Gold-decorated magnetic nanoparticles design for hyperthermia applications and as a potential platform for their surface-functionalization

The integration of noble metal and magnetic nanoparticles with controlled structures that can couple various specific effects to the different nanocomposite in multifunctional nanosystems have been found interesting in the field of medicine. In this work, we show synthesis route to prepare small Au nanoparticles of sizes  = 3.9 ± 0.2 nm attached to Fe3O4 nanoparticle cores ( = 49.2 ± 3.5 nm) in aqueous medium for potential application as a nano-heater. Remarkably, the resulted Au decorated PEI-Fe3O4 (Au@PEI-Fe3O4) nanoparticles are able to retain bulk magnetic moment MS = 82–84 Am2/kgFe3O4, with the Verwey transition observed at TV = 98 K. In addition, the in vitro cytotoxicity analysis of the nanosystem microglial BV2 cells showed high viability (>97.5%) to concentrate up to 100 µg/mL in comparison to the control samples. In vitro heating experiments on microglial BV2 cells under an ac magnetic field (H0 = 23.87 kA/m; f = 571 kHz) yielded specific power absorption (SPA) values of SPA = 43 ± 3 and 49 ± 1 μW/cell for PEI-Fe3O4 and Au@PEI-Fe3O4 NPs, respectively. These similar intracellular SPA values imply that functionalization of the magnetic particles with Au did not change the heating efficiency, providing at the same time a more flexible platform for multifunctional functionalization.

exposed to an electromagnetic field. The enhancement in the power absorption must be due to the interfacial exchange interaction between magnetic phases in core-shell systems. It must be noted that Fe 3 O 4 NPs by themselves have also a photothermal response to a ≈800 nm laser excitation, as demonstrated in vitro and in vivo by Espinosa et al. 5 , although the question of whether the photothermal response of magnetite could be tuned to different optical wavelengths still remains open. Due to these different requirements the synthesis of Au-Fe 3 O 4 nanoparticles with a core-shell structure has been difficult to achieve in a reproducible way [7][8][9] . The approach of synthesizing Au-decorated magnetic NPs offers the flexibility to the plasmonic responsiveness by selecting the appropriate size of the Au NPs and its surface functionalization due to the strong adsorption ability 10 and its reactivity with thiolated 11 or disulphide groups that makes easier to functionalize them 12,13 . A recent approach to coat polymeric magnetic nanostructures with Au by attaching gold seeds to the NPs surface followed by the reduction of Au has been reported 3 . However, it is somewhat difficult to control the particle aggregation and the uniformity and thickness of the gold shell. Some attempts have been reported about the synthesis of Fe 3 O 4 /Au hybrid structures using different polymers as a platform to attach Au NPs 14,15 , including polymers such as poly(ethylenimine) (PEI), poly(acrylic acid) or dextran. The branched form of PEI is an appealing choice due to the exposed amine groups that provide abundant active sites for chemical modification 16,17 . It also prevents the aggregation driven by the dipolar interaction between magnetic cores, providing better stability to the colloid 11,18,19 and could provide the advantage of preventing the Fe 2+ ions to interact with cytoplasmic enzymes promoting the generation of reactive oxygen species through Fenton reaction 20 .
In this work, we reported our initial results of the synthesis and characterization of gold-decorated magnetic nanoparticles to magnetic hyperthermia and as a potential non-toxic carrier for biomedical applications. The reproducibility and morphology of the Au decorated Fe 3 O 4 NPs was confirmed via high resolution transmission electron microscopy (TEM) and HAADF-STEM images and their magnetic properties are conserved. Subsequently, the Au@PEI-Fe 3 O 4 NPs were used for hyperthermia on experimental in vitro.

Synthesis of Au@PEI-Fe 3 o 4 nanoparticles
PEI-coated Fe 3 O 4 nanoparticles (PEI-Fe 3 O 4 NPs) were synthesized in one-step using the oxidative hydrolysis method reported elsewhere [21][22][23] . Subsequently Au nanoparticles were grown on the surface of PEI-Fe 3 O 4 NPs by a modified method previously reported 11,24 . Briefly, a gold solution was prepared using trisodium citrate (0.068 mmol), sodium borohydride (0.019 mmol) as the reducing agent and gold (III) acetate (0.027 mmol) in 50 mL of deionised water. 40 mL of PEI-Fe 3 O 4 NPs in a concentration of 0.085 mg/mL was mixed with the Au solution, stirred and heated up to 60 °C for 10 min. Then a solution of 0.1 g trisodium citrate dispersed in 5 mL of deionised water was added. After the formation the Au@PEI-Fe 3 O 4 NPs, the solution was cooled to room temperature and stirred for 2 hours more. Once the Au was reduced, the solution turned into a deep-red colour indicating the presence of metallic Au NPs and the magnetic separation was an indication of their attachment onto the surface of PEI-Fe 3 O 4 NPs. The magnetic separation of the Au@PEI-Fe 3 O 4 NPs were used to wash several times with distilled water until a final pH = 7 was attained.

Results and Discussion
Our simple procedure to obtain gold-decorated Fe 3 O 4 NPs in aqueous medium comprises the initial synthesis of PEI-Fe 3 O 4 NPs through a mild hydrolysis route and subsequent growth of the Au particles onto the magnetic nuclei by the citrate reduction of Au 3+ . The major advantage of this method is the short reaction time and the straightforward growth of Au NPs directly onto the magnetic PEI-Fe 3 O 4 NPs 11,25 in aqueous medium 26 . This sequential process allows to independently tune the size and morphology of both Au and Fe 3 O 4 phases as required by any specific application. The transmission electron microscopy TEM images of as-prepared magnetic nuclei (i.e., before addition of the Au particles) showed an octahedral morphology of the particles, with an amorphous layer of ~2 nm thickness at the surface (see Fig. S1 of the supplementary material), corresponding to the coating of the PEI polymer. Besides stabilizing NPs in colloidal solution, PEI molecules on the surface of magnetic NPs could also act as a reducing agent for the gold ions in solution 27 . Figure S1 also shows the morphology of isolated Au NPs, as obtained by the same synthesis protocol but without adding the PEI-Fe 3 O 4 NPs.
When the two-stage synthesis of Fe 3 O 4 and Au NPs was performed, the corresponding TEM images for the final Au@PEI-Fe 3 O 4 system showed that the Au NPs have the same size distribution, and were homogeneously distributed onto the PEI surface (see Fig. 1a,b). The fact that the Au NPs remained attached to the magnetite cores after several washes suggests that they are retained by strong electrostatic interactions of the NH 2 + groups of the PEI 22,28 and the carboxylic groups of the citrate ions of the Au surface 19,22 . Statistical analysis of TEM images using log-normal functions to fit the size distributions of both phases yielded mean size <d core > = 49.2 ± 3.5 for the Fe 3 O 4 cores, and <d Au > = 3.9 ± 0.2 for the Au NPs (see also Fig. S1 of the supporting material). HAADF-STEM images (Fig. 1c) confirmed the homogeneous distribution of the Au NPs (bright dots) within the PEI layer at the Fe 3 O 4 surface (darker areas) 11 . The crystal lattice planes spacing (Fig. 1d,e) were indexed within the Fd-3m space group correspond to magnetite phase while the patterns from the Au grain locations were fitted using a face-centered cubic structure (space group: Fm-3m). The magnetic properties of both PEI-Fe 3 O 4 and Au@PEI-Fe 3 O 4 NPs were found to be very similar regarding the coercive field (H c ), saturation magnetization (M s ) and blocking temperatures (T B ). This is consistent with the fact that both samples were synthesized from the very same magnetic cores and that no major influence is expected from the non-magnetic Au NPs. The M(T) data measured after zero-field-cooling (ZFC) and field-cooling (FC) protocols (applied field H FC = 2.39 kA/m) are shown in Fig. 2a for PEI-Fe 3 O 4 and Au@ PEI-Fe 3 O 4 NPs. Both samples showed irreversible behaviour (i.e., separated ZFC and FC branches) up to 300 K, indicating that the magnetic cores are blocked even at room temperature. Consistently, the ZFC curves showed no maximum in temperatures up to 300 K (i.e., the blocking temperature is not below 300 K) as previously reported on similar Fe 3 O 4 NPs with size ≥ 50 nm 31 . Two distinct 'shoulders' were observed in ZFC-mode curves at temperatures T 1 ≈ 45 K and T 2 ≈ 98 K. The shoulder at T 2 has been previously related to the Verwey transition, which occurs at T V = 122 K in bulk Fe 3 O 4 [31][32][33] . The small bump observed in the FC branch at the same temperature supports this interpretation. The shift of the Verwey transition to lower temperatures has been already reported in Fe 3 O 4 NPs with sizes smaller than ≈15 nm, and attributed to size 34 or shape 35 effects. The origin of the second bump at T 1 ≈ 45 K is not clear and might be related to thermal relaxation/unblocking processes of the smallest Fe 3 O 4 (<10 nm) cores observed in TEM images, which is consistent also with the increase of the FC curves at low temperatures due to weak inter-particle interactions of small particles.
Magnetic  www.nature.com/scientificreports www.nature.com/scientificreports/ activation approaching the unblocking temperature, which should be therefore close above 300 K. The estimated effective magnetic anisotropy constant ≈ . for the power absorption has been given by Rosensweig, where τ is the relaxation time of the magnetic moment, and . Therefore, at fixed frequency Eq. (1) reduces to = SPA AH 0 2 , where A is a constant that includes all magnetic parameters of the sample. This quadratic dependence given by the LRT is expected to be valid for is the anisotropy field of the MNPs. This condition is valid working with highly anisotropic particles or very small applied fields. We have performed a systematic investigation of the SPA(H 0 ) dependence with applied field at a fixed frequency (at f = 571 kHz), using PVA to block the Brownian contribution to the magnetic relaxation and thus mimic the high viscosity at the intracellular medium to compare the results to the in vitro measurements (see below). The experimental SPA vs. H 0 data (shown in Fig. 3)were fitted using a phenomenological equation derived from Eq. (1) by assuming all parameters constant except the applied field H 0 (with H 0 ≪ H k ), yielding a power law form where λ is an empirical parameter that allows estimating eventual deviations from the LRT regime (i.e. λ = 2) 40  www.nature.com/scientificreports www.nature.com/scientificreports/ different particle size distributions on the measured SPA cannot be discarded, since no detailed information on the size distributions of the magnetic cores was provided. On the other hand, the work by Bell et al. 43 reported a nearly three-fold increase on the SPA of iron oxide NPs after incorporating Au nanoparticles.
It can be also noticed from Fig. 3 that the fit of the data using Eq. (3) yielded λ ≈ 4.4 ± 0.1 for both Au@ PEI-Fe 3 O 4 and PEI-Fe 3 O 4 samples, as expected for samples in high viscosity media and having the constituent magnetic cores from the same batch preparation. The similar behaviour regarding magnetic relaxation of the two samples reflects the same average particle sizes and distribution.
The UV-vis absorption spectra of PEI-Fe 3 O 4 and Au@PEI-Fe 3 O 4 NPs dispersed in water exhibit a clear variation of the optical properties (Fig. 4) with the PEI-Fe 3 O 4 NPs without significant absorbance in the visible region 44 . In contrast, the Au@PEI-Fe 3 O 4 NPs exhibited a broad absorption band at ≈530 nm that correspond to the absorbance band of the Au NPs 45 . The weak intensity of this broad band is consistent to the small size (<d> = 3.9 nm) of the Au particles produced 46 . In vitro experiments. A major requirement for the nanosystems to set up a feasible biomedical therapy or protocol is to display low toxicity. To assess the extent of these effects after uptake of MNPs, the toxicity of Au@   Fig. S3 in supporting material), consistent with previously reported data 22,28,47 . We mention here that an exception to the above results are connected to those MNPs with some particular NP-coatings (e.g. dextran) that yield to lysosomal incorporation. In these cases, it is well known that iron liberation from NPs and subsequent generation of reactive oxygen species (ROS) within the cell cytoplasm usually result in a significant increase of the cytotoxicity in microglial cells 48 .
A series of quantitative uptake experiments were performed by co-incubating for 24 h with increasing mass of Au@PEI-Fe 3 O 4 NPs added (from 0 to 200 μg). The results are shown in Fig. 5 indicating a linear trend of the uptake with added mass of NPs. This dependence could be fitted with a linear function = . y x 0 868 (53) , where y is the mass of NPs uptaken per cell (in pg) and x is the concentration of NPs added in μg/mL. At the highest concentration, the BV2 cells were able to incorporate 87 pg/cell after 24 h incubation, consistent with previously reported data using neuroblastoma cells (SH-SY5Y) incubated with PEI-MNPs 27 .
The surface chemistry of the particles is the main factor influencing cellular uptake, the PEI coating in our particles seems no to hinder the phagocytic activity in BV2 microglial cells, and protein adsorption on the positively charge polymer might be one of the reason for this observed behaviour 49 . The surface charge of the Au@ PEI-Fe 3 O 4 NPs were assessed by zeta potential measurements. The as prepared colloidal PEI-Fe 3 O 4 NPs in water at pH 7 showed a value of +20.5 mV, as expected for the presence of positively charged amine groups in the polymer backbone. After 24 h of incubation in cell culture medium (complete DMEM), this value changed to −11 mV due to adsorption of proteins onto PEI-Fe 3 O 4 NPs, in agreement with previous reports 3 . After gold coating, the surface charge of Au@PEI-Fe 3 O 4 NPs showed a zeta potential of −25 mV in water due to the carboxyl groups of citrate-adsorbed molecules and this value dropped to −12 mV, after the incubation in DMEM cell medium.
Regarding the final distribution of the particles after incubation, the analysis using FIB-SEM dual beam microscopy showed large amounts of NPs attached to the cell membrane (Fig. 6) for both types of NPs, forming large (~2-5 μm) agglomerates. We did not find any noticeable morphological or adherence changes in the cells before and after incubation with Au@PEI-Fe 3 O 4 NPs (Fig. 6a). The analysis of the cell cross-sections confirmed the presence of NPs at the intracellular space with the same kind of agglomeration observed at the cell membrane (see Fig. 6c,d). The existences of MNPs were confirmed by EDX spectroscopy through detection of Fe and Au signatures from the cross sections of the intracellular aggregates (see Fig. 6e,g).  ± . y x (0 868 0 053) . Dotted lines represent the 95% confidence interval.
www.nature.com/scientificreports www.nature.com/scientificreports/ the λ > 2 values measured in all cases to the non-linearity of the initial magnetization with H 0 that precludes the validity of LRT for the present experimental conditions. We note that for a given set of (f, H) parameters, the SPA values measured in vitro were systematically lower than for the as prepared colloids. Since the inhibition of particle rotation (Brown relaxation) in vitro due to the high intracellular viscosity was also present for the PVA-fixed as prepared NPs, the lower value from in vitro experiments is most likely originated in the dipolar interactions within the NP agglomerates observed from FIB-SEM Dual Beam images. The dipolar interactions within clusters change the magnetic relaxation dynamics 50 . www.nature.com/scientificreports www.nature.com/scientificreports/ It has been demonstrated by numerical calculations that the dipolar interactions in three-dimensional agglomerates can considerably reduce the SPA in densely-packed clusters. Moreover, in this case the optimal particle size for maximum SPA is shifted towards lower values compared to isolated NPs. It remains to be investigated whether these SPA values can be improved by tuning the particle size and the Au coating for the best use of these NPs as nanoheaters.

Conclusions
We have obtained Au@PEI-Fe 3 O 4 NPs by a simple two-step reaction in aqueous medium, with good performance as nanoheaters for magnetic hyperthermia. These particles have very low in vitro cytotoxicity, and provided an interesting multifunctional nanoplatform for bimodal application of light and magnetic hyperthermia. It has been demonstrated that the behaviour of SPA with applied field H 0 is governed only by the properties of the magnetic cores, being experimentally identical for blocked NPs within solid matrix and/or within the intracellular space. However, the natural agglomeration occurring in cells yield dipolar interactions between NPs to decreases the effective SPA and obliges to recalibrate the optimal particle sizes for maximum heat efficiency.
Transmission electron microscopy (TEM). Average size, distribution and morphology were analysed by transmission electron microscopy (TEM) using a FEI Tecnai T20 microscope, operated at 200 keV. The average particle size (<d>) and size distribution was calculated from histograms after counting N > 500 particles of both Fe 3 O 4 cores and Au NPs. The data could be fitted with a lognormal distribution. High resolution transmission electron microscopy (HRTEM images were taken using a FEI Tecnai F30 microscope, operated at an acceleration voltage of 300 KV. The microscope was equipped with a HAADF (high angle annular dark field) detector for the STEM mode and EDX (X-ray energy disperse spectrometry) pattern was also studied. Lattice fringes were measured from the fast-Fourier transform of HRTEM images, using Gatan Digital Micrograph. Samples of NPs were prepared by placing one drop of a dilute suspension of NPs in ethanol on a carbon-coated copper grid and evaporating the solvent at room temperature. HRTEM images were used for studied the morphology, grain size and structural information of our samples.
X-ray diffraction (XRD) measurement. XRD patterns were obtained using a Rigaku Miniflex 600 diffractometer operating at 30 mA and 40 kV from 20 to 80° (2θ value) using Cu K-α radiation (0.15418 nm). The samples were prepared placing a drop of a concentrated NPs suspension on a zero diffraction silicon wafer. Rietveld method analysis was used to confirm the structural analysis of NPs.
Magnetic measurements. Magnetic properties were determined in dry samples (with nitrogen flow) using a Superconducting Quantum Interference Device (SQUID). Zero-field-cooled (ZFC) and field-cooled (FC) www.nature.com/scientificreports www.nature.com/scientificreports/ curves were measured between 2 to 300 K, with cooling field H FC = 2.39 kA/m. Magnetization as a function of the field was measured at 5 and 300 K in applied fields up to ± 5570 kA/m. Saturation magnetization (Ms) was obtained by extrapolating to infinite field the experimental results obtained in the high range where magnetization linearly increases with 1/H. Values of the magnetic moment were normalized using the mass of the magnetic core of the Au@PEI-Fe 3 O 4 NPs. The concentration of Fe and Au was determined by elemental analysis performed using Inductively Coupled Plasma (ICP) technique.
Specific Power Absorption (SPA) measurements. The parameter to characterize the heating power capacity of our samples is the specific power absorption (SPA), also labelled as specific absorption rate (SAR) or specific loss power (SLP). SPA is described using the expression 51 , where C Liq and δ Liq are the specific heat capacity and density of the solvent carrier, respectively, φ is the mass concentration of the nanoparticles in mg/mL, and ∆T/∆t is the heating rate of the sample during the experiment. In this work the SPA measurements were performed in a commercial magnetic field applicator (nB Nanoscale Biomagnetic S.L., Spain) in a vacuum-insulated Dewar connected to a vacuum pump (10 −7 mbar) and a fibre optic-based thermometer probe placed at the centre of the sample to determine its temperature. To simulate the high-viscosity conditions of the intracellular medium the nanoparticles were dispersed in a PVA polymeric matrix (10% w/w), resulting in final concentrations of 3.

UV-vis spectrophotometry (UV-vis).
UV-vis absorption spectra of the produced nanoparticles were recorded by two spectrophotometers: 1) Thermo Scientific Evolution 220 Diode Array and 2) Jasco (V670). The sample was measured diluted in a 1 mL water solution in a standard quartz cuvette used to quantify the light that is absorbed and scattered by sample. Concentration of Fe in PEI-Fe 3 O 4 and Au@PEI-Fe 3 O 4 NPs was determined by UV-vis spectrophotometry (Shymadzu UV-160) using thiocyanate complexation according to the protocol published elsewhere 21,52,53 . Zeta potential measurements. The Zeta potential were measured using a Zetasizer Nano ZS90 (Malvern instruments) with a He-Ne laser 633 nm working with a scattering angle of 90°. All samples were measured dispersed on supplemented culture media and room temperature and data were obtained using a monomodal acquisition.

Cell culture and viability tests. BV2 cells from a murine microglial cell line were cultured in DMEM for
in vitro studies and maintained at 37 °C 5% CO 2 and 95% relative humidity. For the cell viability assays, BV2 cells were seeded and incubated into a six-well culture plate (25 × 10 4 cell/well) for 24 h at 37 °C with 5% CO 2 . The medium was replaced with fresh media containing increasing concentrations of Au@PEI-Fe 3 O 4 NPs (0, 10, 25, 50, 75 and 100 μg/mL), and incubated overnight. After incubation the medium was removed and the cells were washed twice with PBS. The cells were detached using trypsin and re-suspended in 1 mL of fresh media. Trypan blue was added in equal volume of cell samples. All experiments were conducted in triplicate. Cellular uptake test. BV2 cells were planted into six-well plates (25 × 10 4 cells/well) in a volume of 2 mL.
Then the growth media was replaced by medium with increasing amounts of Au@PEI-Fe 3 O 4 NPs (0, 10, 25, 50, 75 and 100 μg/mL) and incubated for 24 h. The cells were washed with PBS twice times, harvested by trypsinization and suspended in 1 mL of DMEM to count. The pellet precipitated was digested with an acid solution (HCl 6 M and HNO 3 65%, 1:1) to quantify the amount of Fe by UV-vis spectrophotometry using the protocol described above.

Dual Beam (FIB-SEM) analysis. The intracellular distribution of Au@PEI-Fe 3 O 4 NPs in BV2 cells was stud-
ied using a Dual-Beam FIB/SEM analysis (Nova 200 NanoLab, FEI Company) SEM images were taken at 5 and 30 kV with a field emission gun column, and a combined Ga-based 30 kV (10 pA) ion beam to cross-sectioning single cells. This study was complemented by energy-dispersive x-ray spectroscopy (EDX) for chemical analysis. The preparation of the samples was made by seeding BV2 cells on a sterile glass coverslip at a density of 1 × 10 4 cells/well in 0.5 mL of culture media for 24 hours at 37 °C. After 24 h, the growth medium was replaced with the fresh medium with at a concentration of 100 μg/mL of Au@PEI-Fe 3 O 4 NPs. After overnight incubation, the cells were washed two times with PBS and fixed with 4% glutaraldehyde solution for 2 hours. After that the coverslips were washed three times with cacodylate buffer (pH 7.2), and then treated with 1% osmium tetroxide and 2.5% potassium ferrocyanate. After being washed, the samples were gradually dehydrated at room temperature via immersion in increasing concentrations of methanol 30% (x2), 50% (x2), 70% (x2), 90% (x2), and 100%. Finally, the samples were coated with gold for FIB-SEM imaging.