Biomaterials 35 (2014) 6400e6411
Contents lists available at ScienceDirect
Biomaterials journal homepage: www.elsevier.com/locate/biomaterials
Magnetic hyperthermia efficiency in the cellular environment for different nanoparticle designs Riccardo Di Corato a, Ana Espinosa a, Lenaic Lartigue a, Mickael Tharaud b, Sophie Chat c, Teresa Pellegrino d, Christine Ménager e, Florence Gazeau a, Claire Wilhelm a, * a
Laboratoire Matière et Systèmes Complexes, UMR 7057, CNRS and Université Paris Diderot, 75205 Paris Cedex 13, France Institut de Physique du Globe de Paris, UMR 7154, CNRS and Université Paris Diderot, 75205 Paris Cedex 13 France INRA, UR1196 GPL, MIMA2- Plateau de Microscopie Electronique 78352 Jouy-en-Josas, France d Istituto Italiano di Tecnologia, I-16163 Genova, Italy e Laboratoire Physicochimie des Electrolytes, Colloïdes et Sciences Analytiques PECSA UMR 7195, Université Pierre et Marie Curie UPMC-CNRS, 75252 Paris Cedex 05, France b c
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 January 2014 Accepted 11 April 2014 Available online 9 May 2014
Magnetic hyperthermia mediated by magnetic nanomaterials is one promising antitumoral nanotherapy, particularly for its ability to remotely destroy deep tumors. More and more new nanomaterials are being developed for this purpose, with improved heat-generating properties in solution. However, although the ultimate target of these treatments is the tumor cell, the heating efficiency, and the underlying mechanisms, are rarely studied in the cellular environment. Here we attempt to fill this gap by making systematic measurements of both hyperthermia and magnetism in controlled cell environments, using a wide range of nanomaterials. In particular, we report a systematic fall in the heating efficiency for nanomaterials associated with tumour cells. Real-time measurements showed that this loss of heatgenerating power occurred very rapidly, within a matter of minutes. The fall in heating correlated with the magnetic characterization of the samples, demonstrating a complete inhibition of the Brownian relaxation in cellular conditions. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: Nanomagnetism Nanomedicine Magnetic hyperthermia Nanoparticles Cell interactions
1. Introduction Nanomaterials, thanks to their minute size, can cross biological barriers and enter the cells, where they can be activated on demand. Among them, magnetic nanoparticles are particularly promising because they can be activated at a distance by a remote magnetic field, and thus serve as tracers for medical imaging [1], vectors for drug targeting [2], intracellular nanorobots [3], or local heat sources to destroy malignant cells [4]. To achieve such magnetic hyperthermia, an alternating magnetic field (working at a frequency of several hundreds of kHz) is used to oscillate the magnetic moment of each nanoparticle, transforming magnetic energy into heat. The resulting local hyperthermia can damage cancer cells, either alone [5] or in conjunction with chemotherapy [6,7], or be used to promote drug release to a specific target cell or tissue [8,9]. Currently, the
* Corresponding author. E-mail address:
[email protected] (C. Wilhelm). http://dx.doi.org/10.1016/j.biomaterials.2014.04.036 0142-9612/Ó 2014 Elsevier Ltd. All rights reserved.
main limitation of magnetic hyperthermia is the poor heating efficiency of most magnetic nanoparticles, and current therapeutic approaches require large amounts of nanoparticles (in the 1 M of iron range) to be injected locally into the target tumor, in order to obtain a therapeutic effect [10]. Major efforts are therefore underway to optimize nanoparticle heating efficiency, by tuning key parameters such as size, magnetic anisotropy and saturation magnetization of nanoparticles [11,12]. Another way to increase the heating efficiency is to modify the shape or degree of interactions of nanoparticles and thereby alter their anisotropies. For example, solutions of cubed shaped nanoparticles (biogenic magnetosomes [13,14] or synthetic nanocubes [15,16]), coreeshell structures [17,18] or multi-grain assemblies (nanoflowers) [19] show very promising heat-generating potential. However, the intended therapeutic application of nanomaterials as nanosources of heat is tumor hyperthermia, which requires optimal intratumoral heat generation [20e22]. Two strategies are envisaged to deliver nanoparticles to tumor sites [23]. Direct intratumoral injection is feasible for anatomically accessible solid tumors, and this is the approach adopted for ongoing clinical trials.
R. Di Corato et al. / Biomaterials 35 (2014) 6400e6411
Its success depends on intratumoral delivery of large doses of nanoparticles, and also on spatial and temporal control, as well as synergy with other treatments such as radiotherapy or chemotherapy [24e26]. However, this approach cannot be used to treat inaccessible primary tumors or metastases. The second strategy, which involves intravenously injectable or inhalable targeted nanoparticles [27e29], delivers lower concentrations to the targeted cells, even with an optimized targeting, meaning that the chosen nanoparticles must have highly efficient heat-generating potential, at their final destination, inside tumour cells [30]. While major efforts are being devoted to the development of magnetic hyperthermia candidates, only rare studies reported hyperthermia measurements not only in solution but also in biological, and especially cellular environments. Currently, the heating efficiency of hyperthermia nanomaterials is compared by measuring the specific loss power (SLP), also known as the specific absorption rate (SAR), which provides a measure of power dissipated per unit mass of the magnetic material, in watts per gram. The SLP is routinely determined for nanomaterials in solution, and ranges from only few W/g to more than thousands W/g for the most efficient nanoparticles. What is still missing is the evaluation of the SLP in cellular media. Here we tested a broad range of nanomaterials both in solution and in a cellular environment. The particles were chosen among those holding the greatest heat ability for magnetic hyperthermia, and differed in terms of their shape (rock-like spheres, cubes, multicores), size (between 9 and 25 nm), and nature of metallic core. These parameters have different effects on the magnetic properties (superparamagnetism versus ferromagnetism, exchange coupling, dipolar interactions, magnetocrystalline anisotropy). We first designed real-time measurements of magnetic hyperthermia in living cells during the time-course of cellenanoparticle interactions. We then produced samples of fixed cells incorporating the different nanoparticles at controlled localisation (at the membrane or inside intracellular vesicles) allowing performing multiple characterizations including electron microscopy observations, hyperthermia measurements and investigations of the temperaturedependent magnetic properties. The overriding aim is to measure and to understand magnetic hyperthermia in the cellular environment, and thus to guide the development of targeted magnetic nanomaterials for tumor therapy. 2. Material and methods
6401
increased to 220 C and kept reacting for 12 h. The black sediments were separated magnetically and washed with mixture of ethanol and ethyl acetate. Then, an aqueous solution of iron(III) nitrate was added to the nanoparticles. The resulting mixture was heated to 80 C for 45 min to achieve a complete oxidation of the nanoparticles. After another treatment with 10% nitric acid, the particles were washed twice with acetone and diethyl ether and finally dispersed in water [19]. 2.1.4. Iron oxide nanocubes To synthesize nanocubes of about 19 nm in edge length, 1 mmol of iron(III) acetylacetonate and 4 mmol of decanoic acid were mixed in 25 mL of dibenzyl ether. The solution was heated to 200 C (5 C/min) and kept at this temperature for 2.5 h. The temperature was increased to reflux temperature (at a rate of 10 C/ min) and reacted for 1 h. Finally, the nanoparticles were washed three times, dispersed in 15 mL of chloroform and finally transferred in water by polymer coating with poly(maleic anhydride alt-1 octadecene), following a well-established procedure [15]. 2.1.5. Iron oxide/gold dimers To synthesize the dimers consisting of a 2 nm gold nanoparticle and 14 nm iron oxide nanoparticles we used a procedure published by Ref. [33]. Briefly, a mixture of oleic acid (1.9 mL, 90% purity), oleylamine (2 mL, 70%), 1,2-Hexadecanediol (2.58 g, 90%) and 1-Octadecene (20 mL, 90%) was heated up at 100 C and kept under vacuum for 45 min. The temperature was then raised at 120 C and 300 mL of iron pentacarbonyl (99.99% kept in glove box) were injected. Three minutes later, a mixture of Gold(III) chloride trihydrate (99.9%), kept in glove box under nitrogen, 0.5 mL of oleylamine and 5 mL of octadecene is added and the temperature was raised from 120 to 310 C in 10 min and kept at 310 C for 90 min. The flask was then cool down to room temperature and at least three washing steps were performed by precipitation with isopropanol and separation by centrifugation and re-dispersion in hexane. At the last step the sample was dissolved in 15 mL of hexane and 50 mL of oleylamine were also added. The same polymer coating procedure as for the nanocubes was used for the water transfer. 2.2. Preparation of cell samples Human adenocarcinoma SKOV-3 cells (ATCC #HTB-77) were cultured in adhesion in McCoy’s 5A modified medium (SigmaeAldrich #M9309) supplemented with penicillin (50 I.U./mL), streptomycin (50 mg/mL) and 10% fetal bovine serum. Cells were maintained at 37 C in humidified atmosphere at 5% CO2. SKOV-3 cells were cultured until confluence before the incubation with magnetic colloids. In order to obtain approximately 2$107 cell, 4 flask (150 cm2) of SKOV-3 were used for each experimental point. 2.2.1. Experiments with living cells 2$107 non-labeled cell were resuspended in 125 mL of PBS and transferred in a test tube (Vmax ¼ 500 mL). The tube was kept at 37 C and a small volume (25 mL, at 37 C) of nanoparticles was added to the pellet in order to reach an iron concentration of 25 mM. Immediately, the mixture of cells and nanoparticles were inserted in the magnetic hyperthermia coil and the temperature increase was recorded for 2 min. Afterwards, the tube was placed back in a thermostatic bath at 37 C. Hyperthermia was applied every 2 min for the first 30 min, every 10 min in the second half and finally every 15 for the second hour. At minute 120 the magnetic field was applied until the reaching of temperature plateau.
2.1. Nanoparticles syntheses 2.1.1. Iron oxide and cobalt ferrite nanoparticles These nanoparticles were synthesized by alkaline coprecipitation of FeCl2 (0.9 mol, CoCl2 for the preparation of cobalt ferrite NPs) and FeCl3 (1.5 mol) salts, according to Massart’s procedure. Superparamagnetic grains were produced by oxidizing 1.3 mol of particles with 1.3 mol of iron nitrate under boiling. Nanoparticles were washed several times with acetone and ether and suspended in water. After magnetic size-sorting, sodium citrate at a molar ratio nFe/nCit ¼ 0.13 was added to the sorted nanoparticles and the mixture was heated at 80 C for 30 min to promote absorption of citrate anions onto their surface [31]. 2.1.2. Ultra magnetic liposomes Liposomes were prepared by the reverse phase evaporation method. A mixture of DPPC/DSPC/DSPE-PEG 2000 was dissolved in diethyl ether and chloroform (3:1). Afterward above described iron oxide nanoparticles (dispersed in water) were introduced and the mixture was sonicated at room temperature for 20 min to produce a water-in-oil emulsion. Preparation was immediately transferred to a round-bottom flask and remaining organic solvent evaporated with a rotavapor at 25 C. The obtained liposome suspension was first filtered (0.4 mm) and finally magnetically sorted for removing the non-encapsulated nanoparticles [32]. 2.1.3. Iron oxide nanoflowers A solution containing 4 mmol of FeCl3 and 2 mmol of FeCl2 (in a liquid mixture of N-methyldiethanolamine and diethylene glycol) was stirred for 1 h. Afterwards, a sodium hydroxide solution (in polyols) was added to the solution of iron chlorides, and the resulting mixture was stirred for another 3 h. Then, the temperature was
2.2.2. Experiments with fixed cell pellets for complete characterization SKOV-3 cells (approximately 30 millions in 4 flasks of 150 cm2 for each sample), in adhesion, were incubated for 2 h at 37 C with the different colloids, at the following iron concentration in the extracellular medium: [Fe] ¼ 2 mM for iron oxide nanoparticles, single or in liposomes, and for cobalt ferrite nanoparticles, [Fe] ¼ 0.2 mM for iron oxide/gold dimers and iron oxide nanocubes, [Fe] ¼ 0.6 mM for iron oxide nanoflowers. At the end of the incubation, the medium was removed and cells were washed three times with culture medium, and further placed at 37 C for an additional 2 h incubation period, to ensure to total internalisation of all the nanomaterials. Cells were then trypsinized, resuspended first in PBS and washed three times in cacodylate buffer (0.1 M). Then, the cell suspensions were fixed with glutaraldehyde (2.5%) at 4 C for 60 min and washed with cacodylate buffer. For experiments at 4 C (to stop the nanoparticles at the plasma membrane), cells were incubated for 60 min in ice box, at the same iron concentrations above described. After incubation, the cells were immediately trypsinized, resuspended first in PBS and washed three times in cacodylate buffer (0.1 M). Then, the cell suspensions were fixed with glutaraldehyde (2.5%) at 4 C for 60 min and washed with cacodylate buffer. 2.3. Transmission electron microscopy (TEM) 106 cells from fixed cell suspensions were washed once with cacodylate buffer (0.1 M), fixed with glutaraldehyde (2.5%) at 4 C for 60 min and washed with cacodylate buffer. Samples were then postfixed with 1% osmium tetroxide containing 1.5% potassium cyanoferrate, gradually dehydrated in ethanol (30%e100%) and embedded in Epon. Thin sections (70 nm) were collected onto 200 mesh
6402
R. Di Corato et al. / Biomaterials 35 (2014) 6400e6411
cooper grids, and counterstained with lead citrate before examination with a Zeiss EM 902 transmission electron microscope at 80 KV (MIMA2 e plateau de MET e unité 1196 GPL e Jouy-en-Josas). Microphotographies were acquired using MegaView III CCD camera and analysed with ITEM software (Eloïse e SARL e Roissy CDG e France).
2.4. Cell magnetophoresis Magnetophoretic mobilities of magnetically labeled cells were measured in magnetic field and field gradient of 145 mT 17 T/m created by a permanent magnet. For each condition, 100 individual cells were tracked to retrieve the average magnetic load.
Table 2 Mass of iron per cell calculated from the ICP measurement of the cellular samples. Nanomaterial
mFe (pg) at the cell membrane
mFe (pg) inside the cells
gFe2O3 nanoparticles gFe2O3 nanoparticles in liposomes
6.1 14.3 3 0.8 2.1 1.1
10.2 16.5 6.7 4.7 6.3 6
CoFe2O4 nanoparticles AugFe2O3 dimers gFe2O3 nanocubes gFe2O3 nanoflowers
3. Results 2.5. Magnetic hyperthermia measurements For magnetic hyperthermia analyses, approximately 2$107 cells were resuspended in 150 mL of PBS for living cells and cacodylate buffer for fixed samples. A laboratory-made device was used. It consists of a resonant RLC circuit, using a 16 mm coil producing an alternating magnetic field with a frequency ranging from 300 kHz to 1.1 MHz and with amplitudes up to 24 kA/m. Four frequencies were chosen for the study: 320, 500, 700 and 900 kHz. Temperature was probed with a fluoroptic fiber thermometer and recorded every 0.7 s. The magnetic samples are introduced in an eppendorf (Vs ¼ 150 mL) placed into the copper coil. The latter has a variable capacity in the range 10 pF4 nF and a self-inductance of 25 mH. The coil was cooled with circulating nonane. Temperature of the nonane was controlled to obtain an equilibrium temperature of 37 0.5 C in the samples.
2.6. Magnetic properties analysis Nanoparticle or cell suspensions (around 20 mg) were introduced in sample holding capsules for Vibrating Sample Magnetometer analysis (VSM, PPMS, Quantum Design, Inc.). Field-dependent magnetization curves were measured at 310 K as a function of the external field in the range 0e3 104 Gauss and hysteresis curves were recorded for magnetic field between 500 and þ500 Gauss. Temperaturedependent magnetization at a magnetic field of 50 Gauss was recorded in the 5 Ke320 K temperature range for zero-field-cooled (ZFC) and field-cooled (FC) sample (freezing field of 50 Gauss).
2.7. Calorimetry analysis Calorimetry analyses on cellular suspensions were performed by using a Multicell Differential Scanning Calorimeter (TA Instruments). Highly packed cells suspensions (both living and fixed cells) were measured in comparison with ultrapure water and sodium cacodylate buffer in order to determine a relative heat capacity coefficient. Heat flow rates were collected by tuning the temperature from 25 to 45 C.
2.8. Iron measurement The ion concentration in the samples was determined by using a Spectro ARCOS ICP-AES. In the processing of cellular samples, an expected value (about 10 pg of iron per cell, see Table 2 for reference) was considered. Taking into account the final volume of the solution used for the analysis (10 mL) and the instrument calibration range (10e1000 mg/L of selected ion), the number of cells to be digested was calculated by using: Ncells ¼
500 mLg 0:010 L CST VFIN ¼ ¼ 5$105 cells; mg Ccell 10$106 cell
where CST is a value of concentration centered in the calibration curve, VFIN is the solution final volume and Ccell is the iron amount hypothesized per cell. The samples were first digested by boiling cell (or nanoparticle) suspensions in concentrated nitric acid (SigmaeAldrich, trace metals basis grade) for 1 h. The solutions were finally diluted with filtered ultrapure water for the analysis.
3.1. A broad range of nanomaterials tested We selected six different magnetic nanomaterials among the prime candidates for magnetic hyperthermia. Citrate-coated nanoparticles composed of maghemite or cobalt ferrite were synthesized by coprecipitation and then size-sorted to select diameters in the 10 nm range [34]. Maghemite nanoparticles synthetized by coprecipitation are the most promising for biomedical applications, for their well-established biocompatibility [35] and for their highyield synthesis. The same maghemite nanoparticles were also encapsulated in liposomes, generating a high-confined assembly of nanoparticles in the core of the composite [32]. These liposomes were tested in order to determine whether nanomaterials protected by a lipid bilayer prior to cell delivery should behave differently as heating intracellular mediators because of the intraparticles interactions as well as the phospholipids coating. Larger (14 nm) maghemite nanoparticles sharing an interface with a single small (2 nm) sphere of gold (gold/iron oxide heterodimer) were obtained by a modified version of the thermal decomposition method reported by Ref. [33]. Thermal decomposition method was used as well to produce maghemite nanocubes (approximately 20 nm in edge), previously described as excellent nano-heaters with SLP reaching the 1000e2000 W/g range [15]. Because of their hydrophobic surface, the latter two preparations were transferred in water by amphiphilic polymer coating [36]. Finally, equally efficient citrate-coated nanoflowers (20e25 nm) were synthetised by hydrothermal polyol process [19]. The size, shape and composition of these nanomaterials actually modulate their magnetic dynamics. Table 1 summarizes the saturation magnetization and mean diameters for all the nanostructures tested and electron microscopy images are presented on Fig. 1a. The SLP were measured in triplicate, in the frequency range of 320e900 kHz and amplitudes of the magnetic field of 10e24 kA/m, for the nanomaterials originally dispersed in water or resuspended in high viscosity glycerol (Fig. 1b). For all materials, the SLP values retrieved were high, over 100 W/g, and increased with the applied frequency and amplitude of the magnetic field. This dependence is well predicted by the linear response theory (LRT) [37] of heat generation taking into account both Néel and Brown relaxation processes, and the fitting with LRT is further examined in the discussion. Measurements were also performed in presence of serum
Table 1 Saturation magnetization and diameter for all nanomaterials, volumic magnetic anisotropy extracted from the SLP curves, assuming the validity of the linear response theory LRT, relaxation times sB and sN, for nanomaterials in solution. For nanocubes and nanoflowers, values of anisotropy constants K and Néel times obtained through the fitting of the data with the LRT, are presented only for comparison, but must be considered false. In cells, sB and sN retrieved from the fitting of the SLP curves (using LRT) are also shown. Nanomaterial
gFe2O3 nanoparticles gFe2O3 nanoparticles in liposomes CoFe2O4 nanoparticles AugFe2O3 dimers gFe2O3 nanocubes gFe2O3 nanoflowers
Ms (A/m) 2.8 2.8 3.2 2.8 3.9 3.9
5
10 105 105 105 105 105
Diameter (nm)
Anisotropy K (J/m3)
sB, sN (s) in Solution
sB, sN (s), in cells
10 10 9.2 14 18 22
2.6 104 2.8 104 9.2 104 0.6 104 (8 102) (1.1 103)
2 106, 2.4 108 2 106, 7.5 109 1.5 106, 1.2 106 6 106, 7.5 109 8 106, (3 109) 2 105, (4.6 109)
>2 >2 >2 >6
104, 104, 104, 104,
8 109 8 109 105 3 109
R. Di Corato et al. / Biomaterials 35 (2014) 6400e6411
6403
Fig. 1. All nanomaterials tested in the study: TEM images (left) and heating capacities (right, SLP (W/g) as a function of the frequency and intensity of the applied magnetic field) for co-precipitated maghemite nanoparticles (w10 nm, A), assembly of the same maghemite nanoparticles in liposomes (w250 nm, B), cobalt ferrite nanoparticles (w10 nm, C), iron oxide/gold dimers (w15 nm, D), iron oxide nanocubes (w18 nm on edge, E), iron oxide nanoflowers (w25 nm, F). The SLP were measured for the nanomaterials resuspended in water (black plain circles), or in glycerol (grey plain circles). The black and grey lines correspond to the fitting with the linear response theory (LRT), according to Equation (2), using the water viscosity or the glycerol one, respectively. Whenever the lines are dotted, it indicates that the adjustment with LRT is poor and should not be trusted.
6404
R. Di Corato et al. / Biomaterials 35 (2014) 6400e6411
(20%), in stable conditions, demonstrating no impact on the heating efficiency (Supporting Information Figure S1). These six nanomaterials, all excellent candidates for an efficient hyperthermia, and exhibiting different size, composition, and relaxation mechanisms, were tested in the cellular environment. 3.2. Real-time measurement of nanoparticles heating efficiency in live cells First, in order to mimic a tumor environment, we created “minitumors” consisting of dense masses of tumor cells (20 million
cells in 150 ml of culture medium). Microcalorimetry experiments showed that the heat capacity of such cellular samples was unchanged relative to water (calorific capacities Cwater ¼ 4185 J/L/K, Ccells ¼ 4125 12 J/L/K). Two of the iron oxides nanomaterials (10 nm spherical nanoparticles or 19 nm nanocubes) were then dispersed almost instantaneously in the tumor cell suspension, at iron concentrations of 50 or 25 mM (corresponding to 0.42 and 0.21 mg of total iron for nanoparticles and nanocubes, respectively). Immediately following nanoparticle dispersion (t ¼ 0) the cells were not magnetized (not attracted by a magnet, see Supporting Information Figure S2), and the nanomaterials remained outside
Fig. 2. A. Real-time analysis of hyperthermia during the interaction of the nanomaterials with living cells. The nanoparticles were rapidly dispersed and homogenized within a mass of 20 million cells at time 0. The sample was then alternated between periods at 37 C and periods of 30 s in the hyperthermia coil, which was itself thermostated at 37 C, and the temperature of the sample was measured (B). C, D: The initial increases of the heating are presented at different times (just after dispersion t ¼ 0, and at t ¼ 5 min, 15 min, 1 h and 2 h) for nanoparticles (C) and nanocubes (D), as a function of the time s starting at the application of the magnetic field. E, F: The SLP were calculated from the heating slopes, averaged over eight measurements for nanoparticles (E) and nanocubes (F) at all the time points.
R. Di Corato et al. / Biomaterials 35 (2014) 6400e6411
the cells in the carrier medium. Two hours after adding nanoparticle dispersion to cell sample thermostated at 37 C, the cells could be attracted with a magnet, demonstrating their interaction with the nanomaterials in this tumor cell model system. The analysis of the magnetic velocity (magnetophoresis) [38] was then used to quantify iron mass per cell after these 2 h of interaction, yielding values of 8 pg and 20 pg of iron per cell (0.4 mg and 0.16 mg total iron content) for nanoparticles and nanocubes, respectively, and indicating that almost all the nanoparticles had been transferred to the cells. We thus obtained a system mimicking in situ the interaction of nanomaterials with living tumour cells. We then evaluated the heating in the sample by measuring the temperature increase in response to an alternating magnetic field (700 kHz, 20 kA/m), right after nanoparticles dispersion, every 5 min for the first 15 min and every 15 min for the next 90 min, as illustrated in Fig. 2B. Each temperature measurement lasted no longer than 30 s, at temperatures not exceeding 42 C, in order to limit the impact on the cellenanoparticle interaction due to heating. The measurement of the iron mass per cell at the end of the heating experiment yielded the same values as those obtained in the previous experiment involving no heating steps, further validating the approach. For the last measurement, the temperature increase was recorded for 5 min, yielding a plateau value (Supporting Information Figure S2) of 46 C for both nanomaterials used. By contrast, in solution, nanocubes generated plateau values of 56 C and nanoparticles of 52 C. Finally, from the initial slope of each temperature measurement dT/dt, at the different times (examples shown in Fig. 2C and D), we calculated the heating power (expressed as the Specific Loss Power SLP) according to:
SLP ¼ CV=m*dT=dt;
(1)
where C is the specific heat capacity of the sample (Cwater ¼ 4185J/L/ K, Cglycerol ¼ 3086J/L/K, Ccell ¼ 4125J/L/K), V is the sample volume, and m is the mass of iron in the sample. Fig. 2E and F shows the SLP values averaged over 8 measurements. The SLP of both the nanoparticles and the nanocubes fell rapidly after their dispersion in the tumor cells sample, to less than half their initial value. These results demonstrate that the heating efficiency of magnetic nanomaterials decreases with their interaction with cells, and that the reduction occurs very rapidly. 3.3. Controlling the interaction of the nanomaterials with cells In order to unravel heating efficiency in the cellular environment, it became necessary to produce stable cellular samples allowing multiple measurements over time. To do so, a fixation with glutaraldehyde was envisaged: this crosslinking fixative creates covalent chemical bonds in the membranes, killing the cell
6405
instantaneously while preserving it in its initial structure. Before moving further with fixed samples, we ensured that the cell fixation process did not affect the calorimetric measurements. Figure S3 (Supporting Information) shows the temperature increase observed with two identical samples (cells containing nanocubes), one composed of fixed cells and the other of living cells: the superposition of the two curves validates the approach. It was then possible to control the interaction between the cells and the nanomaterials, and to visualize their subcellular localization. In particular, it may be important to distinguish between extracellular, membrane-interacting nanomaterials and fully internalized nanomaterials (Fig. 3). To do so, the different nanomaterials were administered to adherent tumor cells, which were further incubated for either 1 h at 4 C (to inhibit the internalization process) or for 1e2 h at 37 C. For the 4 C incubation, the cells were immediately detached with trypsin after incubation and fixed, to maintain the nanomaterials at their cell location, on the outside of the membrane. For the 37 C sample, after washing of the nanomaterials incubation medium, the cells were placed at 37 C overnight in complete medium to allow total internalization of the nanomaterials before detachment and fixation. For all conditions, stock cellular samples of 30 million cells were produced and further examined. First, the location of the nanomaterials was checked by TEM (Fig. 4). Most of the tested nanomaterials interacted strongly with the cell membrane, to which they adhered in the form of small aggregates (see left-hand column of Fig. 4). Only dimers and nanoflowers showed a weak association with the cell membrane. The membrane localization (without internalization) was confirmed by images of whole cells in all the different conditions (see Supporting Information Figures S4, S6, S8, S10, S12 and S14 for all the nanomaterials tested). Following the internalization phase, all six nanomaterials were found to be highly concentrated in intracellular vesicles (endosomes), while no more nanomaterials were found anymore at the membranes, as shown in the right-hand column of Fig. 4 and by the various magnifications of multiple cells in Figures S5, S7, S9, S11, S13 and S15 (Supporting Information). Each cellular system was also characterized in terms of the iron mass per cell (Table 2). Membrane-associated nanomaterials represented only a few pg of iron per cell, with the exception of liposomes (14 pg), and very low values were observed for dimers (0.8 pg) and nanoflowers (1.2 pg). Internalization led to higher values, of between 6 and 20 pg of incorporated iron per cell. Overall, all the nanomaterials fall in the same range of iron quantity per cell was found. 3.4. Systematic fall of the SLP in the cellular environment As a second step, we systematically measured the heating efficiency of these cell samples, and compared the SLP values obtained
Fig. 3. Stock samples of fixed cells were prepared, controlling both the amounts of iron per cell, and the localization of the nanomaterials: blocked at cell plasma membrane (left) or confined in intracellular compartments (right).
6406
R. Di Corato et al. / Biomaterials 35 (2014) 6400e6411
Fig. 4. TEM micrographs of cells incubated with all nanomaterials (from top to bottom: spherical maghemite nanoparticles (w10 nm), assembly of the same maghemite nanoparticles in liposomes, cobalt ferrite nanoparticles (w10 nm), iron oxide/gold dimers (w15 nm), iron oxide nanocubes (w18 nm on edge), iron oxide nanoflowers (w25 nm, F). As introduced in Fig. 2, the localization of magnetic nanoparticles into/onto cancer cells was modulated by incubation conditions. Almost exclusively membrane localization (left columns) was obtained by keeping cells at 4 C during incubation period (1 h). When the incubation was performed at 37 C, the intracellular uptake (right columns) was promoted. The nanomaterials were predominantly confined in endosomal/lysosomal compartments, generating high-concentrated assemblies into cytoplasms.
with the ones of the nanomaterials in solution. Also the fiability of the SLP measurements for cell samples was assessed by producing independently three cell samples following the same incubation condition (intracellular nanocubes). Figure S16 (Supporting Information) shows the three temperature measurements. The SLP retrieved were identical with an error of less than 15%. Then, for all cell samples, the SLP was measured systematically three times, at multiple frequencies. The results are summarized in Fig. 5. All the nanomaterials showed a systematic drop in SLP when placed in the cellular environment. Interestingly, the observed drop was the same whether the nanoparticles were attached to the cell
membrane or contained within intracellular vesicles. Heating power was thus never enhanced by cellular association, but its drop was less marked with some nanomaterials than with others: the SLP values for maghemite nanoparticles and maghemite-gold dimers only fell by about one-half in cellular conditions compared to the corresponding solution, while that of cobalt nanoparticles and nanoflowers was reduced by a factor of almost 10. Nanocubes showed an intermediate drop in SLP, by a factor of about 4. Surprisingly, the fall in SLP in cellular conditions did not necessarily correlate with that observed when the nanomaterials were suspended in high viscous glycerol (reported in Fig. 1). Only cobalt
R. Di Corato et al. / Biomaterials 35 (2014) 6400e6411
6407
Fig. 5. Measurement of heat-generating capacity (in terms of the SLP (W/g)) for the different nanomaterials (maghemite nanoparticles (A), in liposomes (B), cobalt ferrite nanoparticles (C), iron oxide/gold dimers (D), iron oxide nanocubes (E), iron oxide nanoflowers (F)), in water and within cell model systems, according to the frequency of the applied magnetic field. All data were fitted according to the linear response theory (LRT) using Equation (2). Fits represented in plain lines are valid, while for dotted line, the LRT hypotheses are not verified, and the fitting is only indicative. The parameters deduced from the fits (effective magnetic anisotropy and for the cellular samples the effective viscosity) are given in Table 1.
nanoparticles behaved similarly in the two conditions, while nanoflowers showed a drop in SLP of only 1.5-fold in glycerol but close to 10-fold in the cellular environment. Maghemite nanoparticles protected within liposomes also behaved in an interesting way: while the encapsulation within liposomes increased their SLP by a factor of 1.3e1.4 when measured in solution, after celle membrane association or internalization the SLP fell quite sharply, by a factor of almost 3, reaching a value strikingly similar to that obtained with non-encapsulated maghemite nanoparticles of the same type. In cells, the decrease recorded for the SLP can be explained by either the introduction of magnetic interparticle interactions (impacting sN) as a consequence of confinement of the particles, or
by the increase of the cellular surrounding viscosity and impeded mobility of the particles (sB increases). The fitting of the SLP curves for the four nanomaterials correctly described by the linear response theory LRT, gives some insights in these roles of magnetic interactions and intracellular viscosity (Table 1), as further commented in the discussion. 3.5. Thermal dependence of the magnetization reflects massive changes in magnetic dynamics of cell-interacting nanomaterials The temperature dependence of the magnetization upon field cooling (FC) and zero field cooling (ZFC) is the signature of the superparamagnetic properties. At low temperature (5 K), no
6408
R. Di Corato et al. / Biomaterials 35 (2014) 6400e6411
magnetization is recorded for the ZFC (magnetic moments immobilized in random directions). When temperature increases, thermal fluctuations unlock the magnetic moment when overcoming the anisotropic energy barrier, and the ZFC magnetization increases proportionally to the number of nanoparticles that have transited to this superparamagnetic state. ZFC magnetization will then reach a maximum at a temperature TB (called blocking temperature)
when most of magnetic moments are no longer blocked. Finally, when temperature further increases over TB, thermal fluctuations increase and magnetization decreases. Fig. 6 shows the FC/ZFC curves for all nanomaterials in all conditions tested. In solution, for maghemite nanoparticles, free or in liposomes, and for gold/ maghemite dimers, TB is well defined, with a maximum in magnetization in the range of 100e200 K (see Table 3 for exact
Fig. 6. Magnetism of the 6 nanomaterials (in solution, on the cell membrane, and inside cells). For each nanomaterial, field and thermal-dependence magnetization curves have been measured: hysteresis loops at T ¼ 300 K (left) and zero-field-cooled (ZFC) and field-cooled (FC) magnetic susceptibility, measured at 50 Oe (right).
R. Di Corato et al. / Biomaterials 35 (2014) 6400e6411 Table 3 Blocking temperature (TB) and anisotropy constant estimated from the thermal magnetization curves (ZFC) for nanomaterials in solution. Nanomaterial
TB (K)
Anisotropy (J/m3)
gFe2O3 nanoparticles gFe2O3 nanoparticles in liposomes
135 165 >>320 175 300 280
8.9 10.9 2.9 3.4 4.8 1.6
CoFe3O4 nanoparticles AugFe2O3 dimers gFe2O3 nanocubes gFe2O3 nanoflowers
104 104 105 104 104 104
values). By contrast, for nanocubes and nanoflowers in solution, the onset of a maximum is detected, but over 270 K (estimated TB are given in Table 3). However, at 273 K, the water becomes liquid, and Brown rotation is restored: all moments that did not undergo Néel relaxation before immediately rotated towards the field, creating a well defined jump in magnetization. These nanomaterials are thus at the transition regime between superparamagnetism and ferromagnetism at room temperature (300 K). Finally, for the cobalt ferrite nanoparticles in solution, no maximum is yet initiated when attaining 273 K, most of the nanoparticles are still blocked in the ice phase, leading to a massive jump in magnetization upon thawing. In this case, the nanoparticles are mostly in the ferromagnetic regime at 300 K. The ZFC curves when the nanomaterials are membraneassociated or internalized changed significantly. For the superparamagnetic cases (maghemite nanoparticles and dimers), the ZFC curve crossed a maximum at similar temperature that for the nanomaterials in solution. For the materials in e or at the transition of e the ferromagnetic regime, the Brown jump was no longer observed when the nanoparticles were attached to the membrane or confined within endosomes; In parallel, the hysteresis loops opened and the initial susceptibility diminished.
Brownian mobility is completely inhibited and the hysteresis loop opened, as demonstrated by the magnetic measurements. This directly shows that the nanoparticles are unable to rotate in their cellular environment and represents a magnetically-based evidence of nanoparticle immobilization in a cellular environment. Let us now focus on the quantitative structural parameters exported from both hyperthermia and magnetism measurements, especially the anisotropy constant K. In the framework of the linear response theory (LRT), the heat generation results either from Néel or from Brown relaxation processes. Néel relaxation corresponds to nanoparticles in the superparamagnetic regime, where thermal fluctuations can overcome the anisotropy energy barrier. The magnetic moment of the nanoparticle rotates within the magnetic core with a characteristic time sN¼s0exp(KV/kBT), where K is the anisotropy constant, V the nanoparticles volume, and with s0 generally equal to 109s. When KV becomes large compared to kBT, the Néel Relaxation is blocked and losses are mainly due to the Brown fluctuations of the nanoparticles itself, according to the characteristic time sB¼3hVhyd/kBT/, where Vhyd is the hydrodynamic volume of the nanoparticles, and h is the viscosity of the surrounding medium. Generally, the two relaxations coexist and the effective relaxation time s is given by 1/ s ¼ 1/sNþ1/sB: the shorter time determines the dominant mechanism of relaxation and the crossover between Néel (superparamagnetism) and Brown (ferromagnetism) regimes of relaxation depends on the anisotropy constant and carrier fluid viscosity. The LRT then describes thermal losses as a function of both the nanoparticles structural properties (size, anisotropy) as well as the frequency F and amplitude H0 of the applied magnetic field [11]:
SLP ¼ 4. Discussion In order to understand the mechanisms of nanomaterial-based magnetic hyperthermia in the cellular environment, we systematically measured heating power and magnetic properties both in solution and in cell samples, while controlling the interaction and localization of the nanomaterials with the cells. We tested a wide range of candidate magnetic nanomaterials (nanoparticles, dimers, nanocubes, nanoflowers and liposomes encapsulating nanoparticles) exhibiting different sizes, shapes, magnetic properties and compositions. All the nanomaterials lost some of their heating efficiency in the cell environment. The key findings are the following: - Heating power starts to fall as soon as the nanoparticles attach to the cell membrane. This explains the very rapid drop in SLP observed following internalization, which itself depends on the kinetics of nanoparticle attachment to the membrane (typically of the order of tenths of seconds). This also precludes any role of nanoparticle degradation in the initial decrease of heating efficiency. - All the nanomaterials tested exhibit a fall in their SLP in cellular conditions (following membrane association or internalization), but the amplitude of this fall is highly dependent on the type of nanomaterial: at best, the SLP in cellular conditions is half than obtained in solution, while at worst it is only one-tenth. - The effect of cellular confinement on the magnetic properties of the nanomaterials is variable: superparamagnetic particles are minimally affected by the cellular environment, while for nanomaterials in or close to the ferromagnetic regime, the
6409
m2o Ms2 VHo2 1 ð2pf sÞ2 3kB T r s 1 þ ð2pf sÞ2
(2)
Measurements were performed both in water and in glycerol in order to tune the viscosity contribution to sB, and thus to determine the anisotropy constant K. Fitting of experimental SLP with Equation (2) could be managed for nanoparticles of maghemite (in liposomes or individuals), of cobalt ferrite, and for goldeiron oxide dimers (plain blacklines in Fig. 1B). Values for the magnetic anisotropies retrieved are presented in Table 1. By contrast, for nanocubes and nanoflowers, LRT is no more valid. Indeed, it is applied under the condition [37] m0*Ms*V*H0
