Ferritin-metallodrugs nanocomposites: light and shadow of drugs loading methods

1 Introduction

Last decades have seen a thriving development of metal-based drugs with different pharmacological properties due to intriguing structural properties conferred by the metal center (Allardyce and Dyson, 2016; Temesgen et al., 2023). The interest in designing new metal compounds relies on their involvement in multiple biological processes that makes them eligible as potential drugs (Mjos and Orvig, 2014; Miranda, 2022; Peña et al., 2022; Vitali et al., 2025b; Wei et al., 2025). Nonetheless, the growing appeal for metallodrugs travels together with a series of deficiencies, such as poor solubility in aqueous media, general toxicity and non-specific delivery to the final target, which create an indistinct barrier between therapeutic and toxic doses (Medici et al., 2018). For these reasons, scientists have been encouraged to explore strategies able to address those limitations while preserving drug efficacy. The use of a drug delivery system is one of the most common approaches. Ideally, a delivery system exhibits different abilities including biocompatibility, stability in blood stream, capability of solubilizing insoluble drugs while preserving their structure and integrity, tumor targeting capacity and suitable pharmacokinetic properties. All these features allow safe driving to the final target (Tibbitt et al., 2016; Chamundeeswari et al., 2019). Different biomaterials boast those features. Natural polymers are often more suitable than synthetic polymers because of their intrinsic biocompatibility (Prestwich and Luo, 2001; Trucillo, 2024) and ability of transport-barriers crossing and site-specific binding.

In this contest, there is a growing interest in designing protein-based biomaterials as tools for drug transport within the body. Proteins are endogenous molecules and can guarantee all the desired features for a nanocarrier (Elzoghby et al., 2012; Jain et al., 2018; Hua et al., 2024). Among them, ferritins, hollow cage spherical proteins responsible for iron storage within cells (∼2,500 Fe atoms/cage), have been tested as drug nanocarriers because of their clear advantages over traditional inorganic/organic nanocarriers, such as superior safety profile, established in vivo pharmacokinetics, easy scale-up, and high drug loading efficiency. The high storage capacity, properly related to the peculiar architecture, is coupled with a series of advantages. For instance, human H-ferritin (HFn), a ferritin subtype made of only heavy (H) chains, emerged as TfR1 (transferrin receptor 1) ligand, a receptor frequently overexpressed on many cancer cell types (Li et al., 2010; Cheng et al., 2020). L-ferritin, made instead of only light (L) chains, has been indicated as a preferential partner of Scara5, an ubiquitarian receptor responsible for iron supply in specific tumors (Alkhateeb et al., 2013; Geninatti et al., 2015). Moreover, ferritin (Fn) surface can be chemically or genetically functionalized with peptides or small organic molecules to improve its target specificity (Wang et al., 2025). Literature offers several examples of efficient delivery upon metallodrug loading/complexation with ferritin. This mini-review aims to explore the use of Fn-based formulations for metallodrug delivery highlighting strength points and weaknesses of this potential nanocarrier.

Fn has been used as a carrier for several molecules, including drugs (Lv et al., 2021; Yazdian-Robati et al., 2022; Bhatt et al., 2024), nutrients (Zang et al., 2017; Hu et al., 2023), catalysts (Maity et al., 2015), nanoparticles (Bellini et al., 2020), peptides (Li et al., 2021) and even proteins (Tetter and Hilvert, 2017; Shuvaev et al., 2018; Chakraborti et al., 2019). There are different approaches that can be employed for the loading of bioactive molecules (He et al., 2019; Mohanty et al., 2022; Zhang and Fan, 2025). These methods often depend on structure, chemical nature and size of cargos. Examples of common loading strategies include: i) passive diffusion, which allows the entering of small drugs within Fn cage through protein channels towards a natural process, ii) passive loading, which promotes the loading upon channel enlargement using temperature, pH or ionic strength, and iii) encapsulation, which exploits Fn ability to be disassembled and reassembled in the presence of extreme conditions (alkaline/acid pHs, denaturant agents like urea or SDS) (Figure 1) (Lucignano and Ferraro, 2024).

Different works proved the efficiency of each method. For instance, Cosottini and co-workers developed HFn-conjugates (HFn-AF) with the antiarthritic and anticancer gold(I) compound Auranofin (AF) by simple addition of a molar excess of AF with respect to ferritin (Cosottini et al., 2023). They found 14 gold atoms per cage. The gold atoms covalently bound to the side chains of solvent exposed Cys90 and Cys102. Despite gold distribution on the protein external surface, which could interfere with HFn recognition by TfR-1, HFn-AF was able to be internalized by A2780 cells, to preserve AF cytotoxicity, killing A2780 cells at a concentration lower than that of free AF, and to overcome cisplatin resistance in A2780cis cells. Similar results were obtained preparing a nanocomposite formed by HFn with Aurothiomalate (AuTM), a gold(I) drug approved by FDA for the treatment of rheumatoid arthritis (Kean and Kean, 2008) and recently tested as an anticancer agent (ClinicalTrials.gov ID: NCT00575393, NCT01383668) (Cosottini et al., 2024).

Notably, when the same approach was used to prepare HFn conjugates with the well-known anticancer drug oxaliplatin (OxaPt) a loss of cytotoxic activity on A2780 cell line was observed (Vitali et al., 2025a). The different behavior of AF, AuTM and OxaPt, when bound to HFn, was explained considering the fate of the metal-containing fragments in the bioconjugates once internalized by cells. In fact, platinum, even if bound to the same residues of gold (Cys90 and Cys102), is not released by HFn and thus is unable to reach its final target, i.e., the nuclear DNA. On the contrary, HFn-AF efficiently metalated the C-terminal dodecapeptide of thioredoxin reductase, which is the main target of AF.

The irreversible binding of metal moieties to HFn residues can completely silence the biological activity of a metallodrug. This probably represents the major limit of the drug passive diffusion method. One of the strategies that can be adopted to overcome this issue is to move from passive diffusion to passive loading or encapsulation.

HFn possesses a “natural drug entry channel” on its shell (Jiang et al., 2020), the opening of which is thermosensitive. A series of Pt (IV) prodrugs were loaded within HFn (Jiang et al., 2024) using high temperatures (55 °C). Pt (IV) compounds have a reduced reactivity towards thiol groups when compared to Pt (II) complexes like cisplatin. This different behavior could prevent biological inactivation caused by Pt irreversible binding to protein residues. As an example, a Pt-HFn nanocomposite called HFn-Pt (IV)-3, containing 21 Pt atoms per cage, was prepared using this technique. Its X-ray structure revealed only two Pt binding sites (Cys90 and Cys102 side chains) suggesting the presence of free Pt-containing fragments within the bulk. It is likely that these species are responsible for the cytotoxicity of the nanocomposite on human ESCC (esophageal squamous cell carcinoma) cell lines.

Similarly, the incubation of recombinant L-chain apo-Fn from horse liver with a Ru-based CO releasing molecule (CORM) at 50 °C led to the formation of a nanocomposite able to increase the half-life for CO release from the CORM (Fujita et al., 2014). The complexation also improved the biological activity of the CORM. Indeed, the nanocomposite had NF-κB (nuclear factor) activation ability 10-fold higher than that of the free compound.

Prof. Merlino’s research group gained experience in the development of ferritin-based nanoconjugates obtained by loading metallodrugs within horse spleen Fn (L-chains, hsFn hereafter) by using the alkaline encapsulation protocol (Petruk et al., 2019; Ferraro et al., 2016; Monti et al., 2017; Ferraro et al., 2018b; Ferraro et al., 2021). This Fn subtype lacks the three reactive cysteines present in HFn and has an intrinsic advantage. The encapsulation protocol consists of cage disassembly at alkaline pH (∼13.0), incubation of disassembled protein with the metallodrug in large molar excess, and reassembly at physiological conditions. Upon reassembly, drug molecules remain trapped inside the cage. Even if this approach does not prevent metal fragments from covalently binding protein residues, it can preserve metallodrug biological properties counting on the huge number of free molecules that can be trapped in the bulk. In this respect, X-ray crystallography has been a powerful tool to assess the potential binding of metal moieties to protein residues and has been used to characterize several ferritin-metal complexes conjugates. Examples are reported in Table 1. A successful example is the encapsulation of three oxo-bridged gold (III) compounds within hsFn, which allowed the loading of an amount of gold between 400 and 700 atoms per cage (Ferraro et al., 2016; Monti et al., 2017). The X-ray structures of the three gold-loaded nanocomposites showed gold atoms, with partial occupancies, covalently bound to His and Cys residues located on the inner protein surface, suggesting that 80–90% of the encapsulated drug was free in the bulk. The evaluation of the biological activity on three cancer cell lines, HepG2, HeLa and MCF-7, in comparison to three normal cell lines, demonstrated the effective cytotoxicity of the three gold complexes, also when encapsulated, with IC₅₀ values lower than those of the free forms but in favor of selectivity toward tumor lines. Notably, hsFn external surface was unaffected by drug encapsulation, thus the protein should retain its ability to be recognized by Scara5 receptor.

TABLE 1. Examples of X-ray structures of ferritins upon complexation/encapsulation of metal-based compound reported in the protein data bank (PDB).

To preserve Fn external surface and to prevent metal binding, the reactive cysteines of HFn can be substituted by site directed mutagenesis. Cosottini et al. used this approach as indirect proof of gold binding site localization (Cosottini et al., 2023). The mutant turned out to manifest a complete loss of the gold binding ability and, consequently, of drug loading.

Conversely, Ueno and co-workers demonstrated how multinuclear metal complexes could be prepared by deletion or introduction of key residues, such as His, Glu, and Cys, at appropriate positions on protein surface (Abe et al., 2008; Maity et al., 2015; Lu et al., 2022). Apo-ferritin from horse liver (rHLFr) was engineered by replacing His114 with a non-coordinating alanine to fabricate a trinuclear Pd cluster at the 3-fold axis of the protein (Abe et al., 2008). By varying the positions of Cys and His residues, various coordination arrangements of the Pd complexes were achieved (Wang et al., 2011).

Applying the same approach, the mutant apo-E45C/R52C-rHLFr served as scaffold to build a sub-nano Au₁₀ cluster at the same site (Maity et al., 2017).

Thus, the obtainment of an efficient Fn-based carrier needs a combination of several factors including the loading method, the type of Fn and possible surface modifications. In this respect, it should be noted that Fn surface modifications involve not only site directed mutagenesis but also chemical functionalization. Many diseases are characterized by overexpression of specific markers often manifestation of higher metabolic requests from unhealthy cells compared to those of healthy cells. These markers, particularly those exposed to the cellular external surface, can be exploited for the development of nanocarriers able to specifically recognize them and to promote targeted drug delivery. Fn external surface is a blank canvas and can be functionalized genetically or chemically to form chimeric nanocages with improved selectivity.

Metabolically reprogrammed cancer cells overexpress the folate receptor α (FRα) which is an attractive anticancer drug target for a range of solid tumors, including ovarian, lung and breast cancers (Scaranti et al., 2020). Tesarova et al. took advantage of this metabolic glitch to produce Fn nanocarriers whose external surface was conjugated with folic acid (FA) with the aim of further improving the uptake of an anticancer Ni compound, Ni2 (Tesarova et al., 2019). From one side, encapsulation significantly decreased haemotoxicity of Ni2 compound, reducing the occurrence of hemolysis as main drug side effect. On the other hand, FA conjugation allowed a selective uptake of the loaded drug. In fact, Ni2-Fn-FA was mostly internalized from malignant cells with high expression of FR (T-47D, human breast cancer cell line) and avoided non-malignant cells (HBL-100, human breast epithelial cell line) and malignant cells with low expression of FR (MCF-7, human breast cancer cell line).

2 Discussion

The last years have seen a great interest in the use of Fn as a biocompatible nanocarrier for bioactive molecules. Despite the huge number of works focused on the loading of molecules with intriguing pharmacological properties within Fn, only few of them explore the use of this protein as a delivery system for metallodrugs. There are three protocols commonly used for metallodrug loading within Fn, each one characterized by pros and cons.

The easiest loading way exploits passive diffusion of small drug molecules through the two types of protein channels that connect the outer surface with the inner cavity (Zhang et al., 2017). As a protein involved in iron storage, Fn internal surface is rich in acidic residues, which can be leveraged to bind and retain metal complexes such as those of Pt, Au, or Ru. So, the combination of the protein cage structure and its natural affinity for metal ions facilitates metallodrug loading through this route. As a matter of fact, the diffusion process within the protein is not purely passive, it does not depend exclusively on the concentration gradient between the interior and the exterior of the protein, it is instead tightly regulated by electrostatic interactions and molecular polarity. Positively charged probes are selectively able to penetrate the protein through the 3-fold channels, whereas non-polar molecules preferentially associate with hydrophobic regions located on the protein external surface. The passive diffusion of drug molecules within Fn core becomes more efficient in the passive loading methods, which enables entrance of drug molecules within the cage upon channels diameter enlargement. Nonetheless, these routes cannot stabilize therapeutic molecules with large molecular weights, which prevalently adsorb on the surface, likely impairing ferritin-receptor recognition.

A widely employed strategy for loading molecules within Fn relies on the disassembly–reassembly process, in which the protein cage is first dissociated under extreme physico-chemical conditions and then reassembled at physiological conditions in the presence of the drug which remains trapped inside the bulk. This approach ensures the loading of a huge amount of drug molecules but shows some drawbacks. Because of the harsh conditions, this method frequently results in partial denaturation of the protein structure which causes a huge loss of protein, significantly lowering the amount of recovered sample (Belletti et al., 2017). In addition, it is not obvious that the loaded drug is stable in these strict experimental conditions.

The subtype of Fn chosen as a potential carrier or the design of an engineered protein can deeply influence the loading. Indeed, some ferritins expose reactive residues on the external surface that covalently, and sometimes irreversibly, anchor metal-containing fragments. So, a successful result seems to rely on the combination of several factors which create a balanced drug-loaded nanocomposite.

Despite the consistent amount of information available in literature on this system, a clearer understanding of the pharmacological behavior of the drug–nanocarrier complex is still required. It is crucial to determine whether the therapeutic effect depends on the release of the drug from the carrier or whether the drug-loaded complex itself is the active therapeutic unit. Additionally, predicting the interactions between the drug–carrier complexes and the organs responsible for clearing foreign materials from the body is essential, as improper biodistribution may significantly compromise therapeutic efficacy. Taking these factors into account would contribute to the rational design of highly efficient anticancer metallodrug nanocarriers.

Another critical aspect to consider is the short plasma half-life of Fn after systemic administration, which limits its tumor-targeting capability (Falvo et al., 2016; Mazzucchelli et al., 2017).

Over the past decade, different nanodrug delivery systems based on Fn have been reported. Multiple studies have confirmed that Fn nanocarriers can enhance the bioavailability of insoluble drugs, promote tumor-specific accumulation and reduce the drug toxicity toward healthy tissues. Nonetheless, major challenges persist. To address these issues various chemical and genetic modification strategies have recently been proposed aimed at improving therapeutic applicability of ferritin-based nanosystems (Sevieri et al., 2023).

Therefore, achieving an optimal balance between structural stability, drug-loading efficiency, and biological performance remains the key challenge in translating Fn-based metallodrug carriers into clinically viable therapeutics. Indeed, further efforts are still needed to fully exploit ferritin potential as a reliable and versatile platform for metallodrug delivery.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Sylvia Mary Draper, Trinity College Dublin, Ireland

Reviewed by: Rabindra Behera, National Institute of Technology Rourkela, India