Engineering iron oxide nanoparticles for enhanced radiosensitization

Abstract

Iron oxide nanoparticles (IONPs) have recently demonstrated considerable potential for enhancing the effectiveness of radiotherapy through radiosensitization. In this study, the radiosensitizing performance of commercial IONPs (Sinerem® and Endorem®) was compared with that of synthesized carboxylated IONPs with different core sizes (5–12 nm) in A549 lung carcinoma cells. A comprehensive evaluation of key biological mechanisms, including cellular internalization, reactive oxygen species (ROS) generation, lysosomal degradation and thioredoxin reductase (TrxR) inhibition, was conducted. Carboxylated IONPs, particularly those with a 7 nm core diameter, exhibited superior radiosensitizing effects, associated with enhanced cellular uptake and substantial TrxR inhibition. By contrast, commercial IONPs exhibited poor internalization and minimal radiosensitizing capacity. Interestingly, no detectable increase in basal ROS levels was found across the different formulations under the tested conditions. Instead, the stronger radiosensitizing response observed with carboxylated IONPs was associated with higher TrxR inhibition and redox imbalance. These findings suggest that sustained ROS overproduction induced by IONP exposure alone is unlikely to be the main driver of the observed effect. These results highlight how nanoparticle size, surface chemistry and intracellular degradation are critical parameters to consider in the development of efficient radiosensitizing agents.

Engineered iron oxide nanoparticles internalized by cancer cells and degraded in lysosomes induce thioredoxin reductase inhibition, leading to redox imbalance and enhanced radiation-induced DNA damage, thereby promoting radiosensitization.

Introduction

Due to their nanoscale dimensions, typically ranging from a few to several hundred nanometers, nanoparticles exhibit distinct physicochemical properties that include a high surface-area-to-volume ratio or enhanced reactivity.¹ These features significantly differentiate them from their bulk counterparts and have enabled their integration into diverse fields such as energy, environmental science and electronics.² In the biomedical domain, their capacity for surface functionalization and molecular-level interactions have spurred the development of nanoparticles as versatile tools for a wide array of applications. Their small size allows for efficient circulation through complex biological environments, while their tunable surface facilitates targeted interactions with specific cells or tissues. Consequently, nanoparticles have emerged as promising platforms for drug delivery, diagnostic imaging and therapeutic applications, where precise biodistribution and optimized cellular interactions are key determinants of clinical success.^1,3–5

Initially developed as magnetic resonance imaging (MRI) contrast agents, iron oxide nanoparticles (IONPs) have played a pivotal role in enhancing soft tissue visualization, particularly in oncological imaging, by increasing the contrast between healthy and pathological tissues.⁶ Depending on their hydrodynamic diameter, IONPs can typically be classified into two main categories: superparamagnetic iron oxide nanoparticles (SPIOs; 50–200 nm) and ultrasmall superparamagnetic iron oxide nanoparticles (USPIOs; <50 nm).³ This size-based classification reflects their differing pharmacokinetics and clinical applications. For example, commercial SPIOs such as Ferumoxides (Endorem®/Feridex®) and Ferucarbotran (Resovist®) were predominantly employed for liver imaging, owing to their dextran and carboxydextran coatings.⁴ In contrast, USPIOs like Ferumoxtran (Sinerem®/Combidex®) were developed for lymph node imaging, while Feruglose® found applications in vascular imaging.⁷ Despite their initial promise, many of these IONP-based agents faced limited commercial success and were subsequently withdrawn from the market, as exemplified by the discontinuation of Sinerem® and Endorem®. Currently, Ferumoxsil remains the only FDA-approved IONP-based contrast agent for gastrointestinal imaging. Interestingly, Feraheme® (ferumoxytol), initially explored for imaging, has been repurposed and approved by the FDA for treating iron-deficiency anemia in patients with chronic kidney disease. These outcomes underscore the challenges of translating nanoparticle-based technologies into clinical practice.⁸

While initially introduced as a diagnostic tool, IONPs have since garnered considerable attention for their therapeutic potential, particularly in oncology. Among the most studied applications, magnetic hyperthermia has emerged as a promising strategy.^9,10 When exposed to an alternating magnetic field, IONPs generate localized heat, inducing selective tumor cell death while minimizing damage to surrounding healthy tissues. This thermal effect has shown synergistic benefits when combined with conventional treatments such as radiotherapy and chemotherapy, primarily through the disruption of DNA repair mechanisms and modulation of the tumor microenvironment. In addition to hyperthermia, IONPs have also been shown to enhance radiotherapy efficacy by acting as radiosensitizers. One of the key mechanisms involves the increased production of reactive oxygen species (ROS) upon irradiation, leading to oxidative damage of cellular components such as DNA, proteins and lipids.^11,12 Elevated ROS levels can also impair DNA repair and disrupt redox homeostasis, further sensitizing tumor cells to radiation.¹³ Notably, inhibition of the antioxidant enzyme thioredoxin reductase (TrxR) by surface-functionalized IONPs has been identified as a contributing factor in amplifying oxidative stress. In our previous study, we demonstrated that modifying the nanoparticle surface not only modulates cellular uptake but also intensifies oxidative damage through TrxR inhibition, thereby improving radiosensitization outcomes.¹⁴

Despite growing evidence supporting the therapeutic potential of iron oxide nanoparticles, the radiosensitizing capabilities of clinically approved or commercial IONPs remain relatively underexplored. Yet, leveraging these well-characterized systems, already known for their biocompatibility and low toxicity, could significantly streamline the clinical translation of new theranostic applications that combine imaging and therapy. In this context, repurposing existing IONP formulations for radiotherapy enhancement represents a promising and pragmatic approach.

Building on this rationale, the present study investigates how variations in IONP formulation influence their radiosensitizing performance. Specifically, we assessed the impact of particle size (ranging from 5 to 12 nm) on key biological parameters, including reactive oxygen species (ROS) generation, TrxR inhibition and cellular uptake. To further contextualize our findings, we benchmarked the radiosensitizing efficacy of synthesized IONPs against two commercial formulations, Sinerem^® and Endorem^®, which are well established in terms of their magnetic and biocompatibility profiles. This comparative approach not only provides a clearer understanding of how physicochemical characteristics influence biological responses but also offers a translational pathway toward the development of next-generation nanotherapeutics for cancer treatment.

Results

Nanoparticle synthesis and characterization

IONPs were synthesized by the alkaline co-precipitation of ferrous and ferric chlorides in a polyol medium, followed by stabilization with a silanization process using a specific organosilane which, after hydrolysis exhibits carboxylated functions (specifically, TEPSA). As shown in Fig. S1 and 1A, TEM images reveal spherical iron oxide nanoparticles with increasing core size from image A to C, corresponding to average diameters of 4.9 ± 0.9 nm, 7.0 ± 1.9 nm, and 11.6 ± 4.6 nm, respectively. In contrast, images D and E of the Fig. S1 correspond to the commercial formulations Sinerem^®, and Endorem^®, corresponding to average diameters of 5.0 ± 1.4 nm and 5.2 ± 0.6 nm respectively.¹⁵ Also, Sinerem^® and Endorem^® formulations exhibit poorly defined morphologies compared to the synthesized carboxylated IONPs.

Fig. 1. Physicochemical characterization of iron oxide nanoparticle (IONP) formulations. (A) Size distribution of IONP cores obtained from TEM analysis, expressed as particle counting histograms and fitted with Gaussian distributions. (B) Hydrodynamic diameter distributions of the different IONP formulations measured by dynamic light scattering (DLS) at an iron concentration of 50 µg Fe per mL in water. (C) Magnetization curves of the different IONP formulations measured by vibrating sample magnetometry (VSM). (D) ¹H nuclear magnetic relaxation dispersion (NMRD) profiles of the different IONP suspensions, reflecting their relaxometric behavior over a broad range of magnetic field strengths. The IONP formulations investigated include 5 nm, 7 nm, and 12 nm carboxylated IONPs, as well as the commercial formulations Sinerem® and Endorem®.

The data in Fig. 1B highlights the strong influence of nanoparticle formulation on their hydrodynamic behavior in aqueous media. As expected, the hydrodynamic diameter of carboxylated IONPs increased with core size, ranging from approximately 11 nm, 18 nm to 32 nm for 5 nm, 7 nm, and 12 nm cores, respectively. These particles displayed a relatively narrow size distribution, reflecting good colloidal stability driven by electrostatic repulsion from their negatively charged surfaces. In contrast, the commercial formulations Sinerem® and Endorem® exhibited significantly larger hydrodynamic diameters, consistent with the presence of a dextran coating and the presence of nanoparticle aggregates.

Particle stability

Fig. S2-A illustrates the evolution of the hydrodynamic diameter of the different IONP suspensions over time in MEM and MEM supplemented with FBS. Regardless of the magnetic core size, the carboxylated IONPs exhibited a similar colloidal behavior, with a rapid destabilization observed in MEM alone. In this medium, the nanoparticles showed a marked increase in hydrodynamic diameter over time, indicating significant aggregation. This instability suggests that electrostatic repulsion alone is insufficient to prevent interparticle interactions in the absence of stabilizing agents. In contrast, when FBS was added in the medium, the hydrodynamic diameter remained more stable over time, suggesting that protein adsorption onto the nanoparticle surface provided additional steric stabilization. This trend was also observed in the commercial formulations Sinerem® and Endorem®, although it was more pronounced for Endorem®, which showed a clearer reduction in aggregation rates in MEM + FBS.

Relaxometric and magnetic properties

The relaxivity measurements at 20 MHz (Table 1) revealed clear trends that associated with both nanoparticle size and formulation. For carboxylated IONPs with the same surface coating but different core sizes, an expected increase in both longitudinal and transverse relaxivities was observed as the nanoparticle diameter increases. Accordingly, the r₂/r₁ ratio rose from 1.7 for 5 nm carboxylated IONPs to 3.4 for 12 nm carboxylated IONPs. These observations were in line with the Nuclear Magnetic Relaxation Dispersion (NMRD) profiles of IONP suspensions, which displaying the frequency-dependent relaxation characteristic of superparamagnetic systems (Fig. 1D). At low frequencies, the longitudinal relaxivity (r₁) is strongly size-dependent: the low-field dispersion was more pronounced for the 5 nm particles, reflecting the dominant contribution of fast Néel relaxation due to their smaller magnetic cores. In contrast, the 12 nm IONPs showed a broader dispersion profile, consistent with their higher magnetic moment and slower Néel relaxation dynamics. Fitting of the NMRD curves with classical relaxation models yielded values of the effective magnetic core diameter that corroborated this size dependence, values of this parameter increasing with nanoparticle diameter. Regarding the dextran-coated formulations (Sinerem® and Endorem®), Sinerem® displays a moderate r₂/r₁ ratio compatible with the absence of nanoparticles clusters, while the high r₂/r₁ ratio associated with Endorem® may indicate the presence of nanoparticle clusters stabilized by the organic coating contributing to their distinct relaxometric behavior.

Table 1. Comparison of the longitudinal and transversal relaxivities (measured at 20 MHz, 37 °C), parameters extracted from the theoretical fitting of NMRD profile (NMRD data), magnetization curve (VSM data) and the core size diameter (obtained by TEM).

Sample	Relaxivities			NMRD data		VSM data	TEM
	r 1 (s−1 mM−1)	r 2 (s−1 mM−1)	r 2/r1	M sat (A m2 kg−1)	D NMRD	M sat (A m2 kg−1)	D TEM
5 nm carboxylated IONPs	16.6	28.2	1.7	65	7.8	56.4	4.9
7 nm carboxylated IONPs	27.4	67.8	2.5	49.6	11.8	58.4	7
12 nm carboxylated IONPs	38.5	134.1	3.5	52.1	13.8	79.8	11.6
Sinerem®	23.1	67.4	2.9	45.2	12.72	50.2
Endorem®	21.6	122.2	5.6	43.2	12.62	50.5

The magnetization curves (Fig. 1C) displayed typical superparamagnetic behavior for all samples, with no hysteresis observed at room temperature, confirming the absence of remanence and coercivity. As shown in Table 1, the saturation magnetization (M_sat) significantly differed depending on the nanoparticle formulation. The carboxylated IONPs exhibited size-dependent saturation magnetization (M_sat) values increasing from 56.4 to 79.8 A m² kg⁻¹ as the core size increased. In contrast, the commercial formulations Sinerem^® and Endorem^® displayed lower M_sat values, measured at 50.2 and 50.5 A m² kg⁻¹, respectively. This decrease was consistent with the low crystallinity of the magnetic core contributing to reduced net magnetic moment.

Cell uptake and associated toxicity

Preliminary analysis of the cytotoxic effects of the various IONP formulations on A549 cells was carried out using a MTT assay. The results are summarized in Fig. S3 which shows cell metabolic activity reflecting the number of viable cells percentages under different conditions (A to E). In most cases, no significant change in the number of metabolically active cells was observed, indicating minimal cytotoxicity following incubation with IONPs. A very low decrease in metabolically active cell number was only detected for the 12 nm carboxylated IONPs at concentrations of 50 and 100 µg Fe per mL. However, this reduction remained below 20%, which was under the toxicity threshold defined by ISO 10993-5 standards. These results indicated that the tested nanoparticles exhibited negligible toxicity toward A549 cells, supporting the biocompatibility of the formulations.

IONP internalization by A549 cells was systematically evaluated for all formulations at different time points (6 h, 24 h, and 48 h) using the Perl's Prussian Blue colorimetric assay to quantify the iron uptake per cell after incubation with a fixed iron concentration of 50 µg of Fe per mL. This concentration was chosen because the MTT assay indicated that it did not reduce the number of metabolically active cells for any of the formulations while providing sufficient sensitivity to detect differences in cellular uptake between formulations. The results (Fig. 2A) demonstrated distinct uptake patterns based on nanoparticle size, surface composition and incubation time. The iron uptake was relatively low at 6 h for all formulations, while a significant increase was observed at later time points for all IONPs as shown in Fig. 2A, except for Sinerem^®. Among the different types of IONPs, the 7 nm carboxylated IONPs showed the highest uptake, reaching a peak at 4.6 ± 1.2 pg Fe per cell after 24 h. However, a decrease in internalization was noted at 48 h across most samples. These results underscored the influence of nanoparticle size and surface chemistry on the kinetics of cellular uptake, with carboxylated IONPs generally exhibiting higher uptake than commercial formulations.

Fig. 2. Comparative evaluation of cellular uptake, oxidative stress response, and thioredoxin reductase inhibition induced by distinct IONP formulations in A549 cells. (A) Iron concentrations in A549 cells assessed after 6 h, 24 h and 48 h of incubation with 50 µg of Fe per mL for different suspensions of IONPs. Cellular iron content was quantified by Perl Prussian blue staining method. Iron concentrations are expressed as mean values ± S.D. for three independent experiments (Tukey test, ***p < 0.001, **p < 0.01, *p < 0.05, ns = not significant p > 0.05). (B) Relative reactive oxygen species (ROS) production in A549 cells after incubation with different IONP formulations. A549 cells were treated for 6, 24, and 48 hours with 50 µg Fe per mL. Intracellular ROS levels are expressed as a percentage relative to untreated control cells. Data are presented as mean ± SD from three independent experiments. Statistical analysis was performed using a two-way ANOVA followed by Dunnett's multiple comparisons test; no significant differences were observed compared to the control group. (C) TrxR inhibition calculated from the slope ratio corresponding to TrxR activity curves extracted from absorbance measurements at 412 nm during 10 min for A549 cells treated with and without IONPs. Data are plotted as mean ± SD of 3 independent experiments. One-way ANOVA analysis was performed for each result (Dunnett's multiple comparisons test, ***p < 0.001, ns = not significant). The IONP formulations presented include 5 nm carboxylated IONPs; 7 nm carboxylated IONPs; 12 nm carboxylated IONPs; Sinerem^® and Endorem^®.

ROS generation upon IONP treatment

Fig. 2B represents the relative ROS production in A549 cells after 6, 24 and 48 h of incubation with different types of IONPs at the concentration of 50 µg Fe per mL. Relative ROS levels were calculated in comparison to the untreated control cells, which represents the baseline ROS production without nanoparticle exposure. As shown in Fig. 2B, no significant changes in ROS production were detected in A549 cells following exposure to either the carboxylated IONPs or the commercial formulations.

Degradation in artificial lysosomal fluids (ALF)

The degradation profiles presented in Fig. S2-B reflect the combined influence of nanoparticle core size and surface coatings on the stability of these particles in simulated lysosomal acidic conditions. Nanoparticles with smaller core sizes, such as the 5 nm carboxylated formulation, exhibited significantly faster degradation rates (0.1780 h⁻¹) than larger counterparts with similar surface chemistry (12 nm: 0.0399 h⁻¹). This behavior can be attributed to the higher surface-to-volume ratio of smaller nanoparticles, which increased their exposure to the acidic environment and accelerated dissolution. In contrast, compared to the 5 nm carboxylated nanoparticles, the 7 nm and 12 nm formulations required approximately 1.5 and 6.5 fold longer, respectively, to reach comparable levels of degradation after 24 hours of incubation, further supporting the inverse correlation between core size and degradation kinetics.

Beside core size, surface coatings played a crucial role in modulating nanoparticle stability, as previously shown. Nanoparticles with organic protective layers, particularly dextran-coated formulations, exhibited markedly slower degradation rates, approximately 0.045 and 0.039 h⁻¹ for Sinerem® and Endorem®, respectively. This behavior likely reflects the effect of the thicker dextran coating, which provided a more substantial barrier against the acidic lysosomal environment and may have trapped released ions, thereby slowing core dissolution.

TrxR activity upon IONP treatment

Thioredoxin reductase (TrxR) activity was monitored at 50 µg of Fe per mL at different time points (6, 24 and 48 h). Over this period, TrxR inhibition varied for all IONPs (Fig. 2C). In particular, 5 nm and 7 nm carboxylated IONPs exhibited a significant increase in TrxR inhibition, with residual activity reaching approximately 45% at 48 h in both cases. In contrast, 12 nm carboxylated IONPs showed a weaker inhibitory effect, reaching a maximal inhibition of only ∼71% at 24 h, which then slightly diminished by 48 h, indicating that these larger nanoparticles are less effective TrxR inhibitors compared to the smaller IONPs. Commercial formulations exhibited weak inhibition effect. Endorem^® showed negligible inhibition of TrxR activity at all time points, indicating a minimal interaction with this enzyme, whereas Sinerem^® induced a slight but detectable inhibitory effect at 24 h, although this effect was not sustained at 48 h.

X-ray irradiation

The data in Table 2 summarize the amplification factors (AF) at 2 Gy after a pre-incubation period ranging from 6, 24, to 48 h with various IONPs, providing insight into how the various IONPs and incubation times influence their ability to enhance radiation-induced cell damage.

Table 2. Amplification factor for A549 cells pre-incubated for 6, 24 or 48 h with 50 µg of Fe per mL of IONPs before being irradiated with 2 Gy X-rays. Results are expressed as mean values from three independent experiments, each performed in triplicate. Values marked with an asterisk (*) were calculated from survival fractions that did not show a statistically significant difference compared to the corresponding control and are therefore provided for indicative purposes (Fig. S4).

Incubation time (h)	5 nm carboxylated IONPs	7 nm carboxylated IONPs	12 nm carboxylated IONPs	Endorem®	Sinerem®
6 h	2.5 ± 0.6%*	4.9 ± 0.8%*	6.8 ± 2.5%*	1.7 ± 0.4%*	1.7 ± 0.6%*
24 h	6.5 ± 1.9%*	14.8 ± 4.2%	12.2 ± 2.8%	4.7 ± 1.5%*	2.0 ± 0.6%*
48 h	10.8 ± 2.9%	16.8 ± 1.8%	15.0 ± 3.2%	6.0 ± 1.8%*	6.6 ± 1.8%*

To facilitate interpretation, the amplification factor (AF) was calculated at a fixed radiation dose of 2 Gy and defined as the relative decrease in surviving fraction induced by nanoparticle pre-treatment compared with irradiation alone. Positive AF values therefore indicate radiosensitization, whereas values close to zero indicate no measurable enhancement of radiation-induced clonogenic cell death. After 6 h of incubation, the AF values were relatively low and similar across all IONP formulations. At 24 h, a pronounced enhancement in AF was observed for most IONPs, to a higher extent for the 7 nm and 12 nm IONPs, with AF values around respectively 14.8 and 12.2. By contrast, the AF values obtained for Endorem^® and Sinerem^® remained low and were generally based on differences in surviving fraction that did not reach statistical significance.

At 48 h, the AF for most of the nanoparticles remained high, with 7 nm carboxylated IONPs exhibiting the highest AF (16.8 ± 1.8%). In addition, 5 nm carboxylated IONPs showed a substantial increase from 24 to 48 h, reaching 10.8 ± 2.9%, while Endorem^® and Sinerem^®'s AF showed a slight increase to respectively 6.0% and 6.6%. This suggests that incubation time played a significant role in determining nanoparticle efficiency, but not all nanoparticles benefited equally from extended exposure. In addition, the surface composition exerts a stronger influence on the amplification factor than the IONP core size diameter, which still contributes but in a less pronounced manner.

To assess whether the radiosensitizing effect was restricted to A549 cells, additional clonogenic assays were performed in H460 cells, another non-small cell lung cancer model. Cells were exposed to the most effective formulation identified in A549 cells, namely 7 nm carboxylated IONPs for 48 h before irradiation. Under these conditions, a radiosensitizing effect was also observed in H460 cells, with an AF of 6.1 ± 2.4%. However, this effect was lower than that measured in A549 cells. These results indicate that 7 nm carboxylated IONPs can enhance radiation responses in more than one lung cancer cell model, while also highlighting the cell-line-dependent nature of the radiosensitizing effect.

γH2AX immunofluorescence labeling

γH2AX foci formation was monitored before irradiation and 24 hours after X-ray exposure in A549 cells treated or not with 7 nm carboxylated IONPs, which showed the highest radiosensitizing potential in clonogenic assays. After 24 h of nanoparticle exposure, the amount of foci were comparable between control and IONPs-treated cells before irradiation, and no significant difference in residual foci was observed 24 h after irradiation (Fig. 3 and S6). In contrast, after 48 h of exposure to 7 nm carboxylated IONPs, the amount of foci before irradiation irradiation remained similar, but a modest increase in residual γH2AX foci was detected 24 h after irradiation (Fig. 3 and S6). These results indicate that 7 nm carboxylated IONPs did not strongly modify the early γH2AX response following irradiation, but prolonged exposure to IONPs (i.e. 48 hours of exposure) may promote a moderate persistence of DNA damage and/or delayed repair. This observation is consistent with the clonogenic data, where the strongest radiosensitizing effect was observed after 48 h of nanoparticle exposure.

Fig. 3. Quantification of the intranuclear γH2AX foci in A549 cells 24 hours after X-ray irradiation. (A) Residual γH2AX foci (i.e. foci that still observable 24 hours after X-ray irradiation) in A549 cells exposed or not to 7 nm carboxylated IONPs during 24 h before X-ray irradiation. Each dot represents the amount of γH2AX foci detected in an individual A549 cell. The data are expressed as mean ± SD (unpaired T-test, ns = not significant, *p > 0.05). (B) Residual γH2AX foci in A549 cells r exposed or not to 7 nm carboxylated IONPs during 48 h. Each dot represents the amount of γH2AX foci detected in an individual A549 cell. The data are expressed as mean ± SD (unpaired T-test, **p < 0.01).

Internalization pathway

To elucidate the cellular uptake mechanisms of IONPs, we focused on the 7 nm carboxylated nanoparticles, identified as the most effective radiosensitizer. Their internalization in A549 cells was investigated by examining the roles of clathrin-mediated endocytosis and of macropinocytosis using complementary genetic silencing and pharmacological inhibition approaches.

Silencing of the clathrin heavy chain (CHC, 180 kDa) or the AP2 µ2 subunit (50 kDa) was performed using siRNA transfection. Western blot analysis confirmed an efficient reduction of the abundance of both proteins compared with cells transfected with a non-targeting control siRNA (Fig. 4A). While the KD of these two proteins was confirmed, confocal quantification of intracellular fluorescence intensity following nanoparticle exposure revealed no significant difference in uptake upon the inhibition of the clathrin-dependent endocytosis (Fig. 4C–E). The scatter dot plots, in which each dot corresponds to the fluorescence intensity of a single cell, consistently showed overlapping distributions (Fig. 4B). Statistical analysis with the Mann–Whitney test confirmed the absence of significant changes. These findings indicate that clathrin-mediated endocytosis does not substantially contribute to IONP uptake.

Fig. 4. Evaluation of endocytic pathway inhibition on nanoparticle internalization in A549 cells. (A) Efficiency of clathrin-mediated endocytosis inhibition was confirmed by Western blot analysis following siRNA-mediated knockdown of clathrin heavy chain (CHC, 180 kDa) and AP2 µ2 subunit (50 kDa). (B) Quantification of intracellular nanoparticle-associated fluorescence intensity by confocal microscopy comparing control cells (175 cells) with AP2 (176 cells) and CHC (163 cells) knockdown conditions. Each dot represents the fluorescence intensity of a single cell; horizontal bars indicate mean values. (C–E) Representative confocal microscopy images of A549 cells incubated with nanoparticles under control conditions (C), following AP2 inhibition (D), or CHC inhibition (E). Nuclei are stained in blue (DAPI) and nanoparticles are shown in red. (F) Quantification of intracellular fluorescence intensity in control cells (304 cells) compared to cells pretreated with amiloride (312 cells) to inhibit macropinocytosis, represented as scatter dot plots with mean values. (G–H) Representative confocal microscopy images of A549 cells incubated with nanoparticles under control conditions (G) and cells pretreated with amiloride (H), nanoparticles are shown in red. As fluorescence intensity values did not follow a normal distribution, statistical analyses were performed using the non-parametric Mann–Whitney test. Ns: not significant; ***p < 0.001. Scale bar (confocal microscopy images): 20 µm.

The role of macropinocytosis was investigated using amiloride (25 µM), a pharmacological inhibitor of the Na⁺/H⁺ exchanger. Pretreatment of A549 cells with amiloride followed by incubation with IONPs/amiloride mixture resulted in a marked decrease in intracellular fluorescence intensity compared with untreated cells (Fig. 4G and H). Quantification of more than 300 cells per condition revealed an average reduction in uptake of approximately 40% (Fig. 4F). This inhibitory effect was statistically significant. These results strongly support a major contribution of macropinocytosis to nanoparticle internalization.

Discussion

Iron oxide nanoparticles (IONPs) have long been investigated for biomedical applications, particularly as MRI contrast agents. Clinically validated formulations such as Endorem^® and Sinerem^®, dextran-coated IONPs, have demonstrated excellent biocompatibility and safety. However, their potential as radiosensitizers has remained largely unexplored. In the present study, we directly compared these commercial formulations with home-made carboxylated IONPs of various core sizes (5–12 nm) and highlighted the critical influence of nanoparticle size and surface chemistry on radiosensitizing activity.

Consistent with previous reports,¹⁴ none of the formulations significantly affected the viability of A549 cells, irrespective of the incubation time or the coating composition. Nonetheless, important differences in uptake behavior were observed, particularly when comparing carboxylated formulations with dextran-coated ones. These differences reflect their distinct surface properties and colloidal stability in biological media.^16,17 Although comparable amounts of iron were quantified after 6 h incubation, uptake profiles diverged significantly over longer periods, confirming that nanoparticle surface chemistry dictates intracellular trafficking and long-term fate, thereby influencing radiosensitization potential. A key factor may be the formation of protein coronas: negatively or positively charged particles, such as carboxylated IONPs, preferentially adsorb serum proteins,¹⁸ which can influence particle–cell interactions and subsequent internalization. Indeed, carboxylated IONPs displayed increased stability in FBS-containing medium (versus rapid sedimentation in serum-free conditions), supporting the formation of a protective protein layer. In contrast, such stabilization appeared less pronounced, especially for Sinerem^®.

Interestingly, several formulations exhibited measurable radiosensitizing effects, but carboxylated IONPs clearly outperformed the commercial dextran-coated counterparts under the experimental conditions tested. Among them, 7 nm TEPSA-coated nanoparticles reached the highest amplification factor (AF), highlighting the importance of both nanoparticle size and surface chemistry in therapeutic efficacy. The lower AF observed in H460 cells further suggests that the radiosensitizing response is cell-line dependent, possibly reflecting differences in nanoparticle uptake, intracellular processing, redox homeostasis and/or intrinsic radiation sensitivity.

If radiosensitization were mainly driven by ROS overproduction through iron-mediated Fenton chemistry, formulations with distinct lysosomal degradation kinetics would be expected to display markedly different AF values. However, no clear relationship between degradation rate and AF was observed among TEPSA-coated formulations of different core sizes. In line with this, none of the tested IONPs significantly increased detectable basal intracellular ROS levels under non-irradiated conditions, and no correlation was found between this basal ROS signal and radiosensitization efficiency. This contrasts with several reports attributing IONP-mediated radiosensitization to ROS-driven mechanisms, often involving lysosomal processing, iron release, and catalytic ROS production.^11,19–21 Such discrepancies may stem from differences in nanoparticle coatings, degradation kinetics, irradiation conditions, or cell models across studies, underlining the complexity of IONP-cell interactions.

To further investigate whether the enhanced radiosensitizing effect was associated with altered radiation-induced DNA damage or repair, γH2AX immunofluorescence labeling was performed after exposure to the most effective formulation, namely 7 nm carboxylated IONPs. These experiments did not reveal any major alteration of the γH2AX foci presence before irradiation after 24 and 48 hours of exposure to the 7 nm carboxylated IONPs.

However, a modest increase in the number of residual γH2AX foci observed 24 hours after the irradiation of cells exposed during 48 hours to the 7 nm carboxylated IONPs suggests that prolonged exposure may contribute to persistent radiation-induced DNA damage and/or delayed repair. These findings are consistent with a mechanism in which prolonged nanoparticle exposure may weaken cellular stress-response capacity, potentially through altered redox homeostasis, without requiring a strong increase in detectable basal ROS levels.

Overall, our data indicate that radiosensitization is broadly associated with nanoparticle uptake, particularly after prolonged incubation. However, this relationship is not absolute, as illustrated by Endorem^® and 5 nm carboxylated IONPs: despite comparable intracellular iron contents after 48 h, their AF values substantially differed. This highlights that uptake alone is insufficient to explain radiosensitization. Notably, higher AF values were generally associated with longer incubation times, suggesting that sustained intracellular exposure and processing are required to maximize the effect (Fig. 5A).

Fig. 5. Correlative analysis of intracellular iron levels, TrxR activity, and radiosensitization in A549 cells. (A) Correlation between intracellular iron concentration (pg Fe per cell) and amplification factor at 2 Gy, categorized by incubation time for each considered IONPs formulations. Data represents SD from three independent experiments. (B) Amplification factor at 2 Gy obtained in A549 cells as a function of residual TrxR activity for each IONPs formulations at different incubation time. Data are plotted as mean ± SD from 3 independent experiments.

Since basal ROS did not fully account for the observed radiosensitizing effects (while acknowledging that transient or localized ROS production after irradiation cannot be excluded), TrxR inhibition was investigated as a complementary redox-related mechanism. All formulations induced partial inhibition of TrxR activity, although to different extents. Carboxylated IONPs, particularly the 7 nm TEPSA-coated particles, exerted the strongest effect, reducing residual activity to 54% after 48 h, whereas commercial Endorem® and Sinerem® only modestly affected TrxR, with residual activity remaining close to 80%. These trends paralleled AF values, as the formulations inducing the strongest and most sustained TrxR inhibition also produced the most pronounced radiosensitization. Moreover, the enhanced efficacy observed after prolonged incubation was associated with both lower residual TrxR activity and higher AF values, suggesting that sufficient intracellular processing time is required to establish a measurable redox-related effect. This interpretation is consistent with the central role of TrxR in thiol redox homeostasis, where its inhibition may reduce the ability of cancer cells to buffer radiation-induced oxidative stress. However, because the correlations between AF and nanoparticle uptake, and between AF and TrxR inhibition, displayed similar Pearson coefficients and R² values, neither parameter clearly outperformed the other as a predictor of radiosensitization. These relationships suggest that nanoparticle accumulation and thioredoxin system impairment both contribute to the radiosensitizing effect of IONPs. Mechanistically, this effect is unlikely to arise solely from a nanozyme-like surface activity of intact nanoparticles, although such a contribution cannot be excluded. Rather, the stronger TrxR inhibition observed with faster-degrading formulations supports the involvement of intracellular iron redistribution following lysosomal processing. More precisely, under acidic lysosomal conditions, partial dissolution of the spinel iron oxide core may release both Fe²⁺ and Fe³⁺ species, which can be rapidly oxidized, buffered, or redistributed through intracellular iron-handling pathways. These processes may disturb thiol-dependent redox homeostasis and interfere with redox-sensitive enzymes such as TrxR, whose active site contains a highly reactive selenocysteine residue. Overall, these findings support the contribution of intracellular iron-mediated redox imbalance to IONP-mediated radiosensitization, while further studies will be required to establish a direct causality.

Since the 7 nm carboxylated IONPs emerged as the most effective radiosensitizers, we conducted a preliminary mechanistic investigation to better understand their cellular uptake. Silencing of key components of the clathrin pathway (CHC or AP2) had little effect (Fig. 4), whereas pharmacological inhibition of macropinocytosis reduced uptake by ∼40% (Fig. 4). These results indicate that macropinocytosis is the predominant endocytic route for these nanoparticles, promoting their subsequent lysosomal trafficking. Consistent with this, colocalization analyses with lysosomal markers revealed strong overlap (>85–90%, Fig. S7), confirming that internalized nanoparticles are efficiently routed to lysosomes, where degradation and iron release are most probably taking place, potentially contributing to their biological activity and radiosensitizing effect.

While these results identify key formulation-dependent parameters controlling radiosensitization in vitro, their translational relevance must be considered with caution. The nanoparticle concentration and incubation times used in this study should not be directly extrapolated to in vivo tumor exposure. In vivo, IONP accumulation is governed by pharmacokinetics, mononuclear phagocyte system clearance, vascular permeability, tumor architecture and heterogeneous intratumoral distribution. Moreover, the absence of stealth coatings such as PEG may favor protein corona formation and recognition by phagocytic cells, thereby influencing circulation time, biodistribution and ultimately tumor accumulation. We also mention that optimized administration strategies, including loco-regional or tumor-targeted delivery, may help improve intratumoral nanoparticle accumulation compared with systemic injection alone. The present conditions should therefore be regarded as a controlled comparative framework for evaluating formulation-dependent biological responses, rather than as conditions intended to reproduce clinically achievable intratumoral concentrations Establishing clinically relevant exposure–response relationships will require quantitative biodistribution studies and in vivo measurements of tumor accumulation. In this context, magnetic particle imaging (MPI), which enables direct and sensitive quantification of IONPs, may represent a valuable tool to bridge in vitro radiosensitization data with in vivo dose–effect relationships.

Materials and methods

Materials

Ferric chloride solution (FeCl₃, 45%), ferrous chloride tetrahydrate (FeCl₂·4H₂O, >99%) and sodium hydroxide were purchased from Fluka (Belgium). 3-(Triethoxysilyl)propyl succinic anhydride (TEPSA) was purchased from ABCR (Germany). Citric acid, aprotinin, (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (MTT reagent), calcium chloride dihydrate, sodium chloride, magnesium chloride, sodium citrate dihydrate, diethylene glycol (DEG), dimethyl sulfoxide (DMSO), glycerol, thioredoxin reductase assay kit, sodium lactate, sodium pyruvate, sodium phosphate dibasic heptahydrate, sodium tartrate dihydrate, dimethylformamide (DMF), potassium chloride, acetone, diethyl ether, potassium phosphate dibasic, magnesium sulfate, paraformaldehyde 4%, Fluoromount-G and amiloride (EIPA) were purchased from Merck (Belgium).

Lissamine rhodamine B sulfonyl chloride was purchased from Acros Organics (Geel, Belgium).

Potassium ferrocyanide, iron standard solution 1000 µg mL⁻¹, sodium bicarbonate, sodium sulfate, triton X100, BSA and formaldehyde 37% were purchased from VWR (Belgium).

Trypan blue (4%), minimum essential medium (MEM) GlutaMAX supplement, phosphate-buffered saline, Pierce 660 nm protein assay reagent, fetal bovine serum (FBS), 2,7-dichlorofluorescein diacetate (H₂DCFDA), penicillin–streptomycin (10 000 U mL⁻¹), 4′,6-diamidino-2-phénylindole (DAPI), Lipofectamine RNAiMAX, Opti-MEMn RIPA buffer, protease inhibitors cocktail, PVDF membrane, anti-rabbit Alexa 488 antibody, RPMI 1640 Medium (ATCC modification) and ECCL Western blot substrates were purchased from Thermo Fischer (Belgium).

A549 cells were purchased from American Type Culture Collection (ATCC), (USA).

Human H460 NSCLC cells (RRID: CVCL_0459; received from DKFZ Heidelberg). The cells were authenticated by genetic characteristics determined by PCR-single-locus-technology (Eurofins Genomics, Ebersberg, Germany).

AllStars Negative Control, SI02777355: 5′-TGCCATCGTGTGGAAGATCAA-3′; SI00299873: 5′-AAGGAGAGTCTCAGCCAGTGA-3′; SI00299880: 5′-TAATCC AATTCGAAGACCAAT-3′; SI04152372: 5′-AAGGGCTAACGTCCCAAATAA-3′; SI04190417:5′-CCCTGAGTGGTTAGTCAACTA-3′ were purchased from QIAGEN (France).

Sinerem^® and Endorem^® formulations were provided by Guerbet (Aulnay-sous-Bois, Paris, France).

Normal Goat Serum and Phospho-Histone H2A.X (Ser139) (20E3) Rabbit Monoclonal Antibody were purchased from Cell Signaling Technology (Leiden, The Netherlands).

Iron oxide nanoparticle synthesis (IONPs)

The synthesis of TEPSA-modified IONPs required a two-step co-precipitation/silanization method as previously described.²² The alkaline co-precipitation of Fe²⁺ and Fe³⁺ salts typically yields magnetite (Fe₃O₄) nanoparticles with a spinel structure. However, partial surface oxidation during processing and purification commonly leads to the formation of maghemite (γ-Fe₂O₃), resulting in mixed magnetite/maghemite cores. Magnetic cores were formed, followed by their surface modification to incorporate carboxylic acid functions. The complete experimental procedure is provided in the SI.

Synthesis of rhodamine-tagged IONPs

To obtain rhodamine-labelled IONPs, a small amount of aminated-rhodamine (TFA salt; 1 µmol; 0.8 mg) was added to an aqueous dispersion of carboxylated nanoparticles ([Fe] = 100 mM; 2 mL) in the presence of N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (50 µmol; 10 mg) as a coupling agent. After one night under stirring at pH 7.5, the suspension was purified by membrane filtration (membrane cut-off = 30 kDa), then centrifuged (16 500 g; 40 minutes).

IONP characterization

The nanoparticle size distribution was assessed using Zetasizer nano ZS (Malvern Instruments) with He–Ne Laser (633 nm) in aqueous medium. In addition, stability studies were performed by monitoring the evolution of IONP size distributions over a period of 48 hours after dilution in Minimum Essential Medium (MEM), either in its standard form or supplemented with 10% Fetal Bovine Serum (FBS).

Transmission electron microscopy provided detailed information on nanoparticle morphology and size using a Fei Tecnai 10 microscope (Oregon, USA) at an accelerating voltage of 80 kV. To prepare the samples, a drop of the diluted suspension was put onto a copper-grid (300 mesh) and left to dry at room temperature. Statistical analysis was conducted by iTEM (Germany) on multiple TEM images (from 500 to 600 particles) to determine the mean diameter, standard deviation and polydispersity index (PDI).

The relaxometric properties of the IONPs were analyzed by measuring relaxation times at 20 MHz (37 °C) and by recording nuclear magnetic resonance dispersion (NMRD) profiles. These measurements allowed the study of the evolution of longitudinal proton relaxivity (r₁) as a function of the applied magnetic field (Larmor frequency). Longitudinal (R₁) and transverse (R₂) relaxation rate were performed at 0.47 T using a Minispec mq 20 spin analyzers (Bruker, Germany). These relaxation rates were determined as a function of the iron molar concentration at 0.47 T in order to calculate the r₁ and r₂ relaxivities (defined as the enhancement of the water proton relaxation rate in 1 mmol l⁻¹ solution of contrast agent). The relaxivities were obtained by linear regression as the slope of the observed relaxation rates (R_i^obs) versus the iron concentration, according to the following equation:r_i being the relaxivities and T_i^dia being the proton relaxation times in aqueous solutions without nanoparticles.

The magnetic properties (M_sat) of the samples were characterized at room temperature using an AC magnetometer (AC Hyster, Nanotech Solutions, Spain). This device measures the dynamic magnetic hysteresis loops of the nanoparticle solution (below 10 g l⁻¹ according to the manufacturer specification) subjected to a sinusoidal alternating magnetic field at a fixed frequency (150 kHz).²³ The hysteresis loops were recorded as magnetization (M) versus magnetic field (H). To ensure thermal stability and avoid self-heating of the sample during measurement, an integrated fluidic cooling system was employed.

IONP degradation in artificial lysosomal fluid (ALF)

ALF was prepared following the procedure described by Rabel et al.¹³ This solution is composed of calcium chloride dihydrate (1.15 mmol; 0.128 g); citric acid (108.26 mmol; 20.8 g); glycerol (0.64 mmol; 0.006 g); magnesium chloride (0.62 mmol; 0.06 g); sodium citrate dihydrate (0.26 mmol; 0.077 g; 0.08 g); sodium chloride (54.93 mmol; 3.21 g); sodium phosphate heptahydrate (0.67 mmol; 0.179 g); sodium lactate (0.71 mmol; 0.08 g); sodium hydroxide (150 mmol; 6 g); sodium pyruvate (0.78 mmol; 0.086 g); sodium sulfate (0.27 mmol; 0.039 g); sodium tartrate dihydrate (0.39 mmol; 0.09 g), and formaldehyde (0.304 mmol; 2.703 mL), for a final total volume of 1 liter completed with deionized water. The solution is adjusted to pH 4.5.

For degradation studies, IONPs were diluted in ALF medium to reach a final [Fe] of 5 mM. Samples were then incubated at 37 °C under continuous stirring. At specific time intervals, aliquots were withdrawn and centrifuged at 7250 g for 15 minutes using ultracentrifugation cells. The filtrate obtained after centrifugation was collected and the concentration of iron released due to IONP degradation was quantified. The iron concentration was determined using the Prussian blue colorimetric method,²⁴ calculated from a calibration curve created with a known concentration of iron standard (iron standard 1000 µg mL⁻¹, diluted in Milli-Q water). The mineralized solution (100 µL) was mixed with 100 µL of 5 N HCl solution and 100 µL of 5% potassium ferrocyanide solution and the absorbance was measured at 650 nm using a microplate reader (SpectraMax).

The degradation kinetic process was described by Gompertz fitting defined as follows:where y_m describes the maximum percentage of IONP degradation. y₀ is the initial percentage of IONP degradation, K is the degradation rate (h⁻¹) of IONPs and x is the time (h).

Cell culture

Human lung adenocarcinoma cells (A549) were grown in Glutamax™ supplemented Eagle's Minimum Essential Medium with 10% of fetal bovine serum and 1% of Pen–strep under a humid atmosphere with 5% CO₂ at 37 °C. Human lung carcinoma cells (H460) were grown in RPMI supplemented with 10% of fetal bovine serum and 1% of Pen–strep under a humid atmosphere with 5% CO₂ at 37 °C.

Cellular viability

The metabolic activity of cells exposed to nanoparticles was measured using the MTT method by 3-[4,5-dimethylthiazol-2-yl]-3,5 diphenyl tetrazolium bromide. A549 cells were cultured in 96-well plates at 5 × 10³ cells per well and treated with varying concentrations of IONPs (0; 10; 25; 50; 100; 150; 200 µg of Fe per mL of IONPs) for 6, 24 or 48 hours. After incubation, 100 µL of the MTT solution (500 µg mL⁻¹ in PBS) were added to the wells. After 3 hours of incubation, the formed formazan crystals were dissolved and the absorbance was measured at 570 nm using a microplate reader.

Iron dosing method by Perls' Prussian blue reaction

A549 cells were plated at 3 × 10⁵ cells per well in 6-well plates and incubated overnight at 37 °C with 5% CO₂. The day after, the culture medium was replaced with MEM + 10% FBS with or without 50 µg Fe per mL of IONPs. After a given incubation time (6, 24 or 48 h), cells were washed three times with PBS, detached using trypsin and collected by centrifugation (1000 rpm, 5 min, 25 °C). Cell numbers were determined and cell pellets digested in 5 M HCl for 72 hours at 37 °C. The resulting solution was then mixed with 5% solution of potassium ferrocyanide in water and the absorbance was measured at 650 nm using a microplate reader. The amount of iron was quantified as pg Fe per cell based on a calibration curve generated from a known iron standard concentration.

Evaluation of intracellular reactive oxygen species (ROS) production

The evaluation of intracellular ROS production in response to nanoparticle exposure was evaluated with the 2,7-dichlorofluorescein diacetate (H₂DCFDA) probe. A549 cells were cultured in 96-well opaque walled plates at 5 × 10³ cells per well with six wells per formulation and treated with IONPs (50 µg of Fe per mL of IONPs) or not (control group) for 6, 24 or 48 hours. Following incubation, the culture medium was removed, and cells were washed twice with HBSS. Half of the wells were then filled with 100 µL of a 15 µM H₂DCFDA solution in Hank's Balanced Salt Solution (HBSS), while the remaining wells were filled with 100 µL of HBSS solution alone. After 1 hour of incubation, fluorescence was measured using a SpectraMax M2 microplate reader (Molecular Devices) with excitation at 485 nm and emission at 535 nm.

HBSS buffer was prepared by dissolving calcium chloride dihydrate (1.67 mmol; 0.185 g), magnesium sulfate (0.81 mmol; 0.0977 g), potassium chloride (5.37 mmol; 0.4 g), potassium phosphate dibasic (0.34 mmol; 0.06 g), sodium bicarbonate (0.71 mmol; 0.35 g), sodium chloride (136.86 mmol; 8 g), d-glucose (17 mmol; 1 g) and anhydrous sodium phosphate (0.34 mmol; 0.0478 g) in demineralized water to a final volume of 1 liter.

Thioredoxin reductase (TrxR) activity

TrxR activity was assessed using a commercially available kit. A549 cells were incubated in T25 flasks at 1.26 × 10⁶ cells in the presence of 50 µg Fe per mL of IONPs or without (control group). After the incubation (6, 24 or 48 h), the cells were washed with PBS, trypsinized and collected after centrifugation (1000 rpm, 5 min, 4 °C). Cells were resuspended in lysis buffer (9% w/w sucrose; 5% v/v aprotinin in deionized water) and lyzed using a Dounce homogenizer. The enzymatic activity of TrxR was then measured by monitoring the reduction of 5,5′-dithiobis(2-nitrobenzoic) acid into 5-thio-2-nitrobenzoic acid at 412 nm for 10 min using a spectrophotometer (SpectraMax M2, Molecular Devices, CA, USA).

X-ray irradiation

The radiosensitizing properties of IONPs were evaluated using a constant X-ray beam from an X-Rad 225 XL (PXi Precision X-ray, CT, USA) machine operating at 225 kV with a dose rate of 2 Gy min⁻¹. Cells were seeded in 24-well plates at a density of 5 × 10⁴ cells per well before irradiation. The wells were initially filled with cell culture medium + 10% FBS and left in the incubator overnight. Before irradiation, the medium was replaced by cell culture medium + 10% FBS, with or without 50 µg Fe per mL of IONPs for an incubation time of 6, 24 or 48 h at 37 °C. Prior the irradiation, the cells were washed with PBS to remove the nanoparticles and the wells were refilled with cell culture medium + 10% FBS.

Clonogenic assay

After irradiation, cells were detached with 0.25% trypsin and subsequently counted. Cells were seeded at a known density in 6-well plates containing cell culture medium supplemented with 10% of FBS and 1% of penicillin/streptomycin to obtain countable colony numbers for different incubation times. To accurately determine the number of cells seeded, cells were also placed in 24-well plates, fixed with 4% paraformaldehyde after 2 h, washed with PBS, and manually counted under an optical microscope. After 11 days, the colonies were stained with violet crystal in 2% ethanol and then counted. The surviving fraction (SF) was determined by dividing the plating efficiency (PE) of irradiated cells by the PE of control cells. To evaluate the radiosensitizing effect of IONPs, the amplification factor (AF) was calculated. The AF reflects the enhancement of cell death in the presence of IONPs compared with radiation alone at a given dose, in this case, 2 Gy.wherewhere

γH2AX immunofluorescence labeling and visualization

DNA damage induction and repair was assessed by γH2AX immunofluorescence labeling before and 24 hours after X-ray irradiation. A549 cells were seeded on a 12 mm round coverslip at a density of 200 000 cells per well (24-well plate) and placed in the incubator at 37 °C with 5% CO₂ overnight. The medium was then replaced by MEM supplemented with 10% of FBS with or without 50 µg Fe per mL of carboxylated IONPs for 24 or 48 h. After this incubation period, the 7 nm carboxylated IONPs are removed and the immunostaining procedure is carried out either before or 24 hours after X-ray irradiation from the plate, using the same procedures as described earlier in this study.

The cells were rinsed with PBS before being fixed with 4% paraformaldehyde for 10 minutes at 4 °C, followed by 5 minutes at room temperature. The cells were then rinsed three times with PBS before being permeabilized for 5 minutes at room temperature in a PBS/Triton 0.1%. Non-specific binding sites were blocked by incubating the cells for 1 hour at room temperature in a PBS/normal goat serum 5% Triton 0.3%. The cells were then incubated with the anti-γH2AX primary antibody overnight at 4 °C (400-fold dilution of the commercial stock in a PBS/BSA 0.1%/Triton 0.3%). Incubation with the secondary anti-rabbit Alexa 488 antibody was carried out for 1 hour in the dark at room temperature (500-fold dilution of the commercial stock in a PBS/BSA 0.1%/Triton 0.3%). Slides were then mounted with Fluoromount-G containing DAPI and imaged by confocal microscopy (Nikon Ti2 microscope) at 60× magnification.

The number of foci per cell was determined using the Foci Analyzer V.1.92 plugin. Optimisation of the plugin parameters values was conducted so that the plugin's output values correspond to those obtained from a manual count of a representative sample of nuclei. Pan-nuclear stained nuclei, a common sign of late apoptosis,²⁵ were manually excluded from the datasets considering that the aspect was incompatible with the quantification of nuclear foci.

Nanoparticle localization

The rhodamine-labeled nanoparticle uptake in A549 cells was analyzed by confocal laser scanning microscopy. A549 cells were seeded at a concentration of 1.1 × 10⁵ cells per well in µ-slide 4 well ibiTreated (Proxylab, Beloeil, Belgium) and placed in the incubator at 37 °C with 5% CO₂ overnight. The medium was then replaced by MEM supplemented with 10% of FBS with or without 50 µg Fe per mL of IONPs for 2 h; 6 h; 24 h; 48 h. Finally, the cells were incubated for 30 minutes with the LysoTracker Blue at a final concentration of 50 nM in growth medium without NPs. Cells were examined with a Zeiss LSM710/AxioObserver Z1 confocal microscope using a plan-Apochromat 63x/NA 1.4 oil DIC M27 immersion objective in a thermostated chamber (XL/LSM incubator, Zeiss; Tempcontrol 37-2, PeCon) at 37 °C. The images of the samples were acquired in z-stack (15 slices) mode with a slice thickness of 0.68 µm.

Investigation of internalization pathways

Understanding the mechanisms by which nanoparticles enter cells was investigated using both pharmacological inhibition and genetic silencing approaches. Cells were maintained at 37 °C in a humidified incubator with 5% CO₂. For internalization studies, A549 cells were seeded at a density of 2 × 10⁵ cells per well in 4-well chamber slides containing glass coverslips and allowed to adhere for 24 h prior to treatment. Cells were then incubated for 1 h with rhodamine-labelled iron oxide nanoparticles at a concentration of 100 µg Fe per mL, at 37 °C. Following incubation, cells were washed three times with PBS and fixed with 4% paraformaldehyde. Slides were then mounted with Fluoromount-G containing DAPI and imaged by confocal microscopy (Zeiss LSM900 confocal microscope equipped with an Airyscan detector, an α Plan Apo 63× numerical aperture (NA) 1.4 oil immersion objective and a stage-top incubator – Morph-Im Platform of UNamur).

Investigation of macropinocytosis-mediated uptake

The contribution of macropinocytosis to IONP internalization was assessed using amiloride (25 µM), a known inhibitor of the Na⁺/H⁺ exchanger. To ensure continuous inhibition throughout the exposure period, cells were pre-treated with amiloride for 30 min, followed by co-incubation for 1 h at 37 °C with a solution containing both IONPs (50 µg Fe per mL) and amiloride. After incubation, cells were carefully washed with PBS, fixed with 4% paraformaldehyde, mounted with Fluoromount-G containing DAPI and imaged by confocal microscopy as described above.

Investigation of clathrin-mediated endocytosis

Clathrin-mediated endocytosis was targeted by RNA interference using small interfering RNAs (siRNAs) against the clathrin heavy chain (CHC) or the AP2 subunit, with a non-targeting siRNA (siCTL) used as a control. Cells were transfected for three consecutive days with 40 nM siRNA using Lipofectamine RNAiMAX (2 µL per well in 400 µL Opti-MEM, a reduced-serum medium optimized for siRNA-lipid complex formation and cellular uptake).

Seventy-two hours after the final transfection, cells were incubated with IONPs following the standard internalization protocol. Knockdown efficiency was validated by Western blot analysis. Cells were lyzed in RIPA buffer supplemented with protease and phosphatase inhibitors and total protein content was determined using a BCA assay. Proteins were separated by SDS-PAGE and transferred to PVDF membranes (Turbo Trans-Blot system, Bio-Rad). Target proteins (CHC, AP2, and GAPDH) were detected using specific primary antibodies, followed by HRP-conjugated secondary antibodies (anti-mouse-HRP or anti-rabbit-HRP, depending on the primary antibody host). Detection was carried out using a chemiluminescent substrate.

Image processing and quantification were performed using (Fiji Is Just) ImageJ v2.16.0/1.54p (NIH).

For each condition, confocal z-stack images were imported into FIJI and converted to maximum-intensity projections using the Z Project function. Regions of interest (ROIs) corresponding to individual A549 cells were manually delineated using the “Freehand selection” tool. Within each ROI, the intracellular fluorescence signal associated with IONP uptake was quantified by measuring the Raw Integrated Density (RawIntDen) on the projected image. Background intensity was measured in an acellular region of each field of view and subtracted from all cellular measurements. For each experimental condition, fluorescence values were normalized to the corresponding control (set as 100%) and pooled across independent experiments for statistical analysis.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). All experiments were conducted independently and repeated at least three times on different days. Data are expressed as mean ± standard deviation (SD).

Prior to statistical testing, data distribution was assessed. Quantitative experimental readouts derived from independent biological replicates (intracellular iron concentration, ROS levels, and TrxR enzymatic activity) showed no major deviation from normality and were therefore analyzed using parametric tests. One-way analysis of variance (ANOVA) was applied when comparing multiple experimental groups with a single independent variable, followed by Tukey's or Dunnett's multiple comparisons tests, as appropriate. Two-way ANOVA was used when two independent variables (treatment and incubation time) were considered simultaneously, including evaluation of potential interaction effects, followed by Dunnett's post-hoc test or Tukey test.

In contrast, single-cell fluorescence intensity data obtained from confocal microscopy exhibited non-Gaussian distributions characterized by high intercellular variability. Normality tests confirmed that these datasets did not follow a normal distribution. Consequently, statistical comparisons between two independent groups were performed using the non-parametric Mann–Whitney test.

Statistical significance was defined as *p < 0.05, **p < 0.01, ***p < 0.001.

Conclusion

Taken together, our findings highlight the multifactorial nature of IONP-mediated radiosensitization. The observed effects cannot be attributed to a single parameter, such as core size or composition, but rather arise from a complex interplay between physicochemical properties and biological responses. Surface coating, colloidal stability, protein corona formation, cellular uptake, intracellular trafficking, lysosomal degradability and interference with antioxidant defenses all emerge as interdependent determinants of efficacy. Among the mechanisms investigated, TrxR inhibition appears to be strongly associated with enhanced radiosensitization, particularly for the most effective carboxylated formulations. This supports the idea that modulation of redox homeostasis contributes to the radiosensitizing response, beyond mere nanoparticle accumulation. In addition, although not directly investigated here, the reported radical-scavenging properties of polysaccharide coatings such as dextran may represent an additional factor contributing to the weaker radiosensitizing activity of the commercial formulations.

Consequently, the rational design of future nanoradiosensitizers should not rely solely on optimizing size or core composition, but should integrate how nanoparticle surface chemistry influences intracellular fate and redox regulation. In this context, carboxylated IONPs, and particularly 7 nm TEPSA-coated nanoparticles, represent promising candidates for further investigation. Future studies should now focus on quantitative in vivo biodistribution and exposure–response relationships to determine whether the most effective formulations identified here can achieve therapeutically relevant tumor accumulation. Such studies will be essential to establish how nanoparticle accumulation, intracellular processing and redox modulation collectively translate into radiosensitizing efficacy in more complex biological settings.

Author contributions

The manuscript was written through contributions of all authors. The experiments were mainly designed by I. T., A. C. H., H.-F. R., C. M., D. S. and S. L. and performed by I. T., V. L., E. B., D. S., C. S. and T. V. I. T., V. L. and D. S. prepared and characterized the nanoparticles. I. T. V. L., E. B., C. S. S. B. and H.-F. R. designed and performed cell culture experiments and related analyses. I. T., V. L., A. C. H., S. P., C. M., E. B. and D. S. designed and performed the X-ray irradiation. I. T. and V. L. contributed equally to this work. All authors have given approval to the final version of the manuscript.

Conflicts of interest

The authors declare that they have no competing interests.

Data availability

The main data supporting the results in this study are available within the paper and the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6ra00750c.