complex structures of rHuHF‐C2 with small molecules can be resolved; therefore, we adjusted our design strategy and proposed a new approach applicable to molecules (Figure 7). In detail, first, the small molecule is placed into the pocket of rHuHF‐C2 through molecular docking or by using the desired conformation to obtain the spatial coordinates of the complex. Then, amino acids within a certain range from the small molecule are selected, and the LigandMPNN program implemented on the Wemol [74] platform is employed for sequence generation. Subsequently, the top 10 sequences generated based on the “ligand_confidence” score are subjected to structural modeling using AlphaFold3. Thereafter, molecular docking is performed to dock the small molecule with each predicted structure. For each model, the small molecule with the highest docking score is selected, and its conformation is compared with that of the pre‐designed small molecule to calculate the RMSD value. If the RMSD is less than or equal to 1.5 Å, the protein corresponding to this sequence may specifically bind to the small molecule, and experimental validation is conducted. If the RMSD value is greater than 1.5 Å, this conformation and the protein are taken as the initial template for the next round of design until the RMSD value of the small molecule conformation before and after design is less than or equal to 1.5 Å. Using the method above, whether for hydrophobic or hydrophilic molecules, the binding mode and position of the molecule in the pocket will gradually stabilize during multiple rounds of design. Therefore, whether the molecule in the molecule‐protein complex used in the first design can spontaneously bind to the pocket and whether its conformation is optimal are not directly related to the conformation of the small molecule in the final design.

FIGURE 7.

FIGURE 7

Open in a new tab

Rational design strategy for engineering the rHuHF‐C2 pocket toward specific molecular binding using LigandMPNN, AlphaFold3, and molecular docking.

This strategy is not limited to ferritins. Our engineering method should, in principle, be applicable to numerous naturally occurring proteins, particularly enzymes, which possess catalytically active sites typically composed of long‐side‐chain amino acids. The mutation or substitution of these sites significantly expands the native catalytic center's volume, ultimately forming a remarkably large pocket at the original active site via the proposed method for either non‐specific or specific small‐molecule binding. A key advantage of this strategy lies in its utilization of naturally existing proteins as templates instead of de novo synthetic ones. As compared to artificially designed counterparts, these natural templates offer superior structural diversity (far exceeding de novo design's structural space), intrinsic folding stability (evolved through natural selection, eliminating extensive stability optimization), excellent biocompatibility (low immunogenicity, ideal for in vivo applications), and cost‐effective engineering (inherent cavities reduce design complexity, avoiding high computational/experimental costs of de novo construction). In essence, this strategy expands the template pool for small‐molecule‐binding protein design from limited synthetic scaffolds to a vast library of natural proteins, unlocking unparalleled diversity, stability, and applicability for biotechnology and biomedicine.

Ferritin has been widely applied in the delivery of anti‐tumor drugs due to two natural advantages. First, its hollow cage structure and controllable self‐assembly regulation allow drug molecules to be easily encapsulated into the internal cavity. Second, the ferritin receptor TfR1 is overexpressed on the surface of numerous tumor cells, endowing it with natural tumor‐targeting property and the ability to penetrate the blood‐brain barrier. Studies have shown that the recognition and binding sites of ferritin with TfR1 are all located on the outer surface of ferritin. Meanwhile, another receptor of ferritin, Nuclear Receptor Coactivator 4 (NCOA4), which is responsible for ferritin autophagy and related to the release of drugs encapsulated by ferritin, interacts with rHuHF at sites also located on the outer surface [26, 50]. The main objective of this work is to explore new binding sites for drug molecules to improve the encapsulation and delivery efficiency of ferritin for drug molecules. Notably, the catalytic oxidation active center of ferritin is located on the internal surface of rHuHF nanocage, and our design and modification of the rHuHF ferroxidase center will not affect the recognition and binding ability of ferritin with TfR1. Therefore, the designed rHuHF not only retains the advantages of natural ferritin in drug delivery but also provides a larger encapsulation space and even specific binding sites for drug molecules, effectively improving the drug encapsulation efficiency.

3. Conclusions

We have successfully completed the design of the ferroxidase center of rHuHF, enabling it to serve as a universal pocket that can widely bind a range of hydrophobic molecules, which was validated by resolving the crystal structures of complexes with seven hydrophobic molecules. Furthermore, using artificial protein design approaches, we further engineered the ferroxidase center to achieve specific binding of CUR molecules, with a significant enhancement in affinity. Ultimately, we propose a universal strategy for designing specific binding sites in the ferroxidase center of rHuHF, enabling the specific recognition and binding of target molecules within rHuHF.

Based on rHuHF‐C2 and other rHuHF mutants, an enhanced ferritin‐based drug encapsulation system can be further constructed. Without altering the fundamental properties and inherent advantages of the original rHuHF encapsulation system, this engineered strategy can effectively expand the internal space for molecular encapsulation, thereby increasing the encapsulation capacity of small‐molecule drugs. Furthermore, by combining the redesign of the interaction interfaces between rHuHF subunits, the transformation of rHuHF from its native 24‐mer assembly to dimeric, octameric, or other oligomeric forms can be achieved. This transformation shifts the drug loading mode of rHuHF from non‐specific encapsulation within the internal cavity to specific binding via the redesigned pockets, thereby greatly improving drug loading efficiency as compared to traditional ferritin‐based encapsulation systems. In addition, rHuHF possesses inherent advantages, including high recombinant expression yield, thermal stability, resistance to acid‐base stress, and biogenic nature. These characteristics make it an ideal template for specific small‐molecule binding in novel protein engineering; hence, this platform also holds broad application prospects in diverse fields such as small‐molecule specific recognition, artificial enzyme design, and material separation.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll73922‐sup‐0001‐SuppMat.docx.

SMLL-22-e73922-s001.docx (4.5MB, docx)

Acknowledgements

The work was supported by the National Natural Science Foundation of China (Grant Nos. 22577068 and 32372320), and the Research Project of Shanxi Province (Grant Nos. YDZJSX20231A009, 202404041101041). The authors thank the staff from the BL18U1, BL19U1 beamline of the National Facility for Protein Science in Shanghai (NFPS) and the staff of the BL17U1 beamline at Shanghai Synchrotron Radiation Facility (SSRF) for their assistance in x‐ray data collection.

Contributor Information

Nan Zhang, Email: zhangnan@nankai.edu.cn.

Guanghua Zhao, Email: gzhao@cau.edu.cn.

Yang Yun, Email: yunyang@sxu.edu.cn.

Hongfei Wang, Email: wanghf@sxu.edu.cn.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article. The atomic coordinate and structure factors for rHuHF‐C1, rHuHF‐C2, rHuHF‐C2‐CUR‐1, rHuHF‐C2‐CUR‐2, rHuHF‐C2‐ALY, rHuHF‐C2‐AZB, rHuHF‐71‐CUR, rHuHF‐C2‐NIS, rHuHF‐C2‐SEM, rHuHF‐C2‐DAC and rHuHF‐C2‐CAR under accession number 9VO7, 9VOC, 9VOD, 9VOE, 9VOF, 9VOM, 9VON, 21HO, 21KW, 21KV, and 21LE have been deposited in the Protein Data Bank (http://wwpdb.org/).

References

  • 1. Morris R., Black K. A., and Stollar E. J., “Uncovering Protein Function: From Classification to Complexes,” Essays in Biochemistry 66 (2022): 255–285, 10.1042/EBC20200108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Ferraro C., Dattilo M., Patitucci F., et al., “Exploring Protein‐Based Carriers in Drug Delivery: A Review,” Pharmaceutics 16 (2024): 1172, 10.3390/pharmaceutics16091172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Elzoghby A. O., Samy W. M., and Elgindy N. A., “Protein‐Based Nanocarriers as Promising Drug and Gene Delivery Systems,” Journal of Controlled Release 161 (2012): 38–49, 10.1016/j.jconrel.2012.04.036. [DOI] [PubMed] [Google Scholar]
  • 4. Edwardson T. G., Mori T., and Hilvert D., “Rational Engineering of a Designed Protein Cage for siRNA Delivery,” Journal of the American Chemical Society 140 (2018): 10439–10442, 10.1021/jacs.8b06442. [DOI] [PubMed] [Google Scholar]
  • 5. Xie L., Bai H., Song L., et al., “Structural and Photodynamic Studies on Nitrosylruthenium‐Complexed Serum Albumin as a Delivery System for Controlled Nitric Oxide Release,” Inorganic Chemistry 60 (2021): 8826–8837, 10.1021/acs.inorgchem.1c00762. [DOI] [PubMed] [Google Scholar]
  • 6. Bai H., Gong W., Pang Y., et al., “Synthesis, Cytotoxicity, and Biomacromolecule Binding: Three Isomers of Nitrosylruthenium Complexes With Bidentate Bioactive Molecules as Co‐Ligands,” International Journal of Biological Macromolecules 243 (2023): 125009, 10.1016/j.ijbiomac.2023.125009. [DOI] [PubMed] [Google Scholar]
  • 7. Jiang Y., Liang M., Svejkar D., et al., “Albumin‐Micelles via a One‐Pot Technology Platform for the Delivery of Drugs,” Chemical Communications 50 (2014): 6394, 10.1039/c4cc00616j. [DOI] [PubMed] [Google Scholar]
  • 8. Makwana H., Mastrotto F., Magnusson J. P., et al., “Engineered Polymer–Transferrin Conjugates as Self‐Assembling Targeted Drug Delivery Systems,” Biomacromolecules 18 (2017): 1532–1543, 10.1021/acs.biomac.7b00101. [DOI] [PubMed] [Google Scholar]
  • 9. Wong A. S. M., Czuba E., Chen M. Z., et al., “pH‐Responsive Transferrin‐pHlexi Particles Capable of Targeting Cells In Vitro,” ACS Macro Letters 6 (2017): 315–320, 10.1021/acsmacrolett.7b00044. [DOI] [PubMed] [Google Scholar]
  • 10. Miodragovic D., Merlino A., Swindell E. P., et al., “Arsenoplatin‐1 is a Dual Pharmacophore Anticancer Agent,” Journal of the American Chemical Society 141 (2019): 6453–6457, 10.1021/jacs.8b13681. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Lai W.‐F., Reddy O. S., Zhang D., Wu H., and Wong W.‐T., “Cross‐Linked Chitosan/Lysozyme Hydrogels With Inherent Antibacterial Activity and Tuneable Drug Release Properties for Cutaneous Drug Administration,” Science and Technology of Advanced Materials 24 (2023): 2167466, 10.1080/14686996.2023.2167466. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Tabe H., Fujita K., Abe S., et al., “Preparation of a Cross‐Linked Porous Protein Crystal Containing Ru Carbonyl Complexes as a CO‐Releasing Extracellular Scaffold,” Inorganic Chemistry 54 (2015): 215–220, 10.1021/ic502159x. [DOI] [PubMed] [Google Scholar]
  • 13. Santos‐Silva T., Mukhopadhyay A., Seixas J. D., Bernardes G. J. L., Romão C. C., and Romão M. J., “CORM‐3 Reactivity Toward Proteins: The Crystal Structure of a Ru(II) Dicarbonyl−Lysozyme Complex,” Journal of the American Chemical Society 133 (2011): 1192–1195, 10.1021/ja108820s. [DOI] [PubMed] [Google Scholar]
  • 14. Vernekar A. A., Berger G., Czapar A. E., et al., “Speciation of Phenanthriplatin and its Analogs in the Core of Tobacco Mosaic Virus,” Journal of the American Chemical Society 140 (2018): 4279–4287, 10.1021/jacs.7b12697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Chung Y. H., Cai H., and Steinmetz N. F., “Viral Nanoparticles for Drug Delivery, Imaging, Immunotherapy, and Theranostic Applications,” Advanced Drug Delivery Reviews 156 (2020): 214–235, 10.1016/j.addr.2020.06.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Zhang J., Cheng D., He J., Hong J., Yuan C., and Liang M., “Cargo Loading Within Ferritin Nanocages in Preparation for Tumor‐Targeted Delivery,” Nature Protocols 16 (2021): 4878–4896, 10.1038/s41596-021-00602-5. [DOI] [PubMed] [Google Scholar]
  • 17. Ferraro G., Ciambellotti S., Messori L., and Merlino A., “Cisplatin Binding Sites in Human H‐Chain Ferritin,” Inorganic Chemistry 56 (2017): 9064–9070, 10.1021/acs.inorgchem.7b01072. [DOI] [PubMed] [Google Scholar]
  • 18. Deng Q., Lin P., Gu H., Zhuang X., and Wang F., “Silk Protein‐Based Nanoporous Microsphere for Controllable Drug Delivery Through Self‐Assembly in Ionic Liquid System,” Biomacromolecules 25 (2024): 1527–1540, 10.1021/acs.biomac.3c01104. [DOI] [PubMed] [Google Scholar]
  • 19. Polizzi N. F. and Degrado W. F., “A Defined Structural Unit Enables De Novo Design of Small‐Molecule–Binding Proteins,” Science 369 (2020): 1227–1233, 10.1126/science.abb8330. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Lu L., Gou X., Tan S. K., et al., “De Novo Design of Drug‐Binding Proteins With Predictable Binding Energy and Specificity,” Science 384 (2024): 106–112, 10.1126/science.adl5364. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. An L., Said M., Tran L., et al., “Binding and Sensing Diverse Small Molecules Using Shape‐Complementary Pseudocycles,” Science 385 (2024): 276–282, 10.1126/science.adn3780. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Chen G., Zhao K., Li M., et al., “Rapid Clearance of Achiral Small‐Molecule Drugs Using De Novo‐Designed Proteins and Their Cyclic and Mirror‐Image Variants,” Nature Biomedical Engineering 9 (2025): 1775–1787, 10.1038/s41551-025-01404-w. [DOI] [PubMed] [Google Scholar]
  • 23. Yu B., Liu J., Cui Z., et al., “De Novo Design of Light‐Responsive Protein–Protein Interactions Enables Reversible Formation of Protein Assemblies,” Nature Chemistry 17 (2025): 1910–1919, 10.1038/s41557-025-01929-2. [DOI] [PubMed] [Google Scholar]
  • 24. Wang W., Xi H., Fu D., et al., “Growth Process of Fe–O Nanoclusters With Different Sizes Biosynthesized by Protein Nanocages,” Journal of the American Chemical Society 146 (2024): 11657–11668, 10.1021/jacs.3c13830. [DOI] [PubMed] [Google Scholar]
  • 25. Wang W., Liu X., Wang Y., et al., “Distinct Structural Characteristics Define a New Subfamily of Mycoplasma Ferritin,” Chinese Chemical Letters 33 (2022): 4952–4955, 10.1016/j.cclet.2022.03.119. [DOI] [Google Scholar]
  • 26. Hoelzgen F., Nguyen T. T. P., Klukin E., et al., “Structural Basis for the Intracellular Regulation of Ferritin Degradation,” Nature Communications 15 (2024): 3802, 10.1038/s41467-024-48151-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Montemiglio L. C., Testi C., Ceci P., et al., “Cryo‐EM Structure of the Human Ferritin–Transferrin Receptor 1 Complex,” Nature Communications 10 (2019): 1121, 10.1038/s41467-019-09098-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Song N., Zhang J., Zhai J., Hong J., Yuan C., and Liang M., “Ferritin: A Multifunctional Nanoplatform for Biological Detection, Imaging Diagnosis, and Drug Delivery,” Accounts of Chemical Research 54 (2021): 3313–3325, 10.1021/acs.accounts.1c00267. [DOI] [PubMed] [Google Scholar]
  • 29. He J., Fan K., and Yan X., “Ferritin Drug Carrier (FDC) for Tumor Targeting Therapy,” Journal of Controlled Release 311‐312 (2019): 288–300, 10.1016/j.jconrel.2019.09.002. [DOI] [PubMed] [Google Scholar]
  • 30. Palombarini F., Fabio E. D., Boffi A., Macone A., and Bonamore A., “Ferritin Nanocages for Protein Delivery to Tumor Cells,” Molecules 25 (2020): 825, 10.3390/molecules25040825. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Jiang B., Fang L., Wu K., Yan X., and Fan K., “Ferritins as Natural and Artificial Nanozymes for Theranostics,” Theranostics 10 (2020): 687–706, 10.7150/thno.39827. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Cosottini L., Geri A., Ghini V., et al., “Unlocking the Power of Human Ferritin: Enhanced Drug Delivery of Aurothiomalate in A2780 Ovarian Cancer Cells,” Angewandte Chemie International Edition 63 (2024): 202410791, 10.1002/anie.202410791. [DOI] [PubMed] [Google Scholar]
  • 33. Jin Y., He J., Fan K., and Yan X., “Ferritin Variants: Inspirations for Rationally Designing Protein Nanocarriers,” Nanoscale 11 (2019): 12449–12459, 10.1039/C9NR03823J. [DOI] [PubMed] [Google Scholar]
  • 34. Wang W., Wang L., Chen H., et al., “Selective Elimination of the Key Subunit Interfaces Facilitates Conversion of Native 24‐mer Protein Nanocage Into 8‐mer Nanorings,” Journal of the American Chemical Society 140 (2018): 14078–14081, 10.1021/jacs.8b09760. [DOI] [PubMed] [Google Scholar]
  • 35. Zhang S., Zang J., Wang W., et al., “Conversion of the Native 24‐mer Ferritin Nanocage Into its Non‐Native 16‐mer Analogue by Insertion of Extra Amino Acid Residues,” Angewandte Chemie International Edition 55 (2016): 16064–16070, 10.1002/anie.201609517. [DOI] [PubMed] [Google Scholar]
  • 36. Zhao S., Zang J., Zhang X., et al., “‘Silent’ Amino Acid Residues at Key Subunit Interfaces Regulate the Geometry of Protein Nanocages,” ACS Nano 10 (2016): 10382–10388. [DOI] [PubMed] [Google Scholar]
  • 37. Zang J., Chen H., Zhang X., et al., “Disulfide‐Mediated Conversion of 8‐mer Bowl‐Like Protein Architecture Into Three Different Nanocages,” Nature Communications 10 (2019): 778, 10.1038/s41467-019-08788-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38. Gu C., Mi Y., Zhang T., Zhao G., and Wang S., “Construction of Robust Protein Nanocage by Designed Disulfide Bonds for Active Cargo Molecules Protection in the Gastric Environment,” Journal of Colloid and Interface Science 678 (2024): 637–647, 10.1016/j.jcis.2024.08.196. [DOI] [PubMed] [Google Scholar]
  • 39. Wang W., Wang L., Li G., Zhao G., Zhao X., and Wang H., “AB Loop Engineered Ferritin Nanocages for Drug Loading Under Benign Experimental Conditions,” Chemical Communications 55 (2019): 12344–12347, 10.1039/C9CC05247J. [DOI] [PubMed] [Google Scholar]
  • 40. Ahn B., Lee S.‐G., Yoon H. R., et al., “Four‐Fold Channel‐Nicked Human Ferritin Nanocages for Active Drug Loading and pH‐Responsive Drug Release,” Angewandte Chemie 130 (2018): 2959–2963, 10.1002/ange.201800516. [DOI] [PubMed] [Google Scholar]
  • 41. Xia H., Xu H., Wang J., et al., “Heat Sensitive E‐Helix Cut Ferritin Nanocages for Facile and High‐Efficiency Loading of Doxorubicin,” International Journal of Biological Macromolecules 253 (2023): 126973, 10.1016/j.ijbiomac.2023.126973. [DOI] [PubMed] [Google Scholar]
  • 42. Swift J., Wehbi W. A., Kelly B. D., Stowell X. F., Saven J. G., and Dmochowski I. J., “Design of Functional Ferritin‐Like Proteins With Hydrophobic Cavities,” Journal of the American Chemical Society 128 (2006): 6611–6619, 10.1021/ja057069x. [DOI] [PubMed] [Google Scholar]
  • 43. Incocciati A., Kubes J., Piacentini R., et al., “Hydrophobicity‐Enhanced Ferritin Nanoparticles for Efficient Encapsulation and Targeted Delivery of Hydrophobic Drugs to Tumor Cells,” Protein Science 32 (2023): 4819, 10.1002/pro.4819. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44. Budiarta M., Roy S., Katenkamp T., Feliu N., and Beck T., “Overcoming Non‐Specific Interactions for Efficient Encapsulation of Doxorubicin in Ferritin Nanocages for Targeted Drug Delivery,” Small 19 (2023): 2205606, 10.1002/smll.202205606. [DOI] [PubMed] [Google Scholar]
  • 45. Lu C., Maity B., Peng X., et al., “Design of a Gold Clustering Site in an Engineered Apo‐Ferritin Cage,” Communications Chemistry 5 (2022): 39, 10.1038/s42004-022-00651-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Maity B., Abe S., and Ueno T., “Observation of Gold Sub‐Nanocluster Nucleation Within a Crystalline Protein Cage,” Nature Communications 8 (2017): 14820, 10.1038/ncomms14820. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Kanekiyo M., Wei C.‐J., Yassine H. M., et al., “Self‐Assembling Influenza Nanoparticle Vaccines Elicit Broadly Neutralizing H1N1 Antibodies,” Nature 499 (2013): 102–106, 10.1038/nature12202. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. Tai W., Chai B., Feng S., et al., “Development of a Ferritin‐Based Nanoparticle Vaccine Against the SARS‐CoV‐2 Omicron Variant,” Signal Transduction and Targeted Therapy 7 (2022): 173, 10.1038/s41392-022-01041-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49. Weidenbacher P. A.‐B., Sanyal M., Friedland N., et al., “A Ferritin‐Based COVID‐19 Nanoparticle Vaccine That Elicits Robust, Durable, Broad‐Spectrum Neutralizing Antisera in Non‐Human Primates,” Nature Communications 14 (2023): 2149, 10.1038/s41467-023-37417-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Fu D., Wang W., Zhang Y., et al., “Self‐Assembling Nanoparticle Engineered From the Ferritinophagy Complex as a Rabies Virus Vaccine Candidate,” Nature Communications 15 (2024): 8601, 10.1038/s41467-024-52908-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51. Liu M., Jin D., Yu W., et al., “Enhancing Tumor Immunotherapy by Multivalent Anti‐PD‐L1 Nanobody Assembled via Ferritin Nanocage,” Advanced Science 11 (2024): 2308248, 10.1002/advs.202308248. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52. Dong Y., Ma Y., Li X., Wang F., and Zhang Y., “ERK‐Peptide‐Inhibitor‐Modified Ferritin Enhanced the Therapeutic Effects of Paclitaxel in Cancer Cells and Spheroids,” Molecular Pharmaceutics 18 (2021): 3365–3377, 10.1021/acs.molpharmaceut.1c00303. [DOI] [PubMed] [Google Scholar]
  • 53. Zhai M., Wang Y., Zhang L., et al., “Glioma Targeting Peptide Modified Apoferritin Nanocage,” Drug Delivery 25 (2018): 1013–1024, 10.1080/10717544.2018.1464082. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54. Wang W., Zhang Y., Zhao G., and Wang H., “Ferritin With Atypical Ferroxidase Centers Takes B‐Channels as the Pathway for Fe2+ Uptake From Mycoplasma,” Inorganic Chemistry 60 (2021): 7207–7216, 10.1021/acs.inorgchem.1c00265. [DOI] [PubMed] [Google Scholar]
  • 55. Ebrahimi K. H., Hagedoorn P.‐L., and Hagen W. R., “Unity in the Biochemistry of the Iron‐Storage Proteins Ferritin and Bacterioferritin,” Chemical Reviews 115 (2015): 295–326, 10.1021/cr5004908. [DOI] [PubMed] [Google Scholar]
  • 56. Ebrahimi K. H., Bill E., Hagedoorn P.‐L., and Hagen W. R., “The Catalytic Center of Ferritin Regulates Iron Storage via Fe(II)‐Fe(III) Displacement,” Nature Chemical Biology 8 (2012): 941–948, 10.1038/nchembio.1071. [DOI] [PubMed] [Google Scholar]
  • 57. Duan M., Wang Y., Zang J., et al., “Construction of an Artificial Photosynthesis System With A Single CdS QDs‐Ferritin Hybrid Molecule,” Small Methods 9 (2025): 2400915, 10.1002/smtd.202400915. [DOI] [PubMed] [Google Scholar]
  • 58. Ebrahimi K. H., Hagedoorn P.‐L., and Hagen W. R., “A Conserved Tyrosine in Ferritin is a Molecular Capacitor,” Chembiochem 14 (2013): 1123–1133, 10.1002/cbic.201300149. [DOI] [PubMed] [Google Scholar]
  • 59. Hempstead P. D., Yewdall S. J., Fernie A. R., et al., “Comparison of the Three‐Dimensional Structures of Recombinant Human H and Horse L Ferritins at High Resolution 1 1Edited by R. Huber,” Journal of Molecular Biology 268 (1997): 424–448, 10.1006/jmbi.1997.0970. [DOI] [PubMed] [Google Scholar]
  • 60. Maglic J. B. and Lavendomme R., “MoloVol: An Easy‐To‐Use Program for Analyzing Cavities, Volumes and Surface Areas of Chemical Structures,” Journal of Applied Crystallography 55 (2022): 1033–1044, 10.1107/S1600576722004988. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61. Banez M. J., Geluz M. I., Chandra A., et al., “A Systemic Review on the Antioxidant and Anti‐Inflammatory Effects of Resveratrol, Curcumin, and Dietary Nitric Oxide Supplementation on Human Cardiovascular Health,” Nutrition Research 78 (2020): 11–26, 10.1016/j.nutres.2020.03.002. [DOI] [PubMed] [Google Scholar]
  • 62. Hewlings S. J. and Kalman D. S., “Curcumin: A Review of its Effects on Human Health,” Foods 6 (2017): 92. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63. Zhang N., Lin R., Gao W., et al., “Curcumin Modulates PTPRZ1 Activity and RNA m6A Modifications in Neuroinflammation‐Associated Microglial Response,” Advanced Science 12 (2025): 2405263, 10.1002/advs.202405263. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64. Fang F., Wang S., Song Y., et al., “Continuous Spatiotemporal Therapy of a Full‐API Nanodrug via Multi‐Step Tandem Endogenous Biosynthesis,” Nature Communications 14 (2023): 1660, 10.1038/s41467-023-37315-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65. Pandolfi L., Bellini M., Vanna R., et al., “H‐Ferritin Enriches the Curcumin Uptake and Improves the Therapeutic Efficacy in Triple Negative Breast Cancer Cells,” Biomacromolecules 18 (2017): 3318–3330, 10.1021/acs.biomac.7b00974. [DOI] [PubMed] [Google Scholar]
  • 66. Gu C., Zhang T., Lv C., Liu Y., Wang Y., and Zhao G., “His‐Mediated Reversible Self‐Assembly of Ferritin Nanocages Through Two Different Switches for Encapsulation of Cargo Molecules,” ACS Nano 14 (2020): 17080–17090, 10.1021/acsnano.0c06670. [DOI] [PubMed] [Google Scholar]
  • 67. Morasso C., Truffi M., Tinelli V., et al., “Exploring the Anti‐Inflammatory Effects of Curcumin Encapsulated Within Ferritin Nanocages: A Comprehensive In Vivo and In Vitro Study in Alzheimer's Disease,” Journal of Nanobiotechnology 22 (2024): 718, 10.1186/s12951-024-02897-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68. Gong W., Pang Y., Wang C., Wang W., and Wang H., “Photoinduced NO Release of [Fe2(µ‐SL)2(NO)4]Complexes and Their Protein Adducts: Insights From Structure, Cytotoxicity, and Photodynamic Studies,” Inorganic Chemistry Frontiers 12 (2025): 3490–3504, 10.1039/D5QI00255A. [DOI] [Google Scholar]
  • 69. Nikolova T., Roos W. P., Kramer O. H., Strik H. M., and Kaina B., “Chloroethylating Nitrosoureas in Cancer Therapy: DNA Damage, Repair and Cell Death Signaling,” Biochimica Et Biophysica Acta: reviews on Cancer 1868 (2017): 29–39. [DOI] [PubMed] [Google Scholar]
  • 70. Wang B., Zhang S., Shen Z.‐T., et al., “Core‐Shell Reactor Partitioning Enzyme and Prodrug by ZIF‐8 for NADPH‐Sensitive In Situ Prodrug Activation,” Angewandte Chemie 135 (2023): 202314025, 10.1002/ange.202314025. [DOI] [PubMed] [Google Scholar]
  • 71. Dauparas J., Lee G. R., Pecoraro R., et al., “Atomic Context‐Conditioned Protein Sequence Design Using LigandMPNN,” Nature Methods 22 (2025): 717–723, 10.1038/s41592-025-02626-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 72. Abramson J., Adler J., Dunger J., et al., “Accurate Structure Prediction of Biomolecular Interactions With AlphaFold 3,” Nature 630 (2024): 493–500, 10.1038/s41586-024-07487-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 73. Crooks G. E., Hon G., Chandonia J.‐M., and Brenner S. E., “WebLogo: A Sequence Logo Generator: Figure 1,” Genome Research 14 (2004): 1188–1190, 10.1101/gr.849004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 74. Liu H., Yan X., Fang H., Ge H., and Hou X., “WeMol: A Cloud‐Based and Zero‐Code Platform for AI‐Driven Molecular Design and Simulation,” Journal of Chemical Information and Modeling 66 (2026): 2369–2381, 10.1021/acs.jcim.6c00014. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supporting File: smll73922‐sup‐0001‐SuppMat.docx.

SMLL-22-e73922-s001.docx (4.5MB, docx)

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article. The atomic coordinate and structure factors for rHuHF‐C1, rHuHF‐C2, rHuHF‐C2‐CUR‐1, rHuHF‐C2‐CUR‐2, rHuHF‐C2‐ALY, rHuHF‐C2‐AZB, rHuHF‐71‐CUR, rHuHF‐C2‐NIS, rHuHF‐C2‐SEM, rHuHF‐C2‐DAC and rHuHF‐C2‐CAR under accession number 9VO7, 9VOC, 9VOD, 9VOE, 9VOF, 9VOM, 9VON, 21HO, 21KW, 21KV, and 21LE have been deposited in the Protein Data Bank (http://wwpdb.org/).

Articles from Small (Weinheim an Der Bergstrasse, Germany) are provided here courtesy of Wiley

RESOURCES