Microwave-assisted synthesis of highly crystalline, multifunctional iron oxide nanocomposites for imaging applications †

We report a reproducible single-step, microwave-assisted approach for the preparation of multifunctional magnetic nanocomposites comprising superparamagnetic iron oxide (Fe 3 O 4 ) cores, a polyelectrolyte stabilizer and an organic dye with no requirement for post-processing. The stabilisers poly(sodium 4-styrenesulfonate) (PSSS) and sodium polyphosphate (SPP) have been thoroughly investigated and from analysis using electron microscopy, dynamic light scattering measurements, magnetic hysteresis and magnetic resonance (MR) imaging, we show that the higher degree of Fe 3 O 4 nanoparticle crystallinity achieved with the PSSS stabiliser leads to enhanced magnetic behaviour and thus better contrast agent relaxivity compared to the less crystalline, poorly de ﬁ ned particles obtained when SPP is employed as a stabiliser. We also demonstrate the potential for obtaining a multifunctional magnetic- ﬂ uorescent nanocomposite using our microwave-assisted synthesis. In this manner, we demonstrate the intimate link between synthetic methodology (microwave heating with a polyelectrolyte stabilizer) and the resulting properties (particle size, shape, and magnetism) and how this underpins the functionality of the resulting nanocomposites as agents for biomedical imaging.


Introduction
The choice of synthetic approach employed for the preparation of nanoparticles is of crucial importance when designing materials for a specic function. Of growing recent interest has been the development of routes to nanoparticles which afford great control over particle shape and composition, vital in our efforts to realise an intimate understanding of their unique properties. One example is the eld of magnetic nanoparticles, a class of materials whose size-dependent magnetic properties opens up their potential applications for hyperthermic cancer therapy, site-specic drug delivery and contrast enhancement in magnetic resonance (MR) imaging. [1][2][3][4][5][6][7][8][9][10][11] Control over the magnetic properties is desirable in order to tailor the candidate nanoparticle for a specic biomedical application. 12 Magnetic nanoparticles may be synthesised in a variety of ways, [13][14][15] including by co-precipitation, 16,17 hydrothermal methods, 18 and the decomposition of precursors at elevated temperatures. 19,20 Aqueous routes, such as co-precipitation, are advantageous in that the particles are prepared in biologically tolerated solvents, but oen lead to polydisperse nanoparticles which may display some loss of particle crystallinity. 21 Advances made in the high temperature decomposition of organometallic precursors have led to highly crystalline, monodisperse nanoparticles, with a great degree of control over particle size. 22 These routes, where organic solvents are employed, generate hydrophobic nanoparticles, with additional work-up required to transfer the particles to aqueous environments for subsequent biomedical use. A synthetic approach which produces nanoparticles with high crystallinity without the need for surface post-processing to induce hydrophilicity is therefore highly desirable.
With these concerns in mind, we have employed microwaveassisted methods for the preparation of functionalised iron oxide nanoparticles, which are becoming increasingly attractive in materials chemistry. Microwave approaches to particle synthesis have led to dramatic decreases in reaction times and greater control over product formation. [23][24][25][26][27][28][29] For example, the Niederberger group has developed microwave-assisted methods to prepare a range of nanoparticles of controlled sizes within minutes. 30 Since the rst reports of the microwave-promoted hydrothermal synthesis of sub-micron haematite powders, 31 methods have been developed to prepare nanoparticles of uniform size distribution. Nanoparticles of haematite (a-Fe 2 O 3 ) have been reported from the irradiation of hydrolysed iron salts, 32 while more exotic a-Fe 2 O 3 morphologies (cubes, rings and spindles) have been realised by Yu and co-workers by tailoring the reaction conditions to provide thermodynamic control over particle growth. 33,34 Superparamagnetic maghemite (g-Fe 2 O 3 ) nanoparticles may be prepared through a microwavetreated co-precipitation reaction with a typical particle size of 10 nm obtained. 35 Increases in nanoparticle crystallinity have been observed upon aging under microwave conditions, with postprocessing using stearic acid resulting in stable organic ferro-uids. 36 Uptake in endothelial cells of superparamagnetic citrate-coated Fe 2 O 3 nanoparticles prepared by microwaveassisted methods have also recently been reported. 37 Variation of reactant concentration in the microemulsion synthesis of Fe 3 O 4 under microwave irradiation has resulted in monodisperse composites of variable sizes. 38 Reaction time for the polyol synthesis of cobalt ferrite nanoparticles has been decreased to just one hour by combining microwave heating, 39 while a variety of MFe 2 O 4 ferrites (M ¼ Zn, Ni, Mn, Co) have also been reported from microwave-hydrothermal methods. 40 For their use as imaging agents, tailoring the surface chemistry of magnetic nanoparticles is essential. 22,[41][42][43][44] For example, Liong et al. have reported the preparation of a multifunctional nanocomposite for imaging and targeted drug delivery where iron oxide nanoparticles are phase transferred from an organic to an aqueous solution before silica coating and co-condensation of a uorescent moiety render them multifunctional. 45 Recently, Liu et al. have reported the use of polyacrylic acid in the synthesis of Fe 3 O 4 nanoparticles to achieve tuneable particle cluster sizes of between 100 nm and 400 nm, which display good biocompatability. 46 We have previously reported the use of polyelectrolytes for the in situ stabilisation of magnetic nanoparticles during a co-precipitation reaction, which has led to new developments in magnetic uid preparation. [47][48][49] Employing polyelectrolyte stabilisers in situ during nanoparticle nucleation and growth confers a high degree of stability induced by strong affinity of the anionic groups to the metal cations, with stable aqueous suspensions in the presence of a 0.5 T magnetic eld obtained.
Here, we report the reproducible, repeatable preparation of bare Fe 3 O 4 nanoparticles and the rst use of microwave synthesis to obtain multi-coordinating polyelectrolytestabilised Fe 3 O 4 nanoparticles and polyelectrolyte-stabilised magnetic uids functionalised with a uorescent dye according to Fig. 1. We are particularly interested in what effect, if any, the choice of polyelectrolyte stabiliser has on the resulting nanoparticle shape, size and crystallinity. The polyelectrolytes used were poly(sodium 4-styrenesulfonate) (PSSS), which has been employed previously as a stabiliser in a traditional coprecipitation reaction, 48 and sodium polyphosphate (SPP) which was employed as an in situ stabiliser for the rst time, where the sulfonate and phosphate groups, respectively, bind to the iron ions in solution prior to particle precipitation. In this manner, the polyelectrolytes act as stabilisers preventing the further growth of the nanoparticles, while at the same time promoting colloidal stability of the particles in water.

Results and discussion
A modied co-precipitation technique has been employed, wherein a precursor solution of ferric and ferrous chlorides in the presence of polyelectrolyte solution was rst prepared. The co-precipitation method involves several processes: nucleation, seed formation and growth. 50,51 Rapid particle nucleation follows the addition of ammonia base, aer which the particle suspension is transferred to a microwave cavity and particle growth occurs under microwave irradiation at 150 C for 20 min. All samples were washed until neutral. In the case of the bare particles, the sample was dried for further analysis. For the polyelectrolyte samples, the nal, neutral h washings were highly stable colloidal suspensions, which have been characterised using dynamic light scattering (DLS), transmission electron microscopy (TEM), and MR imaging, and their cellular interactions were analysed. The remaining solids were dried and analysed using X-ray diffraction (XRD), X-ray absorption spectroscopy (XAS), SQUID magnetometry, thermogravimetric analysis (TGA) and IR.
XRD patterns collected on dried powders of each sample are shown in Fig. 2(a). For comparison, a diffraction pattern was also collected of magnetite nanoparticles prepared using the traditional co-precipitation route without any microwave treatment (red line). The patterns obtained for all samples may be indexed to the cubic spinel, magnetite (Fe 3 O 4 ). The broad peaks observed are typical for nanoparticles and the particle sizes, which may be obtained from the Scherrer equation, are included in Table 1. These are on the order of 10 nm. Also shown in Fig. 2   that the XRD patterns for magnetite and maghemite are similar, Raman spectroscopy measurements were performed. Using Raman spectroscopy, it is possible to assign characteristic peaks for magnetite (which contains both Fe 3+ and Fe 2+ ions) and the oxidised iron oxides maghemite (g-Fe 2 O 3 ) and haematite (a-Fe 2 O 3 ). Fig. 2(b) shows the resulting Raman spectrum for the pure uncoated nanoparticles, displaying a strong peak at 664 cm À1 which can be attributed to the A 1g mode of magnetite. Also visible are the E g and T 2g modes, centered at 301 and 527 cm À1 respectively. All peaks observed are in excellent agreement with previous assignments for magnetite in the literature. 4,52,53 There are no peaks for haematite or maghemite observed.
The comparatively broader peak shapes for the Fe 3 O 4 -SPP sample are reected in the decreased particle size values obtained from the Scherrer equation for these particles (10.9 nm for SPP-coated particles, 11.2 nm for uncoated nanoparticles). HRTEM images for pure Fe 3 O 4 samples reveal single-crystal nanoparticles (see Fig. 3). The Fe 3 O 4 particles are aggregated, with a typical particle size of 12 AE 2 nm (measured for N ¼ 100 particles). Analysis of lattice spacing and SAED patterns, shown in Fig. 3 Interestingly, the nanoparticle shape, size and aggregate nature are signicantly affected by the polyelectrolyte stabilisers employed under otherwise identical reaction conditions. Electron microscopy images of the polyelectrolyte-stabilised samples are shown in Fig. 4, where the presence of the PSSSpolyelectrolyte is evidenced by the core-shell appearance of the nanocomposite ( Fig. 4(a)). The presence of polyelectrolyte on the PSSS-and SPP-Fe 3 O 4 nanoparticle surface has also been conrmed using IR spectroscopy (see ESI Fig. S1 †). Both samples give an Fe-O stretch at $530 cm À1 and a broad stretch at 3400 cm À1 for physically adsorbed water on the particle surface. In the case of PSSS-Fe 3 O 4 ,anFe-O-S stretch is noted at 669 cm À1 , which indicates that the mode of binding is through the sulfonate group to the surface iron atoms of the nanoparticles. There are also stretches noted for the sulfonate groups at 775, 830, 1115, 1160 and 1405 cm À1 . In the case of the SPP-Fe 3 O 4 sample, an Fe-O-P stretch is observed at 992 cm À1 , indicating that the mode of binding is via the phosphate groups to the surface iron atoms. Stretches relating to the phosphate groups are also observed at 869 and 1255 cm À1 . TGA plots for Fig. 2 (a) Powder XRD patterns for samples prepared using microwave-assisted methods: bare magnetite (orange), PSSS-Fe 3 O 4 (navy) and SPP-Fe 3 O 4 (purple). For comparison, a pattern of magnetite nanoparticles prepared using traditional co-precipitation synthesis is included (red) together with a reference ICSD pattern. (b) Raman spectrum of pure Fe 3 O 4 nanoparticles prepared using microwaveassisted method. Peak at 664 cm À1 is assigned to A 1g mode of magnetite, with E g and T 2g modes centered at 301 and 527 cm À1 respectively. No peaks for haematite or maghemite were observed.  In order to probe the behaviour of these polyelectrolyte nanocomposites in the presence of a magnetic eld, samples were dried in a 0.5 T magnetic eld and analysed using microscopy. In the case of the PSSS-stabilised nanoparticles, linear assemblies are observed (Fig. 4(c)); reminiscent of previous reports for samples prepared by co-precipitation alone. 48 Currently, we are focussing attention on understanding how the nature of the polyelectrolyte and the chain length affect the formation of these linear assemblies and the resulting implications on imaging behaviour of nanoparticulate contrast agents.
The aggregation of these stabilised nanoparticles has been investigated further by measuring the average hydrodynamic radii at 298 K using DLS (ESI Fig. S2(b) †). For both samples and for a range of reproduced reactions, this average is found to be close to 100 nm. The polydispersity of the SPP-Fe 3 O 4 sample is greater than that for the PSSS-Fe 3 O 4 sample, which is evidenced by the broader peak shape and the slightly higher polydispersity index (PDI) (0.158 for SPP-Fe 3 O 4 and 0.097 for PSSS-Fe 3 O 4 ). The low values obtained for the PDI values (<0.2) indicate a unimodal distribution of monodisperse clusters for both samples. Upon exposure to a 0.5 T magnetic eld these values do not change, conrming the excellent water stability these polyelectrolyte stabilisers confer on the Fe 3 O 4 nanoparticles prepared using microwave methods. Table 1 summarises the particle sizes as calculated from Scherrer broadening (XRD), average particle size (TEM), hydrodynamic radius and zeta potential (DLS). The observed zeta potentials are below À30 mV for both samples, which conrms the negative surface charge we postulate in Fig. 1 and reaffirms the excellent water stability of these suspensions, which was the case over a six month period as conrmed by DLS analysis. 54 Magnetization curves were measured at 300 K and 10 K in a magnetic eld of up to 2 Â 10 4 G and are shown in Fig. 5. There is negligible coercivity and remanence noted at 300 K, indicative of superparamagnetic, single-domain iron oxide particles. The magnetization is unsaturated up to 2 T, even at 10 K. Uncoated magnetite prepared using our microwave-assisted method gives a saturation magnetisation of 67.6 emu g À1 at 300 K. While this is lower than the theoretical value for bulk magnetite of 98 emu g À1 (most likely due to spin disorder on the particle surface 55 ), this value is higher than previously reported saturation magnetisation values for aqueous routes to iron oxide nanoparticles (40-50 emu g À1 ). 56,57 By employing polyelectrolytes as stabilisers we have found that primary particle sizes and morphologies change and this, in turn, has a marked effect on the resulting magnetic properties. The SPP-stabilised particles are smaller in size, with ill-dened shape, and this is reected in the reduced M s value of 53.9 emu g À1 at 300 K. The PSSS-stabilised samples, on the other hand, have an M s value of 77.1 emu g À1 at 300 K and appear highly crystalline in HRTEM. Interestingly, the magnetisation value here is signicantly higher than M s values previously obtained from NMRD data for PSSS-stabilised Fe 3 O 4 (30-50 emu g À1 ) prepared without the additional microwave synthesis step. 58 The saturation magnetisation value is greatly affected by the crystallinity of the sample. For high temperature decomposition routes using organic surfactants, for example, the highly crystalline and uniform nature of the particles is manifested in similarly higher M s values ($80 emu g À1 ). 59 For biological use, organic-surfactant coated particles require further postprocessing to transfer into aqueous solutions to obtain stable  suspensions. 26,60 The microwave-assisted synthesis reported here achieves both properties in a single step: excellent magnetic properties combined with long-term aqueous stability. From these data, we have shown the combination of the PSSS stabiliser and microwave irradiation results in a waterstable crystalline material, without compromising the magnetic properties.
To evaluate the MR efficacy of the PSSS-Fe 3 O 4 and SPP-Fe 3 O 4 samples, T 1 , T 2 and T * 2 -weighted images were collected at varying Fe concentrations using a 3 T clinical MRI scanner. Relaxivity values are reported in Table 2. Increasing Fe concentrations ($0-2m Mi nH 2 O) were imaged and are displayed in Fig. 6. Higher relaxation rates are noted upon increasing Fe concentration, because PSSS-Fe 3 O 4 and SPP-Fe 3 O 4 shorten the T 2 relaxation time, which reduces the signal intensity, i.e. negative contrast. Interestingly, the contrast displayed by the PSSS-Fe 3 O 4 sample is greater than that of the SPP-Fe 3 O 4 particles. The increased crystallinity of these particles and their enhanced magnetic properties are the major factors for this behaviour. Since it is the magnetic moments of the particles interacting with the water protons that result in image contrast, the crystalline nature of the particles is vital in determining how efficiently this may occur. In the case of the SPP-stabilised nanoparticles, the irregular shape and decrease in saturation magnetization point to greater disorder of the surface spins in these particles which will play a role in decreasing the imaging efficacy. This is in good agreement with an extensive study reported by Vuong et al., who have proposed a method for predicting T 2 relaxation based on the nanocrystal size and magnetization values. 61 For our PSSS-Fe 3 O 4 particles, the excellent magnetic properties and the crystalline nature of the primary particle indicate the promising potential for these contrast agents for MR imaging. MR contrast properties for several other reports of functionalised iron oxide nanoparticles are also included in Table 2. 48,62,63 Enhanced MR contrast has been previously noted for linear assemblies of iron oxide nanoparticles. 48,58,64 In the current case, we observe relaxivities on the order of commercially available contrast agents and nanoparticle suspensions prepared by high temperature decomposition routes, with the added advantage here of no additional work-up required to transfer to aqueous conditions. [65][66][67] Recently, Zboril and co-workers reported extraordinarily high relaxivities of 735 mM s À1 for iron oxide nanoparticles with a 1 nm terephthalic acid coating, postulating that effective spin-transfer to surrounding water protons is mediated via p-conjugation pathways through the organic surfactant. 68 It is enticing to consider the combined use of active spin-transfer surfactants with microwave processing to further optimise relaxivity behaviour.
The effect of stabiliser on cell toxicity was examined using cell viability studies on suspensions of the polyelectrolytestabilised particles co-incubated at increasing concentrations (0.001-1000 mgL À1 ) with a range of cell lines and these results are depicted in Fig. 7. Regardless of the stabiliser employed, all nanoparticle suspensions tested were found to be non-toxic to a variety of mammalian cell lines, demonstrating the non-toxic robustness of these materials. Iron concentration ranges were similar to previous reports. 69 This is determined by the fact they did not affect the viability of UKF-NB-3 neuroblastoma cells, primary human retinal pigment epithelial (RPE) cells, or primary human foreskin broblasts (HFF) even in concentrations up to 1 mg of iron per L.
To demonstrate uptake into mammalian cells, uorescentlylabelled samples were prepared by the addition of Rhodamine B  to the PSSS polyelectrolyte solution before particle precipitation, where association is driven by the electrostatic interactions between the dye and the polyelectrolyte. The uorescent nature of the nanocomposite was conrmed using uorescence spectroscopy (l ex ¼ 545 nm) and an enhanced aggregation of the composite is noted in DLS, where electrostatic interactions drive the formation of larger cluster sizes and this is reected by the larger Z-ave (108.2 nm; PDI 0.169) (see ESI Fig. S4 †). The sample spent several days under a 0.5 T magnetic eld before DLS measurement to remove any larger aggregates and to ensure a stable suspension remained. Removal of any excess dye was conrmed by continuous washing of the sample and evaluation using uorescence spectroscopy before redispersion into water and also by dialysis experiments of the resulting nanocomposite (see ESI Fig. S5 †). Fig. 8 shows the resulting nanoparticle uptake by UKF-NB-3 neuroblastoma cells, which can be clearly located throughout the cytoplasm. Of importance here is that this internalisation process had no effect on viability as described above. Complementary Z-scan images conrm the location of the functionalised nanoparticles throughout the cell cytoplasm (Fig. 8(c)), rather than on the cell surface.

Conclusions
In conclusion, we have demonstrated a reproducible and reliable single step route to multifunctional magnetic nanocomposite materials, using a microwave-assisted synthesis, in order to provide stable magnetic uids for use as MR contrast agents. Here, the underpinning structure-property relationship in these materials is highlighted by the synthetic approach taken and the resulting outcome this has on the functional properties of the nanocomposites. The choice of stabilizerpolysulfate versus polyphosphate -has a substantial effect on the magnetic properties, which translates to their ability to enhance the relaxation mechanism of surrounding water protons in MR imaging. Whilst not the highest relaxivities reported for iron oxide nanocomposites, the comprehensive analysis presented here using a full range of characterisation techniques demonstrates PSSS as an excellent stabiliser for the preparation of multifunctional magnetic nanocomposite materials and our microwave-assisted synthetic approach can be applied to a full range of ferrite-based nanoparticles and polyelectrolyte-stabilised systems. Previous reports have shown that effective contrast agents can be obtained using PSSS as a stabiliser for iron oxide nanoparticles without any microwave treatment. 48,58 In the case of the nanocomposites prepared without microwave heating, the hydrodynamic radius and polydispersity index is greater (136 nm and 0.21, respectively). Most interesting though is that the saturation magnetisation (obtained from NMR dispersion measurements) is consistently lower for samples prepared without the application of microwave heating (30-50 emu g À1 ). This has a resulting implication on the contrast agent efficacy (values of r 1 and r 2 obtained are of the same order for both samples, but for the sample prepared without microwave treatment, the relaxation properties were measured only at 1.5 T). Therefore, the addition of this microwave heating step in the case of the PSSS stabiliser positively affects the subsequent properties and functionality of the nanocomposite. The direct link between particle crystallinity and the resulting magnetic behaviour and the governing effect this has on MR imaging capability is particularly interesting. We show that these nanocomposites are non-toxic to a range of mammalian cells, where their uptake can be confocally imaged. The ease of this approach allows for the preparation of extremely stable magnetic uids for combined MR contrast efficacy and optical imaging and paves the way for a synthetic methodology which allows for greater control over nal functionality.

Materials
All materials were obtained from Sigma Aldrich. Millipore water was deoxygenated by boiling and then cooling under nitrogen gas. A CEM Discover SP system was used for microwave heating. FTIR spectra (400-2000 cm À1 ) were recorded using a Shimadzu  IR Affinity-1 spectrophotometer. X-ray diffraction was performed on a Bruker D8 X-ray diffractometer and powder patterns were analysed using Rietveld renement as embodied in the Fullprof suite. Thermogravimetric analysis was obtained with a Netzsch STA 409 PC Luxx TGA machine. All samples were heated in air to 700 C. Dynamic light scattering measurements were performed using a Malvern Zetasizer nano-ZS. Transmission electron microscopy (TEM) images were taken on a JEOL JEM (200-FX) operating at 120 kV. Samples were prepared on a formvar coated copper grid. Some grids were dried over a 0.5 T magnet. High resolution TEM and selected area electron diffraction (SAED) were performed on an FEI Tecnai TF20 instrument tted with a eld emission gun, operated at 200 keV. TEM samples were prepared by dispersing the sample in deionised water and dropping the solution onto an amorphous holey carbon coated grid. TEM data were obtained and processed using either Digital Micrograph or IMAGEJ 1.41 soware.

Preparation of magnetic nanocomposites
FeCl 3 $6H 2 O (2.70 g; 10 mmol) and FeCl 2 $4H 2 O (0.99 g; 5 mmol) were dissolved in 10 mL of deoxygenated water. Polyelectrolyte stabiliser (0.2 g of either PSSS or SPP) was dissolved in the iron solution. The solution was heated to 80 C. Ammonia solution (10 mL; 28-30%) was injected at a rate of approximately 2.5 mL s À1 and the solution was stirred for 20 minutes before transferring to the microwave cavity to be heated at 150 C for 20 minutes. For uorescently-labelled samples, Rhodamine B (2 Â 10 À4 g; 10 mmol) was dissolved in 2 mL of deoxygenated water and the PSSS polyelectrolyte (0.2 g) was added to this and stirred for two hours. This was transferred to 10 mL of Fe 3+ /Fe 2+ solution (2.70 g; 10 mmol, and 0.99 g; 5 mmol respectively), before addition of 10 mL of ammonia solution at 80 C. The resulting black precipitates were washed with Millipore water (5 Â 20 mL) with the nal two washings being used in TEM, DLS, FTIR, MRI and confocal measurements. The solid precipitate was analysed with XRD, TGA, HRTEM and SQUID measurements.

MR imaging
The tubes of different PSSS-Fe 3 O 4 and SPP-Fe 3 O 4 concentrations were placed in a rack in the centre of the magnet. MR imaging was performed with a standard extremity ex coil on a clinical 3 T MRI scanner (Achieva, Philips Healthcare, Best, The Netherlands). T 2 was determined with a 2D multi-spin-echo sequence (FOV ¼ 120 Â 120 mm 2 , matrix ¼ 432 Â 432, measured slice thickness ¼ 3 mm, echo train length ¼ 5, TE ¼ 10 ms, TR ¼ 725 ms, ip angle ¼ 90 ). The acquired imaging data was transferred to a computer running Matlab and analysed using an in-house Matlab tool to receive the relaxation time T 2 for each Fe concentration. Excel was used to plot the relaxation rate R 2 over the concentration and the relaxivity value was determined using linear regression. Iron concentrations of all MRI scanned serial dilutions of SPIONs were determined using Inductively Coupled Plasma Mass Spectrometry (ICP-MS). Briey, samples were digested in 70% nitric acid overnight at room temperature, followed by dilution in deionized water. A standard curve was acquired with each sample set for iron concentration determination.

Investigating the effect of nanoparticles on cell viability
The effects of the nanoparticles on cell viability were determined in UKF-NB-3 neuroblastoma cells, primary human retinal pigment epithelial cells, and primary human foreskin broblast cells. UKF-NB-3 cells were cultivated in IMDM supplemented with 10% foetal calf serum (FCS), 100 IU mL À1 penicillin, and 100 mg mL À1 streptomycin. Retinal pigment epithelial cells were cultured in DMEM supplemented with 20% FCS, 100 IU mL À1 penicillin, and 100 mg mL À1 streptomycin. Fibroblasts were cultured in DMEM supplemented with 10% FCS, 100 IU mL À1 penicillin, and 100 mg mL À1 streptomycin. All cells were cultivated at 37 C in humidied 5% CO 2 atmosphere as previously described. 70,71 Cell viability upon the addition of nanoparticle preparations at different concentrations was determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) dye reduction assay aer 120 h of incubation as described previously. 70,71 4.5 Determination of cellular location of nanoparticles using uorescence microscopy UKF-NB3 cells were seeded at 2 Â 10 5 cells per well in a 24 well plate, containing a coverslip in each well, and grown at 37 C for 48 hours. Rhodamine conjugated nanoparticles were diluted to 0.099 mg L À1 iron concentration in media, 1 mL added to appropriate wells and incubated for 3 hours before aspirating and xing with methanol at À20 C for 5 minutes. Coverslips were mounted using mowiol and anti-fade and slides examined using a Leica confocal laser scanning (TCS 4; Â63 oil lens) microscope.