Efficient uptake and retention of iron oxide-based nanoparticles in HeLa cells leads to an effective intracellular delivery of doxorubicin

The purpose of this study was to construct and characterize iron oxide nanoparticles (IONPCO) for intracellular delivery of the anthracycline doxorubicin (DOX; IONPDOX) in order to induce tumor cell inactivation. More than 80% of the loaded drug was released from IONPDOX within 24 h (100% at 70 h). Efficient internalization of IONPDOX and IONPCO in HeLa cells occurred through pino- and endocytosis, with both IONP accumulating in a perinuclear pattern. IONPCO were biocompatible with maximum 27.9% ± 6.1% reduction in proliferation 96 h after treatment with up to 200 µg/mL IONPCO. Treatment with IONPDOX resulted in a concentration- and time-dependent decrease in cell proliferation (IC50 = 27.5 ± 12.0 μg/mL after 96 h) and a reduced clonogenic survival (surviving fraction, SF = 0.56 ± 0.14; versus IONPCO (SF = 1.07 ± 0.38)). Both IONP constructs were efficiently internalized and retained in the cells, and IONPDOX efficiently delivered DOX resulting in increased cell death vs IONPCO.


Results
Physical and chemical characterization of the IONP. Synthesis of IONP using a three-step synthesis method leads to the generation of highly crystalline individual iron oxide cores with average diameters of 12.82 ± 2.73 nm (Fig. 1a-c), showing diffraction rings characteristic for face-centred spinel structured magnetite ((220), (222), (400), (422), (333), (440)) ( Fig. 1d). Individually covered NPs with PEG were organized as coreshell-like nano-constructs, as shown by high resolution transmission electron microscopy (HR-TEM; Fig. 1c; PEG shell emphasized with white arrows). HR-TEM confirmed the data on crystallinity and emphasized the presence of the (220) plane of 0.29 nm width (Fig. 1c), characteristic for the magnetite phase. Conjugation with PEG also lead to a stabile dispersion of the NPs stock solution in water, a mean hydrodynamic diameter of 164.2 nm being measured (prior to ultrasound dispersion). Zeta potential measurements showed good stability (14.80 mV for stock solutions with no prior ultrasound treatment). Loading of the DOX resulted in an increase in the hydrodynamic diameter (369.1 nm mean diameter) and a change of surface charge to negative values (−20.9 mV zeta potential). Both of the constructs were monodisperse systems, as the values of the polydispersity index (PDI) were below 0.3 (0.233 for IONP CO and 0.238 for IONP DOX ). Quantitative determination of the DOX-loading content in IONP showed a value of 1.11 wt% (Supplementary Material Section 2).

Release of DOX from IONP DOX .
The release experiments were carried out in three biologically relevant culture media with different pH values: 7.4 pH, which is characteristic for physiologic plasma, 6.5 pH which is relevant for tumour microenvironment 27,28 and 4.8 pH, which is encountered in the intra-lysosomal compartment 29 . IONP DOX showed a rapid, initial release which was not significantly affected by the pH (Fig. 1e).

Internalizing and retention of IONP in HeLa cells. At 24 h after 16 h incubation with nanoparticles,
internalized IONP were shown not to translocate into the nucleus, remaining organized as light blue-coloured agglomerates covering the outer nuclear membrane (Fig. 2c,d). The morphology of the samples was not affected in case of IONP CO -exposed HeLa cells (Fig. 2c). However, the cell density was affected by DOX from the IONP DOX samples (Fig. 2d) and by equivalent amounts of free DOX (Fig. 2b). The morphology of IONP DOX -treated HeLa cells changed, becoming rounder and larger. In addition, nuclei increased in volume. By adjusting the focusing plane, different depth levels of accumulated IONP could be differentiated in the cell (Supplementary Figs. S1, S2). Thus, most nanoparticle agglomerates appeared in a perinuclear location in the cell. Still, some can be seen directly interacting with the cell membrane during the internalizing process.
In case of IONP DOX , localizing of the nano-constructs inside the cells was confirmed by fluorescence detection of DOX 24 h after 16 h incubation with DOX-loaded nanoparticles (Fig. 2e). A signature red fluorescence of the IONP DOX aggregates (sub-micron spherical structures in the peri-nuclear area and the cytoplasm) was observed with a weaker intensity in the remaining cytoplasm and in the nucleus (Fig. 2e).
Transmission electron images were acquired for HeLa cells exposed to different concentrations of IONP (100 and 500 μg/mL) for 16 h in order to show their internalization and localization in HeLa cells at 24 h after NP removal. This technique was also employed to potentially study the mechanisms of entrapment. Figure 3 emphasized the localizing of IONP as agglomerates in the perinuclear area and the cytoplasm. The results (Fig. 3e-g and Supplementary Fig. S4) show that both types of constructs are internalized via macropinocytosis (Fig. 3e) and sometimes smaller aggregates are internalized via endocytosis (Fig. 3f), eventually being transferred in lysosomes (Fig. 3g). At 24 h after the end of NP incubation period, the IONP CO appeared to be entrapped in intracellular vesicles (Fig. 3a,b and, while IONP DOX aggregates were localized in both vesicular structures and appeared free in cytoplasm (Fig. 3c,d).

Effect of IONP on the proliferation kinetics of HeLa cells.
The effect of IONP on the proliferation of HeLa cells was determined for a broad concentration range of IONP CO and IONP DOX (0-200 µg/mL) during 48-96 h incubation. Results were shown relative to controls (untreated HeLa cells = 100%).
With IONP CO , a time and concentration-dependent decrease in proliferation was observed until 72 h of incubation with nanoparticles ( Fig. 4a,b). All results were statistically significant relative to untreated cells (One-way ANOVA, p < 0.05). However, after 96 h of incubation, no significant changes in relative absorbance were observed compared to control samples, a maximum reduction of relative absorbance being 10.88 ± 7.68% for the highest concentration employed (200 µg/mL IONP) (Fig. 4c). www.nature.com/scientificreports www.nature.com/scientificreports/ DOX-free nanoparticles caused a delay in cells proliferation up to 72 h. This effect was diminished at short incubation periods (24/48 h), the effect of IONP CO on the HeLa cells could be rather correlated with an initial cytotoxic or growth inhibitory effect, as control cells normally underwent maximum 2 divisions during this time interval (doubling time of HeLa cells is ~18 h 30 ). The effect on the cell's ability to reduce the tetrazolium salt might  One-way ANOVA statistical analysis revealed a significant difference between treated groups and control; Two-way ANOVA statistical analysis proved significant difference between IONPs and DOX-IONPs (P < 0.0001 for 48 h; P < 0.0001 for 72 h; P < 0.0001 for 96 h). Also, the presence of DOX in the construct induced a significant reduction of proliferation, compared to equivalent concentrations of IONP CO (P = 0.0003 for 48 h; P < 0.0001 for 72 h; P < 0.0001 for 96 h). (d) Clonogenic survival of HeLa cells seeded in the colony formation assay after exposure to 100 μg/mL IONP for 16 h. Data are presented as percentage of untreated control and are shown as mean ± SEM (n = 3).
www.nature.com/scientificreports www.nature.com/scientificreports/ be justified by an initial toxicity with time ( Fig. 4a,b). At later time points (96 h), showing values close to control cells suggested that the initial effect of the IONP was rather based on growth delay/inhibition than cell kill/cytotoxicity (Fig. 4c).
Incubation of HeLa cells with IONP DOX showed a clear cytotoxic effect that increased with the concentration of administered NP and time ( Fig. 4a-c). The calculated IC 50 values were 27.83 ± 7.99 µg/ml for 48 h of NP incubation, 2.31 ± 0.32 µg/ml for 72 h of NP incubation, respectively 9.01 ± 4.68 µg/ml for 96 h of NP incubation. While IONP CO showed nearly no effect on the metabolic activity of HeLa cells at 96 h compared to untreated controls, even at very high doses (200 µg/ml), incubation with IONP DOX for 96 h showed a pronounced effect, showcasing a reduction in signal of 62.62 ± 2.05% (P = 0.002). All data were statistically significant compared to controls (non-treated cells), as shown by One-way ANOVA analysis. Moreover, Two-ways ANOVA statistical analysis showed a significant difference between IONP CO and IONP DOX (P < 0.0001 for 48 h; P < 0.0001 for 72 h; P < 0.0001 for 96 h). Also, the presence of DOX in the construct induced a significant cytotoxic effect in the HeLa cells (P = 0.0003 for 48 h; P < 0.0001 for 72 h; P < 0.0001 for 96 h).
Incubation with IONP for 16 h caused a change in the cell cycle distribution (Fig. 5a-c), evidenced through an increase in the number of cells in G 2 phase, compared to non-treated controls (Fig. 5c). This effect was observed at the time of removing the NP (25.90 ± 1.90 for IONP CO , 26.95 ± 1.45 for IONP DOX , respectively 18.03 ± 6.15 for control cells, where P = 0.05 for IONP CO , P = 0.03 for IONP DOX , compared to control), and again at 16 h after removing the NP (21.00 ± 4.10% for IONP CO , 19.10 ± 0.60% for IONP DOX and 12.90 ± 0.70% for untreated controls; P = 0.013 for IONP CO , respectively P < 0.001 for IONP DOX compared to control). The G 2 /M fraction decreased at 24 h which was paralleled by an increase of the fraction of G 1 cells, suggesting induction and release of a temporary G 2 /M block. Measurements of cell division (number of cell doublings) (Fig. 5d) showed a minimal but statistically significant difference only at 16 h after nanoparticles removal caused by NPs (untreated controls vs IONP CO : P = 0.04; untreated controls vs IONP DOX : P = 0.03) as the effect was independent of DOX loading (IONP CO vs IONP DOX : NS).
The clonogenic survival assay (Fig. 4d) emphasized the biocompatible character of the DOX-free NP with SF(IONP CO ) = 1.07 ± 0.38, while the incorporation of DOX in the IONP polymeric shell caused a reduction in HeLa cells survival with SF(IONP DOX ) = 0.56 ± 0.14 which was attributed to the release of DOX in the cells.

Discussion
In this study we designed and synthesized polyethylene glycol (PEG)-functionalized iron oxide nanoparticles (IONP CO ) for the encapsulation of the chemotherapeutic substance doxorubicin (IONP DOX ). We determined the uptake efficacy of the IONP and evaluated their ability to induce cell death after DOX loading. Results showed that most of the loaded drug was released from IONP DOX within 24 h, with a complete release at 70 h. Internalization of IONP DOX and IONP CO into HeLa cells occurred by pino-and endocytosis, with both IONPs accumulating in the peri-nuclear area. DOX-free nanoparticles proved to be biocompatible for HeLa cells, while the cells treatment with IONP DOX determined a concentration and time-dependent decrease of cells proliferation.
The investigations highlight the intracellular fate of IONP after discontinuing the NP exposure, both quantitatively (through accurate measurements of the retained Fe 3 O 4 per cell) and qualitatively (through electron www.nature.com/scientificreports www.nature.com/scientificreports/ microscopy imaging, in relation to the cellular compartments). Moreover, the study evaluates the IONP influence on the cell cycle and long-term proliferation/clonogenic survival after discontinuing the NP exposure. IONP DOX were made using a three-step synthesis method. The novel method that we have developed is adequate for large scale extension due the advantages like ease, low costs, high yield synthesis and reproducibility. In a first step, bare Fe 3 O 4 nanoparticles were produced by modified room temperature chemical co-precipitation similar to 31 , resulting in highly crystalline face-centred spinel structured magnetite nanoparticles 32,33 (Fig. 1d). Post-synthesis conjugation of the iron oxide cores using PEG (molecular weight 6 KDa) resulted in individual coverage of the IONP with a low crystalline organic phase, forming core-shell-like nano-constructs with high stability and positive surface charge (Fig. 1a-d). Previous results showed that PEG-conjugated nanoparticles have a positive charge in solutions with pH<8 34 , in concordance with our measurements. Dissociation of DOX • HCl in water not only determined a change of IONP charge due to alteration of ions concentration in loading solution, but also led to higher hydrodynamic diameters, following DOX entrapment and interaction with PEG shells.
A challenge in developing nanoparticle-based drug delivery systems is finding an optimal design that enables internalizing, as well as retention in targeted cells. The microscopy investigations emphasized the nanoparticles localization in the cytoplasm of the cells (Fig. 2). In addition, a slight gradient effect of DOX, which might be a result of drug release from the IONP was observed (Fig. 2e and Supplementary Fig. S3). These results correlate with the DOX release data, considering the fact that the fluorescence microscopy investigations were done at 24 h after the end of NP treatment. Moreover, Fig. 2e proves the stability of DOX loading in IONP and suggests that the release is only triggered by the intracellular environment. In case of free DOX ( Supplementary Fig. S3), due to its small size, directly diffused into the nucleus of the cell after few hours of incubation, while IONP DOX samples were showing signal mainly in the cytoplasm and weak signal in the nuclei (Supplementary Fig. S3). Instability of DOX loading in IONP in stock solutions would cause a release of DOX in the buffer solution and thus the direct diffusion of the chemotherapeutic molecule in the cell nucleus accompanied by a lack of fluorescence signal from the cytoplasmic compartment.
Similar results were reported by Zhang Y. et al. 35 for 180 nm Poly(ethylene glycol)-doxorubicin-curcumin nanoparticles which were mainly located in the vicinity of the nucleus at low incubation times, while the free drug diffused into the nucleus of the cells, due to a differentiation in the uptake pathways. This changed with time, as the active substances were released from the constructs.
For all experimental conditions, agglomerates of IONP could be observed both in the peri-nuclear area and the cytoplasm (Figs. 2, 3). The presence of IONP CO was mostly observed in vesicle-like structures (Fig. 3a,b), due to entrapment in the endo-/lysosomal compartments 36 , while IONP DOX were also found in the cytoplasm (Fig. 3c,d). Bypassing the endo-/lysosomal system is a requirement in the design of intracellular drug delivery systems 37 .
Petros et al. 38 stated that nanoparticles having higher hydrodynamic diameter are transported across the cellular membrane via clathrin-mediated endocytosis or macropinocytosis. This was confirmed by Pearson et al. 39 and Jana 40 . Bannunah et al. 41 showed that dimension does not play an important role in the NP mechanism of internalizing, but it is dependent on surface charge; also, there is more than one mechanism involved in the internalizing of the same type of NP. Our results show that both types of constructs are internalized via macropinocytosis ( Fig. 3e and Supplementary Fig. S4c) and sometimes smaller aggregates are internalized via endocytosis ( Fig. 3f and Supplementary Fig. S4d). Both mechanisms were observed for DOX-loaded and DOX-free IONP, meaning that the cells do not use one mechanism or another based on surface charge, but rather on dimension. Extensions of the cellular membrane surround the NP agglomerates in the vicinity of the membrane, forming micrometre-sized vesicles and getting the constructs into the intracellular compartment. Eventually, these get trapped in lysosomes for cellular disposal (Fig. 3g and Supplementary Fig. S4e).
Our observations agree with results from other publications 26 . However, in case of DOX-loaded constructs, the major fraction of NP seem to escape the lysosomal trapping and to be freely located into the cytoplasm at 24 h after discontinuing the exposure of HeLa cells to IONP DOX (Fig. 3c,d). The development and fate of PEG conjugated iron oxide nanoparticles to escape the endo-/ lysosomal trapping in the context of drug delivery is not well studied. This bypassing approach of iron oxide nanoparticles has been investigated for the purpose of radiosensitization, in case of dextran coated iron oxide nanoparticles conjugated with a cell penetrating peptide (TAT) 42 and citrate coated superparamagnetic iron oxide nanoparticles 43 .
Lysosome function is to digest the internalized material taken up by the cell by means of endocytosis 44 , thus escaping or bypassing the lysosome uptake might be a solution to improve organelle specific targeting. In case of nanoparticle formulas, endosome and lysosome inclusions might also be an obstacle for effective treatment, as stated by Lloyd 45 . Whereas nano-carrier degradation by lysosome microenvironment and liberation of the active substance may still be considered one important principle of nanoparticle-based drug delivery systems, lysosome membrane can act as a natural barrier against efficient drug release 46 . In this case, cytosolic delivery of the drugs by nanoparticles escaping lysosomal entrapment 47,48 might be a key to successful killing of the cancer cells.
Besides efficient uptake of nanoparticles in cancer cells, their retention is also important so that an effective high concentration can be reached. While small nanoparticles (diameters lower than 50 nm) undergo exocytosis within 24 h of uptake, larger nanoparticles or aggregates are retained for longer periods of time 49 . Our microscopy investigations emphasized the presence of IONP after 40 h (16 h of incubation with NPs and additional 24 h incubation without NPs), while quantitative measurements showed that, at this time-point, a 3.6-fold higher concentration of Fe 3 O 4 was measured in IONP DOX compared to IONP CO (Fig. 3h). This difference in internalized Fe 3 O 4 in the two groups of IONP-exposed HeLa cells might be due to the difference in hydrodynamic diameter (almost double for IONP DOX compared to IONP CO ), but also due to the induction of cell death in case of DOX-loaded nanoparticles, which can artificially increase the concentration Fe 3 O 4 /cell. To the best of our knowledge this is the first study to evaluate the intracellular retaining and fate of PEG-coated iron oxide nanoparticles (qualitatively, www.nature.com/scientificreports www.nature.com/scientificreports/ in relation to the cellular compartments and also quantitatively, by providing an accurate concentration of NP per individual cell) at periods of time longer than one complete cell cycle, after the termination of NP exposure.
Considering compatibility to human blood, our results showed biocompatibility for both IONP CO and IONP DOX (Supplementary Material Section 7), matching data from the literature [50][51][52] . Concerning the cytotoxicity and proliferation exhibited by HeLa cells after IONP exposure, our data (Fig. 4a-c) showed that, in the first days of interaction between cells and PEG-functionalized NP, a weak cytotoxic effect occurred already at very small concentrations, which did not increase with concentration, after a certain threshold, the amount of cytotoxicity being maintained almost constant. Similar results were obtained by Xia et al. 53 for a redox responsive polyethylene glycol-Fe 3 O 4 nanoparticles self-assembled micelles. However, at later time points (96 h), the cell proliferation was maintained above 80% limit (relative to control), which is a threshold for biocompatibility (ISO 10993-12:2001(E) 54 ). Long-term monitoring (Fig. 4d) showed that IONP CO did not affect the clonogenic survival of HeLa cells after 14 days. Other studies also reported PEG-coated iron oxide nanoparticles as biocompatible 55-57 . Feng et al. 26 showed that such nanoparticles could induce autophagy in vitro, but otherwise showed no obvious signs of in vivo toxicity in BALB/c mice.
Results for IONP DOX showed that DOX-containing NP caused significant cytotoxicity and growth inhibition in HeLa cells compared to untreated control samples (Fig. 4a-c). This effect was concentration and time dependent, showing that these constructs can be efficiently used to induce cell death in human cervical adenocarcinoma cells. The lack of recovery at 96 h (as compared to IONP CO ) suggests that IONP DOX not only caused growth delay but also real cytotoxicity. The clonogenic survival data confirmed this by showing a reduced SF in cells treated with IONP DOX after 14 days compared to IONP CO -treated cells (Fig. 4d).
The cell cycle arrest at 16 h after IONP CO treatment was also observed for IONP DOX (Fig. 5) indicating little effect of DOX released from the nanoparticles. In all, results showed no statistically significant difference in cell cycle distribution between the IONP CO and IONP DOX groups (Fig. 5).
This study shows the generation and characterization of polyethylene glycol-functionalized iron oxide nanoparticle that have shown efficient internalizing and retention in human cervical adenocarcinoma cells. Highly crystalline, bio-and hemocompatible nanoparticles, IONP have the ability to encapsulate and deliver the chemotherapeutic doxorubicin directly into the intracellular cytoplasmic compartment and thereby efficiently causing cell death in the cells. This makes them potential candidates for nanoparticle-mediated and chemotherapy-induced inactivation of tumour cells. The morphology, crystallinity and mineralogical composition of the IONP was examined using transmission electron microscopy (TEM) on a Tecnai G2 F30 S-TWIN HR-TEM (Thermo Fisher Scientific, Hillsboro, OR, USA) equipment with selected area electron diffraction (SAED) module, for which sample preparation has been described previously 31 . The analysis of the loading efficiency of DOX into the nano-constructs is described in Supplementary Material Section 2.

Methods
The hydrodynamic diameter and surface charge (zeta potential) of IONP were characterized using a Delsa Nano C instrument (Beckman Coulter, Brea, CA, USA) and recorded using the DelsaNano 3.73 software (Beckman Coulter). The measurements were done for freshly prepared nanoparticle suspensions in ultrapure water without prior ultrasound dispersion, based on the existing international documentary standards ISO 13321:1996 58 and ISO 22412: 2008b 59 .
The release kinetics of DOX from IONP DOX was measured for media with different biologically relevant pH: 7.4, 6.5 and 4.7. Studies were done in standard conditions of temperature and humidity (37 ± 2 °C, 5 ± 1% CO 2 ). Samples were prepared and analysed as described in the Supplementary Material Section 3 and performed once in triplicate.
Cell culture. The biological evaluation of the IONP was performed on the human cervical adenocarcinoma cell line HeLa, which was obtained from the Tumour Cell Bank of the German Cancer Research Centre (DKFZ, Heidelberg, Germany). This cell line was chosen because it has been previously used in a variety of studies 49 Treatment and incubation with IONP. HeLa cells were seeded in appropriate concentration for each investigation and allowed to attach for 4 h. Afterwards, the culture medium was replaced with fresh medium containing nanoparticles in the respective concentrations and incubated for 16 h.
Uptake and retention of IONP. The uptake and retention of IONP by HeLa cells was evaluated by three microscopic methods generating complementary information on the localizing of the nanoparticles in relation to the cultured tumour cells. The preparation of the samples for microscopy is described in the Supplementary Material Section 4. Optical visualization of IONP in HeLa cells 24 h after incubation (for 16 h) in presence of nanoparticles was performed using a Prussian Blue staining, resulting in a light blue colouring of sub-micron structures. Fluorescence imaging was possible due to native property of doxorubicin. Optical and fluorescence microscopy were performed using a Leica DMRE microscopy equipped with a Leica DFC3000G camera (Leica Mikrosysteme Vertrieb GmbH Mikroskopie und Histologie, Wetzlar, Germany) and an Axio Observer Z1 www.nature.com/scientificreports www.nature.com/scientificreports/ microscope (ZEISS, Oberkochen, Germany) equipped with an Axiocam 506 camera. Images were acquired using ZEN 2 software (ZEISS). Samples for transmission electron imaging were prepared similarly as for optical and fluorescence microscopy imaging (as described in Supplementary Material Section 4). Images were acquired using a Zeiss EM 10 transmission electron microscope (ZEISS) equipped with an Olympus Megaview G2 camera (Olympus Europa SE & Co. KG, Hamburg, Germany).
Particle-induced X-Ray emission (PIXE) was performed using a 3 MV Tandetron accelerator with a 2.7 MeV proton beam and an in-air irradiation setup 62 . The characteristic X-Rays were recorded by an Amptek silicon drift detector (SDD) positioned at 45° with respect to the beam direction and the spectra were processed with the GUPIXWIN 2.2.4 software 63 . The SDD resolution is 130 eV at 5.9 keV (K α line of 55  Proliferation and clonogenic survival. The quantitative effects of the IONP on proliferation kinetics were evaluated using a tetrazolium salt-based proliferation assay (MTT, Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany). Dose-response curves were obtained for concentrations up to 200 μg/mL IONP continuous exposure for 48, 72 and 96 h. Samples were prepared as described in the Supplementary Material Section 5.
Cells for longer term survival evaluation (14 days) were incubated with IONP for 16 h and then detached and reseeded into the colony formation assay at 200 cells/ 25 cm 2 flask as described in Supplementary Material Section 5.
Cell cycle distribution and doubling time. The change in cell cycle distribution and division of HeLa cells exposed during 16 h with IONP was evaluated for cells treated and stained as described in Supplementary Material Section 6. Acquisition was performed on the BD FACSLyric (BD Biosciences, San Jose, CA, USA), and analysed using FlowJo 10.5 software (BD Biosciences).

Statistical analysis.
All data are presented as mean ± SEM from three independent experiments, unless specified otherwise. Statistical analysis was done using t-test, one-way ANOVA (SigmaPlot 12) and two-way ANOVA (Prism 5, GraphPad, San Diego, USA).

Data availability
All data generated or analysed during this study are included in this published article (and its Supplementary Information File).