Ligand biodegradation-induced surface reconstruction of magnetite nanoparticles: potentially overlooked toxicity

Abstract

The coating of nanoparticles (NPs) with biodegradable ligands has been considered an efficient way to improve the biocompatibility of NPs and to decrease their biopersistence. However, the role of ligand biodegradation in reshaping the core material and the resulting change in toxicity have not been fully explored. In this study, magnetite (Fe₃O₄) NPs coated with a high or low loading of hyaluronic acid (termed Fe₃O₄–HA_H and Fe₃O₄–HA_L) were synthesized. Fe₃O₄–HA_R was obtained after the enzymatic degradation of Fe₃O₄–HA_H by a hyaluronidase, which had a coating weight similar to that of Fe₃O₄–HA_L, as verified by thermogravimetric analysis. Interestingly, among the three NPs, Fe₃O₄–HA_R induced the highest rate of cellular damage. In particular, 2 times more hydroxyl radicals were detected in cells stressed by Fe₃O₄–HA_R than in cells stressed by Fe₃O₄–HA_H, despite the highest cellular uptake being observed for Fe₃O₄–HA_H because of the ligand–receptor interactions between hyaluronic acid and membrane receptors. The structural characterization results revealed that ligand degradation induced notable surface reconstruction of the Fe₃O₄ core, showing a characteristic Fe₃O₄@Fe₂O₃ structure with surface-bound ferrous ions. This unique structure endowed Fe₃O₄–HA_R with superior competency in initiating the Fenton reaction and accelerating the Fe(III)/Fe(II) cycle. Our results emphasize the importance of tracing the in vivo biotransformation of NPs from all of the different architectural parts, particularly deciphering the interplay among the core, shell, surface ligands, and surrounding ions.

---

Environmental significance

Generally, coating with biodegradable ligands is regarded as an efficient way to enhance the biocompatibility of nanoparticles and reduce their potential persistence in vivo. However, how the biodegradation of surface coatings is implicated in core material reconstruction is not fully understood and may alter their surface reactivity and potential toxicity. Our results revealed that the enzymatic degradation of hyaluronic acid, a commonly used ligand, induced the surface reconstruction of magnetite nanoparticles. These reconstructed nanoparticles exhibited a high capability for producing hydroxyl radicals, thus boosting cytotoxicity. Therefore, it is imperative to monitor nanoparticle biotransformation from all of the different architectural parts, with particular concern on the interplay among the core, shell, surface coating, and surrounding solutes.

1. Introduction

The design of nanoparticles (NPs) is important for their effective application, including tuning their intrinsic characteristics (i.e., chemical composition,¹ size,² and shape³) and employing a variety of surface ligands.⁴ Ligands play important roles in maintaining the colloidal stability of NPs and the encoding of various functionalities.⁵ For biomedical applications, biodegradability is highly desired, and various biodegradable ligands have been used to decorate NPs, such as poly(ethylene glycol) (which has a high degradation efficacy for low-molecular-weight structures),⁶ hyaluronic acid,⁷ and citric acid.⁸ However, the change in physicochemical properties of NPs due to ligand degradation may jeopardize the function of NPs and change their biodistribution, as well as cytotoxicity.^7,9 From the perspective of nanosafety, it is important to consider the potential complications caused by ligand biotransformation.

The current knowledge with respect to ligand biotransformation is described in the context of degradation mechanisms, resulting in changed NP properties and complicated biological outcomes. Ligand alteration may occur via hydrolysis in an acidic biological medium, cleavage by biogenerated reactive oxygen species (ROS), enzymatic degradation and displacement by other biomolecules.^7,10,11 For example, the shell of an amphiphilic polymer on a gold NP surface, composed of poly(isobutylene-alt-maleic anhydride)-graft-dodecyl, underwent degradation by proteolytic enzymes in the liver.¹² The citrate ligand, which was physically adsorbed on gold NPs, can be replaced by lipid molecules (i.e., 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) during interactions with the membrane.¹³ Consequently, the loss of the surface coating resulted in variations in the surface charges and topologies, decreased colloidal stability and agglomerated NPs.^7,14 The important biological consequences of ligand degradation include impaired target function of NPs, change in persistence in the body,⁷ and increased toxicity.¹⁴ For instance, an intact dense polymer coating endows NPs with excellent colloidal stability, thus exhibiting a longer blood circulation time than NPs with degraded coatings or with less coating.¹⁵ A dense coating on a lead sulfide surface increased particle biocompatibility, whereas the loss of the coating may exaggerate the potential of lead ion release, thus causing long-term in vivo toxicity.¹⁶ Nevertheless, how the degradation of the surface coating influences the reshaping of the core material has not been fully explored. In particular, the dissolution of the core material, surface remodeling, passivation, and crystalline reorganization may occur simultaneously during surface ligand transformation. Systematic monitoring of core material changes during ligand transformation may provide useful information for assessing biodurability, biocompatibility and biosafety in the short and long term.

In this study, magnetite (Fe₃O₄) NPs were chosen because of their potential biomedical applications, such as magnetic bioseparation, detection, clinical diagnosis and therapy.^7,17 Polyanionic hyaluronic acid (HA) was used as a coating ligand, which consists of two alternating units, namely, glucuronic acid and N-acetylglucosamine (Fig. S1†). HA is present in the skin, joints, and cornea at high concentrations, and it has been used to assemble nanoparticles to achieve highly efficient targeted therapy and imaging of tumors.¹⁸ Moreover, hyaluronidase (HAase), which can break down hyaluronic acid (Fig. S1†), is present in various organs and body fluids,¹⁹ and the enzyme activity varies considerably depending on the microenvironment pH, HA content, organ and pathological status.¹⁹ Hence, the effect of HA degradation by HAase on the changes in the Fe₃O₄ core was systematically investigated.

The aim of this study was 1) to investigate the alterations of the physicochemical properties and cytotoxicity of Fe₃O₄ NPs after enzymatic degradation of HA as the coating ligand and 2) to elucidate the reshaping progress of the Fe₃O₄ core during HA degradation. Our results emphasize the importance of assessing NP biotransformation, and considerable efforts should be made to comprehend the essential mechanisms relating to the fate of NPs in biological environments.

2. Materials and methods

2.1 Materials

Ferric chloride hexahydrate (FeCl₃·6H₂O), ferrous dichloride tetrahydrate (FeCl₂·4H₂O) and HA (4.0 kDa) were purchased from Meilun Biology, China. HAase (from sheep testicular tissues) was purchased from Yuanye Biology, China. Ammonia water and hydrochloric acid (HCl) were obtained from Jiangtian Chemical Engineering, China. Standard iron solutions were from Luoen, China. Hydrogen peroxide solution (H₂O₂, 30% v/v) was obtained from Honeywell, USA. A hydroxyphenyl fluorescein (HPF) probe was purchased from Cayman, USA. Murine macrophage cell line J774.A1 was obtained from Tianjin Dingguo Biotechnology Co., Ltd. All reagents used in this study were of analytical grade or better.

2.2 Synthesis of HA-coated Fe₃O₄ NPs

NPs were synthesized according to Huige's method with some modifications⁷ (Fig. 1). Briefly, 1.0 g or 0.4 g of HA was used to synthesize NPs with a high or low surface coating, which were termed Fe₃O₄–HA_H and Fe₃O₄–HA_L, respectively. HA was dissolved in 100 mL of water, and then nitrogen was bubbled through the solution for 30 min. The HA solution was boiled at 100 °C in an oil bath, followed by the addition of iron solution, which was prepared by dissolving FeCl₃·6H₂O (0.29 g, 0.54 mmol L⁻¹) and FeCl₂·4H₂O (0.14 g, 0.36 mmol L⁻¹) in dilute HCl solution (4 mL, 1.0 mol L⁻¹). The process was performed in a nitrogen atmosphere with vigorous stirring. Ammonia solution (30 mL, 28% w/w) was added dropwise to adjust the pH to 9–10. Then, the mixture was refluxed for 40 min, cooled to room temperature and dialyzed for 72 h in a Spectra/Por dialysis bag (with a molecular weight cut off of 12–14 kDa) against deionized water. A total of eight exchanges were performed to remove unreacted reagents and soluble by-products. Finally, the substance in the dialysis bag (pH 7.0) was collected via filtration through a nylon membrane filter (with a diameter of 25 mm and a pore size of 0.22 μm) and then lyophilized for further use.

Fig. 1 Schematic of the synthesis of various NPs.

To obtain the NPs after ligand biodegradation (termed Fe₃O₄–HA_R), 4 g L⁻¹ HAase (300–500 U mg⁻¹) and 200 mg Fe L⁻¹ Fe₃O₄–HA_H were dissolved in 45 mL of pH 4.5 phosphate-buffered saline (PBS), and the mixture was shaken at 37 °C for 6 h. After that, the mixture was centrifuged, washed with 30 mL of ethanol three times, and then lyophilized for further use. If not indicated, the iron concentration (which was determined by inductively coupled plasma mass spectrometry (ICP-MS, Elan drc-e, PerkinElmer, USA)) instead of the overall NP concentration was used throughout the study. Particularly, all subsequent experiments involving the NPs were performed in the dark to exclude light interference.

2.3 Characterization of the various NPs

The morphologies of the NPs were visualized with transmission electron microscopy (TEM) at an accelerating voltage of 200 kV (JEM-2800F, Hitachi, Japan). The X-ray diffraction (XRD) patterns were obtained using a D/Max-2500 diffractometer (Rigaku, Tokyo, Japan) working at 40 kV and 40 mA with a CuKα radiation source. Thermogravimetric analysis (TGA) was performed at a heating rate of 10 °C min⁻¹ in a nitrogen flux with simultaneous thermogravimetric/differential thermal analyzers (TGA/DSC1, Mettler Toledo, Switzerland). The Fourier transform infrared (FTIR) transmission spectra of the materials were obtained using a Bruker TENSOR 27 apparatus (Bruker Optics Inc., Karlsruhe, Germany). The particle size distributions of the NP suspensions and ζ potentials were measured by dynamic light scattering (DLS) using a ZetaPALS (Brookhaven Instruments, Holtsville, U.S.A.). Sonication (at 40 kHz for 10 min) was performed to ensure adequate dispersion before DLS and zeta potential measurements.

2.4 Measurements of the cytotoxicity and intracellular contents of hydroxyl radicals (˙OH) and iron

J774.A1 cells were incubated with 20 mg Fe L⁻¹ Fe₃O₄–HA_H, Fe₃O₄–HA_L and Fe₃O₄–HA_R in Dulbecco's modified Eagle medium (DMEM, pH 7.4) for 12 h at 37 °C. Images of ROS generation were acquired on a confocal microscope (LSM 880, Jena, Germany) using the fluorescence probe 2′,7′-dichlorofluorescin diacetate (DCFH-DA, KeyGEN BioTECH, Nanjing, China). The intracellular ROS content was quantified using an ROS detection kit (KeyGEN BioTECH, Nanjing, China). The total contents of glutathione (GSH) and malondialdehyde (MDA) were measured using a total glutathione assay kit (Beyotime, Shanghai, China) and a micro-MDA assay reagent kit (KeyGEN BioTECH, Nanjing, China), respectively. Cell viability was assessed using a Cell Counting Kit-8 (CCK-8, Beyotime, Shanghai, China).

To determine the intracellular ˙OH content, cells were incubated with various NPs (20 mg-Fe/L) for 12 h, and after centrifugation (2500 rpm × 5 min), the cells were incubated with 2 μL of 5 mmol L⁻¹ HPF (in 2 mL of DMEM) in the dark at 37 °C for 1 h. The cells were then washed with Hanks' balanced salt solution (HBSS) (pH = 7.4) three times. Finally, the fluorescence of the cells in 2 mL of HBSS was determined using a flow cytometer (Accuri C6 Plus, BD, Singapore) at an excitation wavelength of 488 nm and an emission wavelength of 520 nm.

To measure the iron content (termed Fe_tot), the cells were incubated with various NPs (at 20 mg Fe L⁻¹) for 12 h, collected via centrifugation (2500 rpm × 5 min), washed with PBS three times and then resuspended in 1 mL of PBS solution. The cells were lysed by being frozen and thawed three times. After centrifugation, 1 mL of supernatant was mixed with 1 mL of nitric acid solution and 0.5 mL of H₂O₂ solution, and then the mixture was heated in a digestion device (COOLPEX, PreeKem, China) following a temperature program, which was set at 80 °C, 120 °C, 150 °C, and 180 °C for 3 min each, followed by 200 °C for 25 min. After that, the residue was cooled, transferred to a 10 mL volumetric flask, and filled up with ultrapure water. The Fe_tot, corresponding to both membrane-bound and internalized NPs, was determined by ICP-MS.

2.5 Abiotic ion dissolution, characterization of the NP surface composition and abiotic ˙OH production

Various NPs (200 mg Fe L⁻¹) were suspended in PBS (pH 4.5) for 12 h, and the Fe_tot and ferrous ion (Fe²⁺) concentrations in the supernatant were determined with 1,10-phenanthroline, sodium acetate and hydroxylamine hydrochloride.²⁰ To determine the Fe²⁺ concentration, 0.5 mL of supernatant was mixed with 0.3 mL of 2 g L⁻¹ 1,10-phenanthroline and 0.3 mL of 80 g L⁻¹ sodium acetate, and then the mixture was left in the dark for 30 min. The absorbance at 510 nm was measured using an ultraviolet spectrophotometer (Cary 100, Agilent, Germany). To measure the Fe_tot, 0.3 mL of hydroxylamine hydrochloride (2 g L⁻¹) was added to a 0.5 mL sample, which can reduce Fe³⁺ to Fe²⁺, and then the Fe²⁺ concentration was determined as described above. A calibration curve for an Fe²⁺-phenanthroline complex was also constructed (Fig. S2†), and the detection limit was approximately 0.04 mg L⁻¹.

The elemental composition on the surface of Fe₃O₄–HA_R was determined with X-ray photoelectron spectroscopy (XPS) (PHI 5000 VersaProbe, Tokyo, Japan). Ion-beam etching was also employed to remove the surface layer, and the elemental composition in depth profiling was performed.

To test ˙OH generation, the NP suspensions (20 mg-Fe/L) were mixed with 5 μL of H₂O₂ solution (10 mmol L⁻¹) and 2 μL of HPF solution (5 mmol L⁻¹) for 30 min in the dark. After centrifugation, the fluorescence in the supernatant was recorded using a fluorescence spectrophotometer (F-7100, HITACHI, Japan) with an excitation wavelength of 488 nm and an emission wavelength of 515 nm. Moreover, dimethylpyridine nitrogen oxide (DMPO, 250 mmol L⁻¹) was used as a spin trapping agent and was incubated with 20 mg Fe L⁻¹ NPs for 30 min in the presence of 10 mmol L⁻¹ H₂O₂. The electron paramagnetic resonance (EPR) spectra were measured with a Bruker EMX spectrometer (EMX Plus 6-1, Bruker, Germany).

2.6 Measurement of ion release during ligand degradation

The enzymatic degradation procedure was described previously, and the supernatant was collected at 0, 1, 2, 3, 4, 5 and 6 h. The concentrations of Fe_tot and Fe²⁺ were measured as described before. The supernatants of the Fe₃O₄–HA_H suspension without added enzyme were also collected at the indicated times and subjected to ion measurements.

2.7 Characterization of Fe₂O₃ surface layer formation and abiotic superoxide radical (˙O₂⁻) production

Raman spectra were obtained using a Renishaw inVia Raman spectrometer (RM2000, London, U.K.) with a 633 nm He–Ne laser. The selected area electron diffraction (SAED) images and lattice images were obtained by high-resolution TEM (JEM-2100, JEOL, Japan).

To test ˙O₂⁻ production, Fe₃O₄–HA_L or Fe₃O₄–HA_R (20 mg Fe L⁻¹) was suspended in 2 mL of methanol with DMPO (250 mmol L⁻¹) as a spin trapping molecule and then subjected to EPR spectroscopy. If needed, H₂O₂ solution at a final concentration of 10 mmol L⁻¹ was added to the suspension during EPR analysis.

2.8 Statistical analysis

At least three replicates were run per treatment. All measurements were performed at least in triplicate, and the results are expressed as the mean ± standard deviation (SD). Significant differences were determined by an independent t test or one-way ANOVA. A p value less than 0.05 indicated a significant difference.

3. Results and discussion

3.1 Characterization of the three NPs

The three NPs exhibited similar morphologies, physical sizes and crystalline structures. TEM images show that the three NPs were spherical with physical diameters of approximately 4–6 nm (Fig. 2a–c), which agreed well with previous studies showing that Fe₃O₄ NPs synthesized by the chemical co-precipitation method generally had small diameters of approximately 5–10 nm.²¹ Energy dispersive spectroscopy (EDS) (Fig. S3†) also verified the purity of the various Fe₃O₄ NPs. The XRD patterns revealed that the three NPs exhibited similar crystalline structures with characteristic peaks corresponding to the (220), (311), (400), (333), (440), and (533) facets (Fig. 2d), which corresponded well with the XRD pattern of magnetite (JCPD card no. 74-0748). The average grain size was quantitatively calculated by means of the Scherrer equation based on the XRD profiles, and the sizes were determined to be 5.10 nm, 5.60 nm and 5.20 nm for Fe₃O₄–HA_H, Fe₃O₄–HA_L, and Fe₃O₄–HA_R, respectively, which coincided with the sizes determined by TEM counting. The TGA thermograms showed that more HA ligands were coated onto the F₃O₄–HA_H surface relative to the lower weight loss for F₃O₄–HA_L and F₃O₄–HA_R (Fig. 2e). Similar retention rates were obtained for F₃O₄–HA_L and F₃O₄–HA_R (45.28% and 43.17%, respectively), indicating similar coating contents, whereas their different weight loss curves may point to varied HA structures after enzymatic degradation. The thermogram of HA alone was also provided in Fig. S4.†

Fig. 2 Characterization of various NPs. TEM images of (a) Fe₃O₄–HA_H, (b) Fe₃O₄–HA_L, and (c) Fe₃O₄–HA_R, with insets showing the size distribution and mean size. (d) XRD patterns, (e) TGA thermograms, (f) FTIR spectra, (g) zeta potential and (h) DLS analysis (based on the intensity-weighted distribution) of Fe₃O₄–HA_H, Fe₃O₄–HA_L and Fe₃O₄–HA_R. For the zeta potential and DLS analysis, 20 mg Fe L⁻¹ NPs were suspended in water (pH 7.0) or DMEM (pH 7.4). All measurements were performed in triplicate, and the results are expressed as the mean values ± SDs.

The varied structure of the coating ligand after degradation was supported by the FTIR spectra (Fig. 2f, also see Table S1† for the detailed wavelengths of each peak). Regarding free HA, the peaks at 3393 cm⁻¹, 1616 cm⁻¹, 1407 cm⁻¹, and 1047 cm⁻¹ were assigned to the stretching mode of the O–H bond, C=O stretching band of the amide,²² asymmetric vibration of COO⁻,²² and C–O stretching vibration,²³ respectively, which were consistent with the reported HA structure.⁷ Almost identical ligand structures were observed for Fe₃O₄–HA_H and Fe₃O₄–HA_L, whereas peak shifts were detected for Fe₃O₄–HA_R. For instance, the peak at 1616 cm⁻¹ redshifted to 1647 cm⁻¹, which was associated with the stretching vibration of the C=O⋯H hydrogen bond between ligand fragments. The blueshifts from 1407 cm⁻¹ to 1394 cm⁻¹ and from 1047 cm⁻¹ to 1024 cm⁻¹ could be due to the interactions of iron ions with the O and N atoms of the surface ligand.²⁴ Changes in the ligand functional groups also correlated well with the decreased net negative surface charge, thus inducing agglomeration in water as a result of reduced repulsive interactions (Fig. 2g and h). This trend was more evident when comparing Fe₃O₄–HA_L to Fe₃O₄–HA_R with similar loadings of the coating. Fe₃O₄–HA_H exhibited the highest negative charge and the least aggregation in water, and the differences among the three NPs were alleviated in DMEM due to the presence of opposite ions. Thus, HA degradation concurrently changed the assembly pattern of the ligand and surface coordination chemistry.

The enzymatic degradation of HA has been investigated in a few studies for the purpose of biomedical applications, production of low molecular weight HA, and elucidation of the physiological and pharmacological roles of HA.²⁵ Generally, the degradation kinetics and fragmentation were enzyme-,²⁶ pH-²⁷ and time-dependent,²⁸ and little structural change in the residues was perceived under mild conditions and for short degradation times.²⁹ Herein, the structural change of HA coated on the Fe₃O₄ NPs after enzymatic degradation was associated with the modification of surface coordination with the iron core, which possibly stemmed from interactions with fragments.

3.2 Fe₃O₄–HA_R induced high oxidative stress and cytotoxicity

Generally, iron NPs induce cellular damage via oxidative stress due to the potential initiation of Fenton or Fenton-like reactions. Macrophages serve as a first line of defense against exogenous particles in almost every organ system and have been extensively used as a common model in cytotoxicity studies. The CLSM images (Fig. 3a) show that a higher ROS content was induced in the J774 macrophage cell line by Fe₃O₄–HA_R in comparison with Fe₃O₄–HA_H and Fe₃O₄–HA_L, and further quantification analysis evidenced that the ROS content in the Fe₃O₄–HA_R-treated cells was almost 3 times higher than that in J774 cells alone (Fig. 3b). An approximately 50% increase in ROS induction was observed in the presence of Fe₃O₄–HA_H or Fe₃O₄–HA_L, while no difference was perceived between them. Consistent trends were found in terms of decreasing GSH content, increasing MDA levels, and decreasing cellular viability (Fig. 3c–e). Bearing in mind the possible involvement of the Fenton reaction, the intracellular ˙OH content was measured, and more than 2 times the ˙OH content was detected in the cells treated with Fe₃O₄–HA_R relative to that in the J774 cells alone (Fig. 3f). Slight increases in the ˙OH content were noticed in the cells stressed by Fe₃O₄–HA_L and Fe₃O₄–HA_H, while no significant difference was detected between them. Apparently, different ˙OH radical production cannot be ascribed to diverse cellular uptakes since the highest rate of internalization was noted for Fe₃O₄–HA_H due to the ligand–membrane receptor interactions between HA and CD44³⁰ and good dispersion (Fig. 3g). Comparable rates of cellular uptake were detected for Fe₃O₄–HA_L and Fe₃O₄–HA_R because of similar ligand loadings. Although we cannot fully differentiate the intracellular NPs from surface-bound NPs based on the Fe_tot, it is generally thought that a higher level of adhesion ensures higher uptake efficiency.³¹ The iron content was visualized via Prussia blue staining (Fig. S5†). Collectively, the low cellular uptake, high content of generated ˙OH radicals, and high level of ROS induction suggested that Fe₃O₄–HA_R may possess an excellent capability for producing ˙OH radicals.

Fig. 3 (a) CLSM images showing the ROS levels in the J774 cells treated with 20 mg Fe L⁻¹ Fe₃O₄–HA_H, Fe₃O₄–HA_L or Fe₃O₄–HA_R for 12 h in DMEM (pH 7). Quantification of the (b) ROS level, (c) GSH content, (d) MDA content, (e) cell viability, (f) intracellular ˙OH content and (g) Fe_tot. All experiments were carried out at least in triplicate, and the results are presented as the mean values ± SDs. Different letters indicate a significant difference (p < 0.05).

It was suggested that surface coating influenced NP cytotoxicity via modifications in agglomerate size, cellular uptake, surface roughness, dissolution rate, and surface-related reactivity.³² Smaller NPs facilitate a higher rate of cellular uptake, thus potentially inducing higher toxicity. However, in this study, surface coating-related changes in the NP size and uptake showed negligible effects on cellular viability when comparing Fe₃O₄–HA_H with Fe₃O₄–HA_L. The possible reason was that the HA-coated Fe₃O₄ NPs were recognized to be biocompatible, and cellular uptake was not a determining factor on toxicity at low NP concentrations.⁷ Additionally, no damage to membrane integrity (caused by surface roughness) was perceived upon prompt contact with various NPs (as measured by lactate dehydrogenase release, data not shown). Fe₃O₄–HA_L and Fe₃O₄–HA_R with similar sizes and comparable ligand loadings revealed substantially distinctive radical production abilities. Thus, the capability of generating ˙OH radicals, which are closely related to ion dissolution behavior and surface-related reactivity, may be the foremost factor and will be evaluated under abiotic conditions, as described below.

3.3 Fe₃O₄–HA_R released more Fe²⁺ and produced a large number of ˙OH radicals

The dissolution behaviors of the three NPs are compared in Fig. 4a, and the total amount of ions released by Fe₃O₄–HA_H was higher than the amount released by Fe₃O₄–HA_L and Fe₃O₄–HA_R since Fe₃O₄–HA_H was well dispersed. Generally, a higher dispersibility gives rise to a more exposed surface area, thus leading to a higher rate of dissolution.^7,32 Interestingly, almost 100% of the released ions by Fe₃O₄–HA_R were Fe²⁺ compared to 70% by Fe₃O₄–HA_H and Fe₃O₄–HA_L. The high content of Fe²⁺ that was detected may be ascribed to the higher solubility of Fe²⁺ over Fe³⁺ at this pH. Nevertheless, the markedly high level of Fe²⁺ by Fe₃O₄–HA_R was possible due to the reversible adsorption of Fe²⁺ onto the NP surface, as evidenced by the following analysis. The change in surface stoichiometry was confirmed by XPS spectroscopy. As shown in Fig. 4b, the ratio of Fe(II)/Fe(III) on the Fe₃O₄–HA_R surface changed to 1 : 1.75, while the ratio remained at 1 : 2.01 after surface etching (Fig. 4c), which agreed with the stoichiometry of magnetite NPs.³⁴

Fig. 4 (a) Ion release (Fe_tot) from various NP suspensions and the percentage of Fe²⁺ in Fe_tot. Various NPs (200 mg Fe/L) were suspended in PBS (pH 4.5) at 37 °C for 12 h. XPS spectra of Fe₃O₄–HA_R (b) before and (c) after surface etching. (d) UV-vis spectra showing the ˙OH radical production by various NPs (20 mg Fe L⁻¹) incubated with 10 mmol L⁻¹ H₂O₂ using HPF (5 mmol L⁻¹) as a probe. Water instead of a NP suspension was used as a control. (e) EPR spectra showing ˙OH production from various NP suspensions (20 mg Fe L⁻¹) in the presence of 10 mmol L⁻¹ H₂O₂ using DMPO (250 mmol L⁻¹) as a spin trapping reagent. (f) Contents of released Fe_tot and Fe²⁺ at different degradation times. Fe₃O₄–HA_H (200 mg Fe L⁻¹) was incubated with/without 4 g L⁻¹ HAase in PBS (pH 4.5). All experiments were carried out at least in triplicate, and the results are presented as the mean values ± SDs. Different letters indicate a significant difference (p < 0.05).

A high content of surface-adsorbed Fe²⁺ endowed Fe₃O₄–HA_R with a high ability for producing ˙OH radicals since Fe²⁺ is superior to Fe³⁺ in mediating the Fenton reaction. Compared with Fe₃O₄–HA_H and Fe₃O₄–HA_L, Fe₃O₄–HA_R produced nearly 1.5 times the number of ˙OH radicals when reacted with H₂O₂ (Fig. 4d), and this statement was further verified by the stronger peak of the DMPO–˙OH adducts generated by Fe₃O₄–HA_R in the EPR spectra (Fig. 4e). It is worth mentioning that a significant amount of Fe²⁺ was released only during the enzymatic degradation of the ligands (Fig. 4f). One possible reason was that the ligand fragments (which contain COO-, OH- and C(=O)N- groups) may promote ion dissolution (mostly Fe²⁺ ions) at the interface, and the decrease in the amount of Fe²⁺ after 4 h indicated the adsorption, oxidation, and precipitation of aqueous ions. Similar ligand-promoted dissolution has been reported in some studies.^33,35 For instance, oxalate and ascorbic acid significantly promoted the dissolution of nanoscale hematite, which was suggested because metal–organic surface complexes destabilized the Fe–O bond and lowered the energy barrier for dissolution.³³

The effect of ligand coating on core material dissolution has been suggested in some studies.^36,37 Generally, the degradation of inorganic cores is governed by surface corrosion; thus, the protective efficiency of various coatings relies on their ability to prevent access to core surfaces. Therefore, relatively low dissolution rates were observed for the three NPs when considering high ligand loadings, even for Fe₃O₄–HA_R. However, few studies have considered ligand degradation-induced core dissolution,³⁸ in which ligand fragments with various complexation functional groups can strongly bind to the core surface, leading to Fe atoms being pulled out. Thus, a large number of ions were detected during ligand degradation. In particular, the co-presence of Fe²⁺ and Fe₃O₄ NPs can have a significant impact on the stoichiometry and reactivity of magnetite NPs, and the interplay mechanisms include reductive dissolution, the reversible uptake and release of Fe²⁺, bulk electron conduction, electron transfer, atom exchange, and recrystallization.^35,39,40 To our knowledge, the contribution of these processes to the overall cytotoxicity has been decidedly overlooked. Our results reveal a piece of information by decoding the change in surface chemistry after ligand degradation.

3.4 Fe₂O₃ formation on the surface of Fe₃O₄–HA_R facilitates the Fe(II)/Fe(III) cycle via ˙O₂⁻ generation

A close look at the XPS spectra (Fig. 4b) revealed the presence of the Fe₂O₃ layer at the NP interface. Generally, peaks at 710.7 eV and 724.3 eV were reported for Fe 2p_3/2 and Fe 2p_1/2, respectively.⁴¹ The peaks at 709.8 eV and 711.1 eV were attributed to Fe²⁺ and Fe³⁺ in the octahedral coordination of Fe₃O₄, respectively, while the peak at 712.8 eV was assigned to Fe³⁺ in the tetrahedral coordination of Fe₃O₄.^42,43 Notably, a satellite peak at 717.9 eV was observed for Fe₃O₄–HA_R but not for the etched NPs, and this peak was ascribed to Fe 2p_3/2 for Fe₂O₃.^43,44 The appearance of the Fe₂O₃ surface layer was also confirmed by Raman spectroscopy, HRTEM imaging and SAED (Fig. 5a–c). The peaks at 299 nm, 537 nm and 666 nm in the Raman spectra were the characteristic peaks of Fe₃O₄,⁴⁵ and the appearance of new peaks at 231 nm, 410 nm, 616 nm and 1318 nm after ligand degradation was ascribed to Fe₂O₃ (ref. 46) (Fig. 5a). The lattice fringes along Fe₃O₄ (311) in the HRTEM image⁴⁷ have a spacing of 0.25 nm, whereas a lattice spacing of 0.18 nm corresponds to Fe₂O₃ (024) (Fig. 5b). The corresponding SAED pattern of Fe₃O₄–HA_R (Fig. 5c) exhibited visible diffraction rings corresponding to the (533), (220), and (311) planes of Fe₃O₄ and the (113) and (024) planes of Fe₂O₃.⁴⁸ No such surface oxide formation was detected for Fe₃O₄–HA_H and Fe₃O₄–HA_L (Fig. S6†). Accordingly, Fe₃O₄–HA_R was composed of an Fe₃O₄ core and a thin Fe₂O₃ surface layer with embedded Fe²⁺ ions (Fig. 5d).

Fig. 5 (a) Raman spectra of Fe₃O₄–HA_R showing peaks corresponding to Fe₃O₄ and Fe₂O₃. (b) Lattice fringes and (c) SAED image of Fe₃O₄–HA_R. (d) Proposed structure of Fe₃O₄–HA_R. (e) EPR spectra of the DMPO–OOH adducts in the presence of Fe₃O₄–HA_L or Fe₃O₄–HA_R (20 mg Fe L⁻¹ in methanol) with/without the addition of 10 mmol L⁻¹ H₂O₂. DMPO (250 mmol L⁻¹) was used as a spin trapping molecule.

The ferric oxide layer in conjunction with surface-adsorbed Fe²⁺ can play an important role in accelerating the Fe(III)/Fe(II) cycle via ˙O₂⁻ generation, favoring the Fenton reaction.⁴⁹ The ability to generate ˙O₂⁻ was tested for Fe₃O₄–HA_R and Fe₃O₄–HA_L in the presence and absence of H₂O₂. The EPR spectra showed representative peaks of DMPO–OOH adducts, and a stronger peak intensity was observed for Fe₃O₄–HA_R than for Fe₃O₄–HA_L with the addition of H₂O₂ (Fig. 5e). The peak was still discernible even in the absence of H₂O₂ for Fe₃O₄–HA_R due to one electron transfer to molecular oxygen. It is worth mentioning that ˙O₂⁻ generation may also partially contribute to the observed high level of Fe²⁺ during ligand degradation. The mechanism is illustrated by the following equations (eqn (1)–(4)).

≡Fe³⁺ + H₂O₂ → ˙HO₂ + ≡Fe²⁺ + H⁺ (1)

≡Fe³⁺OH + Fe²⁺ + H₂O ↔ ≡Fe³⁺OFe²⁺OH + 2H⁺ (2)

≡Fe³⁺OFe²⁺OH + O₂ → ≡Fe³⁺OFe³⁺OH + ˙O⁻₂ (3)

≡Fe³⁺ or Fe³⁺_aq + ˙O⁻₂ → ≡Fe²⁺ or Fe²⁺_aq + O₂ (4)

In general, the passivation of Fe₃O₄ NPs during bio-nano interactions decreased surface-related activity and mitigated cellular toxicity, including surface oxidation, the absorption of protein or other biomolecules, and the decrease in ion leakage due to surface complexation with thiol groups from proteins.^50,51 Nevertheless, our results revealed the more toxic effect of Fe₃O₄ after surface oxidation because the coexistence of the Fe₃O₄ core and oxidized Fe₂O₃ surface layer in conjunction with surface-adsorbed Fe²⁺ provided a unique reactive surface for the initiation of the Fenton reaction. In line with our studies, it has been shown that Fe₃O₄@Fe₂O₃ nanocomposites significantly improve the charge transport at the Fe₃O₄/Fe₂O₃ interface, boosting their catalytic activities.⁵² The surface-bound Fe²⁺ ions on the Fe₂O₃ shell benefit single-electron molecular oxygen activation, forming a large quantity of ˙O₂⁻.⁵³ Consequently, in view of the particular interest in Fe₃O₄ NPs in bioscience applications, more attention should be given to how passivation and depassivation occur in vivo and how ligand biodegradation concomitantly affects the rate and extent of these processes.

4. Conclusions

Overall, our results demonstrated that the enzymatic degradation of surface HA induced ligand-promoted ion dissolution, followed by the precipitation of Fe₂O₃. Then, a high level of aqueous Fe²⁺ was adsorbed onto the Fe₃O₂ surface layer, forming a complex Fe₃O₄@Fe₂O₃ structure with active surface-adsorbed Fe²⁺ (Fig. 6). This distinctive structure promoted their high competence for producing reactive ˙OH radicals, thus inducing more severe oxidative damage in macrophage cells. In this study, the enzymatic degradation of surface ligands was performed under relatively simple conditions; nonetheless, the reshaping, recrystallization and generation of secondary NPs would be more complicated if considering simultaneous degradation occurring on all the different architectural parts of the NPs, the coexistence of passivation and depassivation by-products and reactions with solution entities in biological environments (e.g., organic/inorganic ligands, redox molecules, and other nutrient metal ions). Overall, tracing the NP biotransformation both in vivo and in vitro would provide valuable information for understanding the fate and toxicity of engineered NPs, which will in turn, allow for the design of more rational and biocompatible NPs for biomedical applications.

Fig. 6 Diagram of the surface reconstruction for Fe₃O₄–HA_R during enzymatic ligand degradation.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grants 21777077 and 21976018). Partial support was also provided by the Fundamental Research Funds for the Central Universities.

References

1. S. Mourdikoudis, A. Kostopoulou and A. P. LaGrow, Magnetic nanoparticle composites: synergistic effects and applications, Adv. Sci., 2021, 8, 2004951–2005008.
2. J. Dolai, K. Mandal and N. R. Jana, Nanoparticle size effects in biomedical applications, ACS Appl. Nano Mater., 2021, 4, 6471–6496.
3. S. Fusco, H.-W. Huang, K. E. Peyer, C. Peters, M. Haeberli, A. Ulbers, A. Spyrogianni, E. Pellicer, J. Sort, S. E. Pratsinis, B. J. Nelson, M. S. Sakar and S. Pane, Shape-switching microrobots for medical applications: the influence of shape in drug delivery and locomotion, ACS Appl. Mater. Interfaces, 2015, 7, 6803–6811.
4. Y. Li, T. M. Krentz, L. Wang, B. C. Benicewicz and L. S. Schadler, Ligand engineering of polymer nanocomposites: from the simple to the complex, ACS Appl. Mater. Interfaces, 2014, 6, 6005–6021.
5. D. Monego, T. Kister, N. Kirkwood, P. Mulvaney, A. Widmer-Cooper and T. Kraus, Colloidal stability of apolar nanoparticles: role of ligand length, Langmuir, 2018, 34, 12982–12989.
6. S. Alexander, Biopolymers, Wiley-VCH Verlag GmbH & Co. KGaA, Berlin, 2005.
7. H. Zhou, J. Tang, J. Li, W. Li, Y. Liu and C. Chen, In vivo aggregation-induced transition between T₁ and T₂ relaxations of magnetic ultra-small iron oxide nanoparticles in tumor microenvironment, Nanoscale, 2017, 9, 3040–3050.
8. Q. Liang, J. Sherwood, T. Macher, J. M. Wilson, Y. Bao and C. J. Cassady, Citric acid capped iron oxide nanoparticles as an effective MALDI matrix for polymers, J. Am. Soc. Mass Spectrom., 2017, 28, 409–418.
9. H. Zhou, J. Ge, Q. Miao, R. Zhu, L. Wen, J. Zeng and M. Gao, Biodegradable inorganic nanoparticles for cancer theranostics: insights into the degradation behavior, Bioconjugate Chem., 2020, 31, 315–331.
10. M. F. A. Lucas, M. Pavelka, M. E. Alberto and N. Russo, Neutral and acidic hydrolysis reactions of the third generation anticancer drug oxaliplatin, J. Phys. Chem. B, 2009, 113, 831–838.
11. S. Athukorale, M. De Silva, A. LaCour, G. S. Perera, C. U. Pittman, Jr. and D. Zhang, NaHS induces complete nondestructive ligand displacement from aggregated gold nanoparticles, J. Phys. Chem. C, 2018, 122, 2137–2144.
12. W. G. Kreyling, A. M. Abdelmonem, Z. Ali, F. Alves, M. Geiser, N. Haberl, R. Hartmann, S. Hirn, D. Jimenez de Aberasturi, K. Kantner, G. Khadem-Saba, J.-M. Montenegro, J. Rejman, T. Rojo, I. Ruiz de Larramendi, R. Ufartes, A. Wenk and W. J. Parak, In vivo integrity of polymer-coated gold nanoparticles, Nat. Nanotechnol., 2015, 10, 619–625.
13. X. Wang, X. Wang, X. Bai, L. Yan, T. Liu, M. Wang, Y. Song, G. Hu, Z. Gu, Q. Miao and C. Chen, Nanoparticle ligand exchange and its effects at the nanoparticle-cell membrane interface, Nano Lett., 2019, 19, 8–18.
14. R. Mo, T. Jiang, R. DiSanto, W. Tai and Z. Gu, ATP-triggered anticancer drug delivery, Nat. Commun., 2014, 5, 3364–3374.
15. X. Cai, X. Liu, J. Jiang, M. Gao, W. Wang, H. Zheng, S. Xu and R. Li, Molecular mechanisms, characterization methods, and utilities of nanoparticle biotransformation in nanosafety assessments, Small, 2020, 16, 1907663–1907682.
16. Y. Kong, J. Chen, H. Fang, G. Heath, Y. Wo, W. Wang, Y. Li, Y. Guo, S. D. Evans, S. Chen and D. Zhou, Highly fluorescent ribonuclease-a-encapsulated lead sulfide quantum dots for ultrasensitive fluorescence in vivo imaging in the second near-infrared window, Chem. Mater., 2016, 28, 3041–3050.
17. T. Gong, X. Song, L. Yang, T. Chen, T. Zhao, T. Zheng, X. Sun, T. Gong and Z. Zhang, Spontaneously formed porous structure and M₁ polarization effect of Fe₃O₄ nanoparticles for enhanced antitumor therapy, Int. J. Pharm., 2019, 559, 329–340.
18. K. Y. Choi, H. Chung, K. H. Min, H. Y. Yoon, K. Kim, J. H. Park, I. C. Kwon and S. Y. Jeong, Self-assembled hyaluronic acid nanoparticles for active tumor targeting, Biomaterials, 2010, 31, 106–114.
19. E. J. Menzel and C. Farr, Hyaluronidase and its substrate hyaluronan: biochemistry, biological activities and therapeutic uses, Cancer Lett., 1998, 131, 3–11.
20. N. Wang, L. Zhu, M. Lei, Y. She, M. Cao and H. Tang, Ligand-induced drastic enhancement of catalytic activity of nano-BiFeO₃ for oxidative degradation of bisphenol A, ACS Catal., 2011, 1, 1193–1202.
21. Z. Zarnegar and J. Safari, Modified chemical coprecipitation of magnetic magnetite nanoparticles using linear-dendritic copolymers, Green Chem. Lett. Rev., 2017, 10, 235–240.
22. S. S. Suner, B. Ari, F. C. Onder, B. Ozpolat, M. Ay and N. Sahiner, Hyaluronic acid and hyaluronic acid: sucrose nanogels for hydrophobic cancer drug delivery, Int. J. Biol. Macromol., 2019, 126, 1150–1157.
23. P. Mahlumba, P. Kumar, L. C. du Toit, M. S. Poka, P. Ubanako and Y. E. Choonara, Fabrication and characterisation of a photo-responsive, injectable nanosystem for sustained delivery of macromolecules, Int. J. Mol. Sci., 2021, 22, 3359–3380.
24. M. R. Swart, B. C. B. Bezuidenhoudt, C. Marais and E. Erasmus, Spectroscopic characterisation of grubbs 2nd generation catalyst and its p-cresol derivatives, Inorg. Chim. Acta, 2021, 514, 120001–120009.
25. M. Yazdani, A. Shahdadfar, C. J. Jackson and T. P. Utheim, Hyaluronan-based hydrogel scaffolds for limbal stem cell transplantation: a review, Cell, 2019, 8, 245–257.
26. N. Hegyesi, E. Hodosi, P. Polyak, G. Faludi, D. Balogh-Weiser and B. Pukanszky, Controlled degradation of poly-epsilon-caprolactone for resorbable scaffolds, Colloids Surf., B, 2020, 186, 110678–110685.
27. H. Lenormand, F. Amar-Bacoup and J.-C. Vincent, pH effects on the hyaluronan hydrolysis catalysed by hyaluronidase in the presence of proteins. Part III. The electrostatic non-specific hyaluronan-hyaluronidase complex, Carbohydr. Polym., 2011, 86, 1491–1500.
28. P. Yuan, M. Lv, P. Jin, M. Wang, G. Du, J. Chen and Z. Kang, Enzymatic production of specifically distributed hyaluronan oligosaccharides, Carbohydr. Polym., 2015, 129, 194–200.
29. R. Nehme, R. Nasreddine, L. Orlic, C. Lopin-Bon, J. Hamacek and F. Piazza, Kinetic theory of hyaluronan cleavage by bovine testicular hyaluronidase in standard and crowded environments, Biochim. Biophys. Acta, Gen. Subj., 2021, 1865, 129837–129844.
30. A. Almalik, S. Karimi, S. Ouasti, R. Donno, C. Wandrey, P. J. Day and N. Tirelli, Hyaluronic acid (HA) presentation as a tool to modulate and control the receptor-mediated uptake of HA-coated nanoparticles, Biomaterials, 2013, 34, 5369–5380.
31. A. Lesniak, A. Salvati, M. J. Santos-Martinez, M. W. Radomski, K. A. Dawson and C. Aberg, Nanoparticle adhesion to the cell membrane and its effect on nanoparticle uptake efficiency, J. Am. Chem. Soc., 2013, 135, 1438–1444.
32. W. Nie, B. Zhang, X. Yan, L. Su, S. Wang, G. Han and D. Han, Degraded hyaluronic acid-modified magnetic nanoparticles, J. Nanomater., 2020, 2020, 1084890–1084898.
33. C. A. Lanzl, J. Baltrusaitis and D. M. Cwiertny, Dissolution of hematite nanoparticle aggregates: influence of primary particle size, dissolution mechanism, and solution pH, Langmuir, 2012, 28, 15797–15808.
34. A. P. Grosvenor, B. A. Kobe, M. C. Biesinger and N. S. McIntyre, Investigation of multiplet splitting of Fe 2p XPS spectra and bonding in iron compounds, Surf. Interface Anal., 2004, 36, 1564–1574.
35. H. Peng, C. I. Pearce, A. T. N'Diaye, Z. Zhu, J. Ni, K. M. Rosso and J. Liu, Redistribution of electron equivalents between magnetite and aqueous Fe²⁺ induced by a model quinone compound AQDS, Environ. Sci. Technol., 2019, 53, 1863–1873.
36. M. N. Martin, A. J. Allen, R. I. MacCuspie and V. A. Hackley, Dissolution, agglomerate morphology, and stability limits of protein-coated silver nanoparticles, Langmuir, 2014, 30, 11442–11452.
37. O. Hakami, Y. Zhang and C. J. Banks, Influence of aqueous environment on agglomeration and dissolution of thiol-functionalised mesoporous silica-coated magnetite nanoparticles, Environ. Sci. Pollut. Res., 2015, 22, 3257–3264.
38. S. Bae, R. N. Collins, T. D. Waite and K. Hanna, Advances in surface passivation of nanoscale zerovalent iron: A critical review, Environ. Sci. Technol., 2018, 52, 12010–12025.
39. C. A. Gorski, R. M. Handler, B. L. Beard, T. Pasakarnis, C. M. Johnson and M. M. Scherer, Fe atom exchange between aqueous Fe²⁺ and magnetite, Environ. Sci. Technol., 2012, 46, 12399–12407.
40. V. Tishchenko, C. Meile, M. M. Scherer, T. S. Pasakarnis and A. Thompson, Fe²⁺ catalyzed iron atom exchange and re-crystallization in a tropical soil, Geochim. Cosmochim. Acta, 2015, 148, 191–202.
41. T. Yamashita and P. Hayes, Analysis of XPS spectra of Fe²⁺ and Fe³⁺ ions in oxide materials, Appl. Surf. Sci., 2008, 254, 2441–2449.
42. D. Wilson and M. A. Langell, XPS analysis of oleylamine/oleic acid capped Fe₃O₄ nanoparticles as a function of temperature, Appl. Surf. Sci., 2014, 303, 6–13.
43. W. Liu, B. Cheng, T. Miao, J. Xie, L. Liu, J. Si, G. Zhou, H. Qin and J. Hu, Room temperature electric field control of magnetic properties for the alpha-Fe₂O₃/Fe₃O₄ composite structure, J. Magn. Magn. Mater., 2019, 491, 165500–165507.
44. T. Yamashita and P. Hayes, Analysis of XPS spectra of Fe²⁺ and Fe³⁺ ions in oxide materials, Appl. Surf. Sci., 2009, 255, 8194–8194.
45. A. G. Nasibulin, S. Rackauskas, H. Jiang, Y. Tian, P. R. Mudimela, S. D. Shandakov, L. I. Nasibulina, J. Sainio and E. I. Kauppinen, Simple and rapid synthesis of alpha-Fe₂O₃ nanowires under ambient conditions, Nano Res., 2009, 2, 373–379.
46. A. Asfaram, M. Ghaedi, A. Goudarzi and M. Rajabi, Response surface methodology approach for optimization of simultaneous dye and metal ion ultrasound-assisted adsorption onto Mn doped Fe₃O₄-NPs loaded on AC: kinetic and isothermal studies, Dalton Trans., 2015, 44, 14707–14723.
47. Q. Xia, Z. Yao, D. Zhang, D. Li, Z. Zhang and Z. Jiang, Rational synthesis of micronano dendritic ZVI@Fe₃O₄ modified with carbon quantum dots and oxygen vacancies for accelerating Fenton-like oxidation, Sci. Total Environ., 2019, 671, 1056–1065.
48. H. Wang, S. Liu, X. Yang, R. Yuan and Y. Chai, Mixed-phase iron oxide nanocomposites as anode materials for lithium-ion batteries, J. Power Sources, 2015, 276, 170–175.
49. J. Shi, Z. Ai and L. Zhang, Fe@Fe₂O₃ core-shell nanowires enhanced fenton oxidation by accelerating the Fe(III)/Fe(II) cycles, Water Res., 2014, 59, 145–153.
50. S. L. Gawali, S. B. Shelar, J. Gupta, K. C. Barick and P. A. Hassan, Immobilization of protein on Fe₃O₄ nanoparticles for magnetic hyperthermia application, Int. J. Biol. Macromol., 2021, 166, 851–860.
51. A. Bajaj, B. Samanta, H. Yan, D. J. Jerry and V. M. Rotello, Stability, toxicity and differential cellular uptake of protein passivated-Fe₃O₄ nanoparticles, J. Mater. Chem., 2009, 19, 6328–6331.
52. X. Tang, R. Jia, T. Zhai and H. Xia, Hierarchical Fe₃O₄@Fe₂O₃ core-shell nanorod arrays as high-performance anodes for asymmetric supercapacitors, ACS Appl. Mater. Interfaces, 2015, 7, 27518–27525.
53. Z. Ai, Z. Gao, L. Zhang, W. He and J. J. Yin, Core-shell structure dependent reactivity of Fe@Fe₂O₃ nanowires on aerobic degradation of 4-chlorophenol, Environ. Sci. Technol., 2013, 47, 5344–5352.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1en00724f

---