Research Article www.acsami.org
Superparamagnetic Fe3O4 Nanoparticles: Synthesis by Thermal Decomposition of Iron(III) Glucuronate and Application in Magnetic Resonance Imaging Vitalii Patsula,† Lucie Kosinová,‡ Marija Lovrić,§ Lejla Ferhatovic Hamzić,§ Mariia Rabyk,† Rafal Konefal,† Aleksandra Paruzel,† Miroslav Šlouf,† Vít Herynek,‡ Srećko Gajović,§ and Daniel Horák*,† †
Institute of Macromolecular Chemistry, Academy of Sciences of the Czech Republic, Heyrovského nám. 2, 162 06 Prague 6, Czech Republic ‡ Institute for Clinical and Experimental Medicine, Vídeňská 1958/9, 140 21 Prague 4, Czech Republic § University of Zagreb School of Medicine, Croatian Institute for Brain Research, Salata 12, HR-10000 Zagreb, Croatia S Supporting Information *
ABSTRACT: Monodisperse superparamagnetic Fe3O4 nanoparticles coated with oleic acid were prepared by thermal decomposition of Fe(III) glucuronate. The shape, size, and particle size distribution were controlled by varying the reaction parameters, such as the reaction temperature, concentration of the stabilizer, and type of high-boiling-point solvents. Magnetite particles were characterized by transmission electron microscopy (TEM), as well as electron diffraction (SAED), X-ray diffraction (XRD), dynamic light scattering (DLS), and magnetometer measurements. The particle coating was analyzed by atomic absorption spectroscopy (AAS) and attenuated total reflection (ATR) Fourier transform infrared spectroscopy (FTIR) spectroscopy. To make the Fe3O4 nanoparticles dispersible in water, the particle surface was modified with α-carboxyl-ω-bis(ethane-2,1-diyl)phosphonic acid-terminated poly(3-O-methacryloyl-α-D-glucopyranose) (PMG−P). For future practical biomedical applications, nontoxicity plays a key role, and the PMG−P&Fe3O4 nanoparticles were tested on rat mesenchymal stem cells to determine the particle toxicity and their ability to label the cells. MR relaxometry confirmed that the PMG−P&Fe3O4 nanoparticles had high relaxivity but rather low cellular uptake. Nevertheless, the labeled cells still provided visible contrast enhancement in the magnetic resonance image. In addition, the cell viability was not compromised by the nanoparticles. Therefore, the PMG−P&Fe3O4 nanoparticles have the potential to be used in biomedical applications, especially as contrast agents for magnetic resonance imaging. KEYWORDS: superparamagnetic, nanoparticles, iron oxide, thermal decomposition, magnetic resonance imaging
■
process,5 sol−gel technique,6 spray pyrolysis,7 and thermal decomposition of organometallic complexes in high-boiling point solvents.8 In contrast to other methods, the advantage of the latter technique involves the preparation of the nanoparticles with uniform sizes and with easy size and morphology control. Iron oxide particles are rapidly engulfed in the bloodstream by macrophages and endothelial cells and collected in the reticuloendothelial system of the liver and spleen.9 This in vivo behavior (passive targeting) of iron oxide nanoparticles has been used for the detection of hepatocellular carcinoma.10 Coating the particles with appropriate agents can enhance the biocompatibility and colloidal stability in aqueous media, as
INTRODUCTION
Due to their unique physical properties and small size, magnetic nanoparticles have attracted considerable interest in the areas of biotechnology and biomedicine. Extensive investigations of the biocompatibility and toxicity of magnetic nanoparticles, especially iron oxides, have resulted in various in vivo biomedical applications. These applications include hyperthermia,1 drug delivery,2 and contrast agents in magnetic resonance imaging (MRI).3 The demands on medical and biotechnological applications are strongly associated with polydispersity, morphology, surface chemistry, and magnetization of the particles. These characteristics are dependent on the method of preparation, type of precursors and surface modifications. Currently, several methods have been proposed for the preparation of magnetic nanoparticles including the coprecipitation of Fe(III) and Fe(II) salts in the presence of an aqueous base (e.g., NH4OH or NaOH),4 hydrothermal © 2016 American Chemical Society
Received: December 28, 2015 Accepted: February 29, 2016 Published: February 29, 2016 7238
DOI: 10.1021/acsami.5b12720 ACS Appl. Mater. Interfaces 2016, 8, 7238−7247
Research Article
ACS Applied Materials & Interfaces
Synthesis of 3-O-Methacryloyl-1,2:5,6-di-O-isopropylideneα-D-glucofuranose (MDG). Methacryloyl chloride (3.22 g, 30.85 mmol) was slowly added to a solution containing 1,2:4,5-di-Oisopropylidene-α-D-glucofuranose (8.02 g, 30.85 mmol) and trimethylamine (3.12 g, 30.85 mmol) in dry acetonitrile (90 mL) at 15 °C. Then, the mixture was stirred at this temperature for 16 h and filtered. The volatiles from the filtrate were removed in a vacuum rotary evaporator at 30 °C. The residue was dissolved in CH2Cl2 (100 mL), and triethylamine hydrochloride was filtered off. Then, the filtrate was washed once with a 0.5 M NaHCO3 aqueous solution (50 mL) and twice with water (50 mL) followed by drying over MgSO4 and concentration under vacuum to afford an oily substance. The MDG was purified by vacuum distillation at 120 °C (∼0.2 Pa). Yield: 52%. RAFT Polymerization of MDG. In a 25 mL round-bottom flask, acetonitrile (9.5 mL), MDG (3.2 g; 9.75 mmol), 4,4′-azobis(4cyanovaleric acid) (0.061 g; 0.22 mmol) and 4-cyano-4(phenylcarbonothioylthio)pentanoic acid (0.12 g; 0.43 mmol) were added under vigorous stirring. The mixture was thoroughly purged with nitrogen for 15 min and heated to 70 °C for 8 h. Then, acetonitrile was removed at 30 °C in a vacuum rotary evaporator. The resulting α-carboxyl-ω-dithiobenzoate-terminated poly(3-O-methacryloyl-1,2:5,6-di-O-isopropylidene-α-D-glucofuranose) (PMDG) was purified by gel filtration in methanol on a Sephadex LH-10 column followed by vacuum-drying at room temperature over phosphorus pentoxide for 6 h. Yield: 60%. Introduction of an Amino Group to PMDG. In a 25 mL roundbottom flask, a dithiobenzoate end group-containing PMDG (1.5 g) was dissolved in methanol (12 mL) under vigorous stirring, and butylamine (0.18 g, 2.5 mmol) was added under nitrogen. Then, the reaction mixture was stirred at room temperature for 2 h. Tris(2carboxyethyl)phosphine hydrochloride (0.16 g; 0.56 mmol) and trifluoroacetate of N-(2-aminoethyl)maleimide (0.15 g; 0.59 mmol) were added under vigorous stirring, and the solution was thoroughly purged with nitrogen for 15 min. The reaction continued at room temperature for 10 h. Next, the reaction mixture was concentrated at room temperature in a vacuum rotary evaporator and the polymer was purified by gel filtration in methanol on a Sephadex LH-10 column. The polymer was dissolved in CH2Cl2 (10 mL) and washed with a 0.5 M aqueous NaHCO3 solution (5 mL) to remove the trifluoroacetic acid. Then, the organic layer was concentrated in a vacuum rotary evaporator, and the resulting α-carboxyl-ω-amino-terminated poly(3O-methacryloyl-1,2:5,6-di-O-isopropylidene-α-D-glucofuranose) (PMDG-NH2) was vacuum-dried at room temperature for 6 h over phosphorus pentoxide. Yield: 88%. Introduction of a Diethylphosphonate Group to PMDG-NH2. The reaction was performed according to previously described procedure with modification.18 In a 50 mL round-bottom flask, PMDG-NH2 (1.3 g) was dissolved in 1,4-dioxane (25 mL), and diethyl vinylphosphonate (0.35 g, 2.1 mmol) and water (5 mL) were added under vigorous stirring. Next, the mixture was heated at 100 °C for 12 h. The solvents and unreacted diethyl vinylphosphonate were evaporated at 60 and 90 °C, respectively, in a rotary evaporator under vacuum (∼0.2 Pa). The resulting α-carboxyl-ω-bis(ethane-2,1diyl)diethylphosphonate-terminated poly(3-O-methacryloyl-1,2:5,6-diO-isopropylidene-α-D-glucofuranose) (PMDG−PEt) was purified by gel filtration in methanol on a Sephadex LH-10 column. Yield: 84%. Preparation of α-Carboxyl-ω-bis(ethane-2,1-diyl)phosphonic Acid-Terminated Poly(3-O-methacryloyl-1,2:5,6di-O-isopropylidene-α-D-glucofuranose) (PMDG−P). In a 25 mL round-bottom flask, PMDG−PEt (1.2 g) was dissolved in CH2Cl2 (10 mL), and the flask was closed with a rubber septum. Trimethylsilyl bromide (0.67 g, 4.4 mmol) was injected using a syringe, and the mixture was stirred at room temperature for 8 h. Then, the volatiles were removed at 30 °C in a vacuum rotary evaporator. The residue was dissolved in a methanol/water mixture (8/1 v/v; 27 mL), and the solution was stirred at room temperature for 14 h. Finally, the solvents were removed at 50 °C in a vacuum rotary evaporator, and the product was purified by gel filtration in methanol on a Sephadex LH-10 column. Yield: 76%.
well as prolong the blood circulation time. Moreover, a suitable polymer coating can endow a particle with specific targeting abilities to provide optimal interactions with the cell membrane. The key to in vivo applications of iron oxide nanoparticles involves a balance between steric and electrostatic repulsive forces. A high surface potential (positive or negative) leads to adsorption of proteins from the plasma and increases particle opsonization. In contrast, a zero potential is associated with low opsonization. Among the various modification agents used for nanoparticle coatings, silica11 and poly(ethylene glycol)12,13 or polysaccharides, such as alginate,14 chitosan,15 dextran,16 or heparin,17 are the most frequently used. In the current study, monodisperse superparamagnetic magnetite (Fe3O4) nanoparticles were synthesized by thermal decomposition of Fe(III) glucuronate. By varying the reaction parameters, such as the temperature, concentration of the stabilizing agent and nature of the high-boiling point solvent (hydrocarbons or polyethers), the nanoparticles size, polydispersity and morphology variations were investigated. In addition, the particle surface was coated with poly(3-Omethacryloyl-α-D-glucopyranose) to make the particles dispersible in water. The coating agent was synthesized by RAFT polymerization of 3-O-methacryloyl-1,2:5,6-di-O-isopropylidene-α-D-glucofuranose and functionalized with phosphonic acid to achieve good attachment of the polymer to the nanoparticle surface. Finally, the uptake of these new waterdispersible nanoparticles by mesenchymal stem cells was investigated to determine the biocompatibility of the particles and their potential for use as a cellular label for MRI applications.
■
EXPERIMENTAL SECTION
Materials. FeCl3·6H2O (98%), octadec-1-ene (OD; 99%), icosane (IS; 99%), squalene (SQ; 98%), monomethyl poly(ethylene glycol) (mPEG; 550 Da), D-glucuronic acid sodium salt monohydrate (97.5%), 1,2:5,6-di-O-isopropylidene-α-D-glucofuranose (98%), diethyl vinylphosphonate (97%), Brij O10 [poly(ethylene glycol) oleyl ether; 709 Da], 4,4′-azobis(4-cyanovaleric acid) (ACVA), 4-cyano-4(phenylcarbonothioylthio)pentanoic acid (CPTA), butylamine (99.5%), tris(2-carboxyethyl)phosphine hydrochloride (98%), and triethylamine (99.5%) were purchased from Sigma-Aldrich (St. Louis, MO) and used as received. Methanol (99.9%) and oleic acid (OA; 95%) were obtained from Lachema (Brno, Czech Republic). Trifluoroacetate of N-(2-aminoethyl)maleimide was obtained from Santa Cruz Biotechnology (Dallas, USA). Ficoll-Paque Premium was obtained from GE Healthcare Bio-Science (Uppsala, Sweden), and luciferin and trypsin were obtained from Sigma-Aldrich (Prague, Czech Republic). All of the other reagent grade chemicals were purchased from Sigma-Aldrich and used as received. Ultrapure Qwater that was ultrafiltered using a Milli-Q Gradient A10 system (Millipore; Molsheim, France) was used throughout the experiments. Preparation of Fe(III) Glucuronate (Fe(Glu)3). The monohydrate sodium salt of D-glucuronic acid (7.24 g) was added to a mixture of methanol (118.5 mL), ethanol (39.5 mL) and water (2 mL) under vigorous stirring followed by the addition of FeCl3·6 H2O (2.79 g). The reaction mixture was refluxed at 67 °C for 4 h and cooled to room temperature. Then, Fe(Glu)3 was separated by filtration and vacuumdried over phosphorus pentoxide for 6 h. Yield: 70%. Preparation of Fe3O4 Nanoparticles. In a typical experiment, Fe(Glu)3 (1.2 g) was added to a solution of OA (0.3 mmol/mL) in OD (25 mL), and the mixture refluxed for 30 min. The resulting black solution was cooled to room temperature, and hot ethanol (70 °C; 100 mL) was added. Then, the particles were separated by a magnet, and this operation was repeated five times. Finally, the oleic acid-coated Fe3O4 nanoparticles were redispersed in toluene. 7239
DOI: 10.1021/acsami.5b12720 ACS Appl. Mater. Interfaces 2016, 8, 7238−7247
Research Article
ACS Applied Materials & Interfaces Table 1. Preparation and Characterization of Fe3O4 Nanoparticles run
solvent
oa (mmol/ml)
t (°c)
fe (wt %)a
coating (wt %)b
Dn (nm)c
PDId
Dh (nm)e
1 2 3 4 5 6 7 8 9 10
sq OD OD IS mPEG/Brij mPEG/Brij mPEG/Brij SQ OD IS
0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.3 0.3 0.3
285 300 320 343 285 300 320 285 320 343
16.6 15.4 13.4 22.6 25.6 25.9 34.5 15.7 13.1 20.3
76.8 78.5 81.3 68.5 64.3 63.9 51.9 78.1 81.8 71.7
32 13 12 18 11 11 12 20 11 17
1.3 1.1 1.03 1.03 1.08 1.09 1.2 1.2 1.03 1.02
60 89 25 74 334 297 350 52 34 81
a Fe content (AAS). bCalculated from the Fe content. cNumber-average particle diameter (TEM). dPolydispersity index (TEM). eHydrodynamic diameter (DLS); SQ squalene; OD octadec-1-ene; IS icosane; mPEG monomethyl poly(ethylene glycol) and Brij is Brij O10.
Preparation of α-Carboxyl-ω-bis(ethane-2,1-diyl)phosphonic Acid-Terminated Poly(3-O-methacryloyl-α-D-glucopyranose) (PMG−P). The di-isopropylidene protection groups of the PMDG−P were removed by a slightly modified conventional method.19 In a 100 mL round-bottom flask, PMDG−P (0.8 g) was dissolved in 80% formic acid (50 mL), and the solution was heated at 50 °C under vigorous stirring for 12 h followed by the addition of water (15 mL). The mixture was stirred for 2 h, and the solvents were removed at 50 °C in a vacuum rotary evaporator. Then, the polymer was purified by gel filtration in water on a Sephadex G-25 column. Finally, the resulting PMG−P was vacuum-dried at room temperature for 6 h over phosphorus pentoxide. Yield: 90%. Modification of Fe3O4 Nanoparticle Surface with PMG−P. The oleic acid-coated Fe3O4 nanoparticles (Run 9 in Table 1; 0.5 g) were dispersed in toluene (12 mL) and 1,4-dioxane (1 mL), and an aqueous PMG−P (0.7 g) solution (8 mL) was added. The mixture was thoroughly purged with argon for 30 min and heated at 70 °C for 72 h under vigorous stirring. The PMG−P-coated nanoparticles (PMG− P&Fe3O4) were purified by precipitation with ethanol (2 × 50 mL), washed with CH2Cl2 (7 × 50 mL) and redispersed in water under sonication for 30 min (Ultrasonic Homogenizer UP200S Hielscher; 45% power). Cell Isolation. Mesenchymal stem cells (MSCs) from transgenic Lewis rats LEW-Tg(Gt(ROSA)26Sor-luc)11Jmsk expressing luciferase were isolated from visceral adipose tissue (epididymal fat pad and perirenal fat).20 Briefly, the fat was excised, washed, and centrifuged. The tissue was digested by collagenase, filtered, and centrifuged. The pellet was resuspended in phosphate buffer, and the cells were layered onto the Ficoll (density 1.077 g/cm3) and centrifuged. The cells at the medium-Ficoll interface were collected after centrifugation, washed and incubated. The cells were cultivated in media that was changed twice each week, maintained at subconfluent levels and released from the tissue flask using trypsin. Cell Labeling. To label the MSCs, a nanoparticle dispersion with 0.1 or 0.5 mM Fe3O4 was added to the cell-containing cultivation media and incubated for 48 h. The viability of the MSCs was determined by counting them in the Bürker chamber using the Trypan Blue exclusion test, which can determine the number of viable cells present in a cell suspension. Independently, the viability of the cell suspensions was verified using an IVIS Lumina XR optical imager (Caliper Life Sciences, Hopkinton, MA) after the addition of luciferin to the medium. The bioluminescent signal after catalyzed luciferin decomposition confirmed the cell viability. MR Relaxometry and Imaging. The magnetic resonance relaxation times were measured using a 0.5 T Bruker Minispec MQ20 relaxometer and a 4.7 T Bruker BioSpec spectrometer (Bruker BioSpin, Ettlingen, Germany), which was also used for MR imaging. A standard CPMG sequence involved an echospacing of 2 ms at 0.5 T and 6.7 ms at 4.7 T, number of echoes between 128 and 2,000, recovery time TR = 5000 ms, and 8 acquisitions. The T2 relaxation times were calculated using a built-in MiniSpec (0.5 T) or ViDi homemade software21 (4.7 T). Identical protocols were employed to
determine the T2 values of both the nanoparticle dispersions and the cell suspensions in the 4% porcine skin gelatin (Sigma-Aldrich). The relaxivity (r2) of the nanoparticles (s−1/mM) was calculated as the reciprocal value of the transverse relaxation time (T2) related to the Fe3O4 concentration (c) after deduction of the gel contribution according to the following eq 1:
r2 = (1/T2 − 1/T20)/c
(1)
where T20 is the transverse relaxation time of the pure gel. Similarly, the R2 relaxation rate of the cell suspensions (s−1/106 cells/ml) was calculated, and the reciprocal value of the transverse relaxation time (1/T2c) was related to the cell concentration (cx) after deduction of the contribution from the unlabeled cells:
R 2 = (1/T2c − 1/T2c0)/cx
(2)
where T2c0 is the transverse relaxation time of the suspension containing the same amount of unlabeled cells. Characterization of the Nanoparticles. TEM micrographs of the iron oxide nanoparticles were obtained on a Tecnai Spirit G2 transmission electron microscope (FEI; Brno, Czech Republic) that was equipped with an energy-dispersive X-ray detector (EDAX; Mahwah, NJ) for determination of the elemental composition. The crystal structure was analyzed by selected-area electron diffraction (SAED). The experimental TEM/SAED patterns were converted into 1D-diffractograms using the ProcessDiffraction software22 and compared to X-ray powder diffraction patterns calculated with the Powder Cell program.23 The calculation was based on known FexOy structures obtained from the Crystallography Open Database.24 The number-average diameter (Dn = ΣDi/N), weight-average diameter (Dw = ΣDi4/ΣDi3) and polydispersity index (PDI = Dw/Dn) were calculated by statistical analysis of at least 900 individual particles from the TEM micrographs using Atlas software (Tescan Digital Microscopy Imaging; Brno, Czech Republic), where Di was the diameter of i-th particle and N was the total number of particles. The hydrodynamic diameter (Dh) and ζ-potential were determined by dynamic light scattering (DLS) using a Nano-ZS Zetasizer ZEN3600 Model (Malvern Instruments, U.K.). A PerkinElmer 3110 atomic absorption spectrometer (AAS) was used to analyze the amount of iron in the samples by measuring the solution obtained after mineralization of the particles with dilute HCl (1:1) at 80 °C for 1 h. The phase composition of the Fe3O4 nanoparticles was determined by X-ray powder diffraction (XRD) on a Bruker D8 diffractometer (CuKα radiation, Sol-X energy dispersive detector). The hysteresis loops were measured using an ADE Magnetics model EV9 VSM magnetometer (MicroSense; Lowell, MA). The relative iron oxide content was determined using a PerkinElmer TGA 7 thermogravimetric analyzer (Norwalk, CT). The surface-modified Fe3O4 nanoparticles were heated in air from room temperature to 800 °C at 10 °C/min. The 1H and 31P NMR spectra were obtained using a Bruker Avance DPX-300 spectrometer at 23 °C, operating at 300.13 MHz. Hexamethyldisiloxane (HMSO) served as an internal standard (0.05 ppm) for 1H NMR analysis, and phosphoric acid (85%) was an external standard in a glass capillary (0 ppm) for 7240
DOI: 10.1021/acsami.5b12720 ACS Appl. Mater. Interfaces 2016, 8, 7238−7247
Research Article
ACS Applied Materials & Interfaces 31
P NMR. CDCl3, D2O and DMSO were used as the solvents. The ATR FTIR spectra were recorded on a PerkinElmer Paragon 1000PC spectrometer using the Specac MKII Golden Gate single attenuated total reflection (ATR) system with a diamond crystal, and the angle of incidence was 45°. The molecular weight (Mw) of the polymers was determined on a modular GPC system equipped with a RIDK-102 ́ refractive index detector (Laboratorni ́ přistroje; Prague, Czech Republic) and a LCD 2084 UV−vis photometric detector (ECOM; Prague, Czech Republic) operated at λ = 254 nm. Two PLgel 104 Å and 50 Å columns (300 × 7.5 mm) with a 10 μm particle size (Polymer Laboratories; Church Stretton, U.K.) were used. The chromatographic data were treated using the Clarity software (DataApex; Prague, Czech Republic). Tetrahydrofuran and toluene served as the mobile phase and flow marker, respectively, (retention time of toluene was 18.2 min) at a flow rate of 1 mL/min. Polystyrene standards (Mw = 500, 1000, 3000, and 10 000) were used for calibration. Characteristics of Fe(Glu)3 and all prepared products are shown in Supporting Information (SI), where GPC chromatograms, ATR FTIR, 1H, and 31P NMR spectra are presented.
Figure 1. TGA of the Fe(Glu)3 complex(full line) and PMG− P&Fe3O4 (dashed line).
RESULTS AND DISCUSSION The preparation of superparamagnetic iron oxides with a narrow size distribution, good magnetic properties and low toxicity remains challenging when the nanoparticles are intended for biomedical applications, such as MRI of labeled cells. In this study, magnetic Fe3O4 nanoparticles were synthesized by thermal decomposition of an Fe(III)/glucuronic acid complex (Fe(Glu)3). Synthesis and Characterization of Fe(Glu)3. The Fe(Glu)3 prepared by reaction of sodium glucuronate with Fe(III) chloride was analyzed by ATR FTIR spectroscopy (Figure S1 in Supporting Information). The distinctive peaks of the COO− groups in the range of 1,650−1,510 and 1,400− 1,280 cm−1 were due to asymmetrical and symmetrical vibrations, respectively.25 The spectrum of Fe(Glu)3 contained two peaks at 1368 and 1418 cm−1, which were due to νs(COO−) symmetric stretching vibrations, and the band at 1,595 cm−1 was assigned to νas(COO−) asymmetric stretching vibrations of the carboxyl groups. The carboxylate orientation in the Fe(Glu)3 complex was determined from the differences (Δ) between the νs(COO−) and νas(COO−) peaks (177 and 227 cm−1), indicating the formation of bridging coordination.26 The peaks located at 2,910 and 2,859 cm−1 were assigned to νas(CH2) asymmetric and νs(CH2) symmetric stretching vibrations, respectively. The bands located at 1,296 and 1,035 cm−1 corresponded to νd(CH) deformations and ν(C−O) stretching vibrations, respectively. The broad band in the 2,500−3,600 cm−1 region was due to ν(OH) vibrations that originated from alcohol groups and water molecules in the Fe(Glu)3 complex. The structure of iron(III) glucuronate was confirmed by ATR FTIR spectroscopy. The formation of the nanoparticles by thermal decomposition of Fe(Glu)3 can be explained by ketonic decarboxylation of carboxylic acids27,28 driven by collision of poly(oxoiron) clusters during rapid nucleation, which resulted in particle growth.29,30 The mechanism of the Fe3O4 nanoparticle formation can be partially deduced from thermogravimetric analysis (TGA) of Fe(Glu)3 (Figure 1). Four main stages were distinguished in the thermogram, which was similar to that of iron(III) complexes with oleic or mandelic acid.31 The first weight loss at temperatures 379 °C (region D in Figure 1). Effect of some Reaction Parameters on the Properties of the Fe3O4 Nanoparticles. In the first set of experiments, the influence of the decomposition temperature ranging from 255 to 365 °C on the particle size and size distribution was studied. Fe(Glu)3 was decomposed in high-boiling point solvents, such as squalene (SQ), octadec-1-ene (OD), icosane (IS), and a mPEG/Brij O10 mixture (1/1 mol/mol). The concentration of the OA stabilizer remained constant (0.15 mmol/mL) throughout these experiments. The reactions in hydrophobic OD at low temperatures (255 and 270 °C) did not produce particles because no nucleation was initiated, which was confirmed by TGA. At 285 and 300 °C, polydisperse particles with sizes of 32 and 13 nm, respectively, were obtained (Runs 1 and 2 in Table 1; Figure 2a,b). This result suggests that these temperatures were still not sufficiently high to ensure quick nucleation that would be followed by the growth stage, which are necessary conditions for the formation of monodisperse nanoparticles.32 Monodisperse cubic nanoparticles that were 12 and 18 nm in size were obtained at 320 and 343 °C, respectively (Runs 3, 4 in Table 1; Figures 2c,d). A big advantage of the thermal decomposition of Fe(Glu)3 complex, in contrast to coprecipitation of iron salts in water, is that it provides a narrow particle size distribution. Moreover, this approach allows for easy particle size control. It is needless to say that only the monodisperse particles can provide uniform physical, chemical, and biological properties, which are so important in biomedical applications. When the particles were synthesized in a polar mPEG/Brij O10 mixture at 285−320 °C, polydisperse irregular particles (11 and 12 nm in size) were obtained (Runs 5−7 in Table 1; Figure 2e). Therefore, at 320 °C, the separation of the nucleation and growth periods did not occur, which may be associated with the higher polarity of PEG in contrast to the hydrocarbon solvents resulting in prolonging of the nucleation at temperatures >300 °C. It is important to note that at 330 °C, the mPEG/Brij O10
■
7241
DOI: 10.1021/acsami.5b12720 ACS Appl. Mater. Interfaces 2016, 8, 7238−7247
Research Article
ACS Applied Materials & Interfaces
Figure 2. TEM micrographs of the Fe3O4 nanoparticles prepared (a, f) in SQ (285 °C), (b) OD (300 °C), (c, g) OD (320 °C), (d, h) IS (343 °C), and (e) mPEG/Brij O10 (320 °C). The concentration of OA was (a−e) 0.15 and (f−h) 0.3 mmol/mL. Insets - particle size distribution histograms.
Figure 3. ATR FTIR spectra of (a) oleic acid-coated Fe3O4 and (b) PMG−P&Fe3O4 nanoparticles. The Fe3O4 particles were prepared at 285 (Run 1), 300 (Run 2), 320 (Run 3), and 343 °C (Run 4) with 0.15 mmol OA/ml and at 320 °C (Run 9) with 0.3 mmol OA/ml.
°C was significantly larger than the Dn (TEM) due to aggregation. The ATR FTIR spectra of the Fe3O4 particles prepared in nonpolar solvents at 0.15 mmol OA/ml contained peaks at 1555 and 1700 cm−1, which were due to symmetric and asymmetric stretching vibrations of the COO− groups, respectively (Figure 3a). The three peaks at 1436, 2851, and 2920 cm−1 were due to νd(CH2) deformations, as well as νs(CH2) symmetric and νas(CH3) asymmetric stretching vibrations, respectively. Similar results were observed in the spectra of the particles prepared with 0.3 mmol OA/ml. ATR FTIR spectroscopy confirmed the presence of OA on the nanoparticle surface. The amount of coating (wt %) on the nanoparticle surface was calculated according to eq 3:
mixture started to decompose. Therefore, further investigation of particle formation was impossible. Moreover, the particles prepared in the mPEG/Brij O10 mixture were poorly dispersible in all organic solvents, which prevent their future application. In the next set of experiments, the OA concentration in the hydrophobic SQ, OD, and IS solvents was increased to 0.3 mmol/mL (Table 1, Runs 8−10). Polydisperse 20 nm iron oxide nanoparticles with irregular shapes were prepared at 285 °C (Figure 2f). Monodisperse 11 and 17 nm particles with a cubic shape were obtained at 320 and 343 °C, respectively (Figure 2g,h). The majority of the nanocubes laid on their faces, which resulted in the preferred nanoparticle orientation observed in the TEM/SAED diffraction patterns and discussed below. It is important to note that the particle size decreased as the OA concentration increased because the higher amounts of OA stabilized more nuclei, which reduced the nanoparticle size. The Dh of the prepared particles was in the range of 25−89 nm for Runs 1−4 and 8−10, suggesting that the particles were separated and did not aggregate (Table 1). The Dh of the particles prepared in the mPEG/Brij O10 mixture at 285−320
coating = 100 − (Fe × 100)/71.68
(3)
where Fe is the percentage of iron in the nanoparticles (according to AAS) and 71.68 is the percentage of Fe in Fe3O4. For example, 18.2 wt % Fe3O4 was calculated for Run 10, which contained 13.1 wt % Fe (Table 1). Therefore, the particles 7242
DOI: 10.1021/acsami.5b12720 ACS Appl. Mater. Interfaces 2016, 8, 7238−7247
Research Article
ACS Applied Materials & Interfaces
Figure 4. (a) TEM/SAED diffraction pattern and (b) comparison of experimental TEM/SAED diffractogram for nanoparticles from Run 9 (Table 1) with the calculated XRD diffraction pattern for Fe3O4. The strongest diffractions were those with the last diffraction index equal to zero (hk0 diffractions; marked in bold) due to the preferred orientation of the cubic nanoparticles. Most of the nanocubes lay on their [001] facets, and therefore, the strongest diffractions corresponded to the zone axis [001]. (c) Hysteresis loops of Fe3O4 nanoparticles from Runs 7 and 9 (Table 1) at 300 K. (d) XRD of the nanoparticles from Runs 1, 3, 7−10 (Table 1). The vertical bars represent the Fe3O4 standard.
particles synthesized by thermal decomposition of Fe(Glu)3 were magnetite. The magnetic hysteresis of the Fe3O4 nanoparticles was measured using a vibrating sample magnetometer. The saturation magnetization (Ms) of 12 and 11 nm nanoparticles from Run 7 and 9 (Table 1) was 25.3 and 9.5 Am2/kg, respectively (Figure 4c). According to AAS, Runs 7 and 9 contained 51.9 and 81.8 wt % of the OA shell, and the Ms of the neat Fe3O4 particles was calculated to be 52.6 and 52.2 Am2/kg, respectively. The obtained Ms values were lower than those in bulk Fe3O4 (Ms = 90 Am2/kg)34 but comparable to previously published data.31 Because no hysteresis and remanent magnetization were observed, the superparamagnetic behavior of the synthesized nanoparticles was confirmed. Preparation of the PMG−P Ligand. In this report, glucose-based PMG−P was chosen as an original and novel ligand for the phase transfer of the hydrophobic Fe3O4 particles into water. Advantage of carbohydrate coatings of Fe3O4 consists in their good biocompatibility and better cellular internalization in contrast to the widely used PEG. Moreover, compared to another commonly applied silica shell, carbohydrates prevent undesirable interactions with proteins reducing thus opsonization and prolonging particle serum and blood half-life that is important for in vitro and in vivo applications.35 The hydrophilic PMG−P ligand was synthesized according to Scheme 1. First, MDG was prepared from 1,2:5,6-di-O-
contained 81.8 wt % coating. The TEM/EDAX spectra (data not shown) were dominated by peaks corresponding to Fe, C, and O, confirming the expected composition of the Fe3O4 nanoparticles. The TEM/SAED diffraction pattern of Run 9 was compared to the calculated diffraction patterns of all common FexOy oxides (Figure 4a). The comparison confirmed that our nanoparticles corresponded to the magnetic form of iron oxide (i.e., Fe3O4) (Figure 4b). Because the Fe3O4 and γ-Fe2O3 crystal structures are very similar, they cannot be distinguished from the TEM/SAED patterns, especially in nanocrystals with broad diffractions.33 As shown in Figure 4b, most of the nanoparticles exhibited a cubic shape and were laid on their faces (crystallographic lattice planes [001]), which resulted in a strong preferred orientation along the [001] lattice direction. Therefore, the diffractions with the last index equal to zero exhibited the highest intensity in the experimental TEM/SAED patterns (Figure 4b). However, the diffractions with a nonzero last diffraction index were much weaker or even extinct, as demonstrated by a comparison of the SAED and XRD diffraction patterns (Figure 4b). The structure of the superparamagnetic nanoparticles was also analyzed by X-ray powder diffraction (XRD). The position and relative intensities of all of the diffraction peaks were in good agreement with those from standard bulk Fe3O4 (Figure 4d). Therefore, the XRD results confirmed that the superparamagnetic nano7243
DOI: 10.1021/acsami.5b12720 ACS Appl. Mater. Interfaces 2016, 8, 7238−7247
Research Article
ACS Applied Materials & Interfaces Scheme 1. Synthesis of Hydrophilic Ligand for Functionalization of Fe3O4 Nanoparticles
and PMG−P were confirmed by ATR FTIR, 1H and 31P NMR spectroscopy (Figures S3−S11). Phase Transfer of Fe3O4 Nanoparticles in Water. The surface modification of the Fe3O4 nanoparticles with PMG−P was monitored by ATR FTIR spectroscopy. In the spectrum of PMG−P&Fe3O4, the broad band located at 3,500 cm−1 was due to ν(OH) stretching vibrations (Figure 3b). The bands located at 1714 and 1036 cm−1 were assigned to ν(CO) and ν(C−O−C) stretching vibrations, respectively. In contrast to the spectrum of neat PMG−P, the peak corresponding to ν(P− O) stretching vibrations in PMG−P&Fe3O4 was shifted to the left and overlapped with the band corresponding to ν(C−O− C) stretching vibrations. This shift may be due to deprotonation of the hydroxyl groups resulting from attachment of the PO3 groups to the Fe atoms. Therefore, the ATR FTIR spectra confirmed that the surface of the Fe3 O4 nanoparticles was successfully modified with PMG−P. According to the TGA, the amount of coating on the PMG− P&Fe3O4 nanoparticles was 72 wt % (Figure 1). The colloidal stability of the PMG−P&Fe3O4 nanoparticles was investigated by DLS at different pH values (5.3−9.2) and NaCl concentrations (10−3 − 1 mol/L). The Dh of PMG− P&Fe3O4 increased from 135 to 950 nm at a pH of 9.6 (ζpotential −25 mV) and a pH of 5.1 (ζ-potential −18.5 mV), respectively. The Dh and ζ-potential of the particles at a pH of
isopropylidene-α-D-glucofuranose and methacryloyl chloride in the presence of trimethylamine. Then, MDG was polymerized to PMDG via RAFT polymerization using the ACVA initiator and CPTA transfer agent. The molecular weight of the polymer was determined to be 4600 g/mol with a polydispersity of 1.06. The dithiobenzoate end groups of PMDG were aminolyzed by butylamine to thiol groups36 that were coupled with N-(2aminoethyl)maleimide to introduce amino groups (denoted as PMDG−NH2). In the next step, the amino groups of PMDG− NH2 reacted with diethyl vinylphosphonate via an aza-Michael addition,18,37 resulting in an α-carboxyl-ω-bis(ethane-2,1-diyl)diethylphosphonate derivative of PMDG (denoted as PMDG− PEt). According to GPC, the elution profiles of the initial PMDG, PMDG-NH2, and PMDG−PEt were unimodal, and the retention times of the eluting polymers were the same (Figure S2). This result confirmed the absence of any coupling or degradation of the polymers during the modification. Finally, the diethyl phosphonate groups were deprotected using trimethylsilyl bromide, and phosphonic acid was introduced as an efficient chelating agent to provide strong covalent bonds with the nanoparticle surface. The resulting α-carboxyl-ωbis(ethane-2,1-diyl)phosphonic acid-terminated PMDG was treated with 80% formic acid to form the water-soluble PMG−P. The structures of PMDG, PMDG-NH2, PMDG−PEt, 7244
DOI: 10.1021/acsami.5b12720 ACS Appl. Mater. Interfaces 2016, 8, 7238−7247
Research Article
ACS Applied Materials & Interfaces 7.2 was 200 nm and −20 mV, respectively. According to the literature, this ζ-potential at a pH of 7.2 should be insufficient to stabilize the nanoparticle dispersion via electrostatic repulsion.38 Nevertheless, our PMG−P&Fe3O4 dispersion was stable for more than 1 week, and no aggregation was observed. This result may be due to additional stabilization of the particles by steric repulsions provided by the PMG−P chains. When the particles were measured in aqueous NaCl solutions (electrolyte), the Dh and ζ-potential associated with the formation of a counterion layer39 increased from 140 to 890 nm and from −15 to −5 mV as the NaCl concentration increased from 10−3 to 1 mol/L, respectively. Moreover, the PMG−P&Fe3O4 colloid was stable for 1 week at 10−3 and 10−2 mol NaCl/l. Therefore, PMG−P can be considered a promising stabilizing agent for surface modification of iron oxide nanoparticles to render them dispersible in water. Labeling of MSCs with PMG−P&Fe3O4 Nanoparticles and MR Experiments. Because cell transplantation requires noninvasive monitoring by MRI, a suitable cell label with sufficient relaxivity is needed. In this study, we took advantage of the newly developed PMG−P&Fe3O4 nanoparticles for these MR relaxometry investigations. The PMG−P&Fe3O4 relaxivities at two magnetic fields (0.5 and 4.7 T) are summarized in Table 2. The PMG−P&Fe3O4 nanoparticles
bioluminescence reaction of luciferin. Therefore, the viable cells emit light after addition of luciferin into the medium. Figure 5 shows a quantitatively similar optical signal from unlabeled and PMG−P&Fe3O4-labeled MSCs after the addition of luciferin at the saturation concentration (3 mg/mL), confirming no substantial differences in the cell viability. The relaxation rate (R2) of the MSCs labeled with PMG− P&Fe3O4 (Table 2) was lower than that of the cells labeled with PLL-, D-mannose- or dextran-coated iron oxide nanoparticles,4,41 which was most likely due to the lower cellular uptake of the PMG−P&Fe3O4 nanoparticles. Figure 6 shows the MR images of suspensions of unlabeled and labeled cells in gelatin. Labeling by PMG−P&Fe3O4 at both 0.1 mM and 0.5 mM Fe3O4 provided visible contrast in the MR images that was higher for the latter concentration. Although the relaxation rate (R2) was lower than in the PLL-, D-mannose- or dextrancoated iron oxide nanoparticle-labeled MSCs due to lower particle uptake, the cells labeled with the PMG−P&Fe3O4 nanoparticles at a high concentration (0.5 mM Fe3O4) still provided a sufficient contrast change in the T2-weighted MR image. In fact, this lower particle uptake may be beneficial with respect to cell viability.
■
CONCLUSIONS Superparamagnetic Fe3O4 nanoparticles are conventionally synthesized by several approaches including physical and chemical methods. They involve coprecipitaion,42 hydrothermal and solvothermal synthesis,43 sol−gel and polyol reactions,44 microemulsion,45 sonolysis,46 biosynthesis,47 and so on. These techniques are often complex; require high-cost starting materials, equipment, or both; produce large particles; do not provide particle shape and size control; particles tend to aggregate or release toxic compounds; and so on. In this study, attention was paid to high-temperature thermal decomposition method, namely thermolysis of Fe(III) glucuronate under different reaction conditions. The Fe(Glu)3, that served as nontoxic alternative to organometallic compounds, such as iron(III) pentacarbonyl and N-nitroso-N-phenylhydroxylamin, was prepared by reaction of environmentally friendly and inexpensive substances, that is, iron(III) chloride and sodium glucuronate.48 Advantage of this technique is production of monodisperse particles with controlled size. According to TEM
Table 2. Relaxivity (r2) of the PMG−P&Fe3O4 Nanoparticles and Relaxation Rates (R2) of the Labeled MSCs −1
r2 (s /mM Fe3O4) R2 (s−1/106 cells/ml) 0.1 mM PMG−P&Fe3O4 in the medium 0.5 mM PMG−P&Fe3O4 in the medium
0.5 T
4.7 T
264 ± 12
339 ± 10
0.27 ± 0.06 1.00 ± 0.09
0.87 ± 0.30 2.21 ± 0.01
had higher r2 relaxivities than similar commercially available contrast agents that are based on iron oxides.40 The Trypan Blue exclusion test did not reveal any differences in the cell viability between the control cells (86%) and the MSCs incubated with the PMG−P&Fe3O4 nanoparticles at 0.1 mM Fe3O4 (87%) or 0.5 mM Fe3O4 (86%). The cell viability was also confirmed by optical imaging of the cells that contained a gene for luciferase, which is an oxidative enzyme that catalyzes a
Figure 5. Luminescent image (in relative color scale) of the MSCs after addition of luciferin to the medium. (a) Unlabeled cells and (b) cells labeled with 0.1 and (c) 0.5 mM PMG−P&Fe3O4 provided similar luminescence signals, confirming that labeling did not affect the cell viability. 7245
DOI: 10.1021/acsami.5b12720 ACS Appl. Mater. Interfaces 2016, 8, 7238−7247
Research Article
ACS Applied Materials & Interfaces
Figure 6. MR cross-sectional T2-weighted image of the test tubes containing (a) unlabeled MSCs and cells labeled with (b) 0.1 mM and (c) 0.5 mM PMG−P&Fe3O4. The cell concentration was 400 000 cells per 0.5 mL sample (i.e., 25 cells per image voxel).
■
ACKNOWLEDGMENTS Support of the Ministry of Education, Youth and Sports of the Czech Republic (project LH14318 and POLYMAT LO1507), BIOCEV (CZ.1.05/1.1.00/02.0109) from the European Regional Development Fund, EU FP7 project GlowBrain (REGPOT-2012-CT2012-316120), and the Ministry of Health of the Czech Republic (project IN00023001) is acknowledged. The authors wish to thank M. Veverka from the Institute of Physics, AS CR, for the magnetometer measurements and J. Hromádková from the Institute of Macromolecular Chemistry for the TEM studies. The authors also acknowledge the Charles University in Prague for the opportunity for doctoral studies for V. Patsula.
analysis, the nanoparticle size was
