Fluorescence-Modified Superparamagnetic Nanoparticles

May 11, 2006 - This report describes the preparation and characterization of new magnetic fluorescent nanoparticles and our success in using them to l...
15 downloads 14 Views 302KB Size
Langmuir 2006, 22, 5385-5391

5385

Fluorescence-Modified Superparamagnetic Nanoparticles: Intracellular Uptake and Use in Cellular Imaging Franck Bertorelle,† Claire Wilhelm,‡ Jacky Roger,† Florence Gazeau,‡ Christine Me´nager,*,† and Vale´rie Cabuil† Laboratoire des Liquides Ioniques et Interfaces Charge´ es, Equipe Colloı¨des Inorganiques, UMR 7612 CNRS/UniVersite´ Pierre et Marie Curie (Paris 6), 4 place Jussieu, case 63, 75252 Paris, Cedex 05, France, and Laboratoire Matie` re et Syste` me Complexes, UMR CNRS 7057/UniVersite´ Paris 7, Baˆ t 11, 140 rue de Lourmel, 75015 Paris, France ReceiVed October 6, 2005. In Final Form: January 16, 2006 This report describes the preparation and characterization of new magnetic fluorescent nanoparticles and our success in using them to label living cells. The bifunctional nanoparticles possess a magnetic oxide core composed of a dimercaptosuccinic acid (DMSA) ligand at the surface and a covalently attached fluorescent dye. The nanoparticles exhibited a high affinity for cells, which was demonstrated by fluorescence microscopy and magnetophoresis. Fluorescence microscopy was used to monitor the localization patterns of magnetic nanoparticles associated with cells. We observed two types of magnetic labeling: adsorption of the nanoparticles on the cell membrane (membranous fluorescence) and internalization of the nanoparticles inside the cell (intracellular vesicular fluorescence). After internalization, nanoparticles were confined inside endosomes, which are submicrometric vesicles of the endocytotic pathway. We demonstrated that endosome movement could be piloted inside the cell by external magnetic fields such that small fluorescent chains of magnetic endosomes were formed in the cell cytoplasm in the direction of the applied magnetic field. Finally, by measuring the critical cellular magnetic load (quantitated by magnetophoresis), we have demonstrated the potential of this new magneto-fluorescent nanoagent for medical use.

1. Introduction Over the past decade, numerous biomedical applications have emerged for superparamagnetic iron oxide nanoparticles dispersed in an aqueous medium.1,2 The combination of the nanometer size with superparamagnetic properties led to their use in labeling and sorting cells or organelles,3 magnetic resonance imaging (MRI),4 targeted drug delivery,5 and hyperthermia.6,7 The addition of fluorescence properties to these magnetic nanoparticles offers new potential for in vitro and in vivo imaging. The present study was motivated by the use of these bifunctional nanoparticles for medical imaging by combined MRI and fluorescence imaging techniques: ex vivo by classical fluorescence microscopy or in vivo via fibered confocal fluorescence microscopy.8 Both methods of detection may be important in diagnostic and therapeutic applications. Although MRI provides anatomically sensitive deep tissue imaging, optical techniques offer higher spatial resolution for visualizing cellular structure and quantifying molecular events. There are increased possibilities for molecular and cellular * Corresponding author. E-mail: [email protected]. Tel: 33-144-27-27-57. Fax: 33-1-44-27-36-75. † UMR 7612 CNRS/Universite ´ Pierre et Marie Curie (Paris 6). ‡ UMR CNRS 7057/Universite ´ Paris 7. (1) Halbreich, A.; Roger, J.; Pons, J. N.; Da Silva, M. F.; Hasmonay, E.; Roudier, M.; Boynard, M.; Sestier, C.; Amri, A.; Geldwerth, D.; Fertil, B.; Bacri, J. C.; Sabolovic, D. Scientific and Clinical Applications of Magnetic Carriers; Ha¨feli, U., Schutt, W., Teller, J., Zborowski, M., Eds.; Plenum Press: New York, 1997; pp 399-417. (2) Halbreich, A.; Roger, J.; Pons, J. N.; Geldwerth, D.; Da Silva, M. F.; Roudier, M.; Bacri, J. C. Biochimie 1998, 80, 379-390 (3) Perrin-Cocon, L. A.; Marche, P. N.; Villiers, C. L. Biochem. J. 1999, 338, 123-130. (4) Koenig, S. H.; Kellar, K. E. Acad. Radiol. 1996, 3, 273. (5) Kohler, N.; Sun, C.; Wang, J.; Zhang, M. Langmuir 2005, 21, 8858-8864. (6) Roger, J.; Pons, J. N.; Massart, R.; Halbreich, A.; Bacri, J. C. Eur. Phys. J.: Appl. Phys. 1999, 5, 321-325. (7) Moroz, P.; Jones, S. K.; Gray, B. N. Int. J. Hyperthermia 2002, 18, 267284 (8) Laemmel, E.; Genet, M.; Le Goualher, G.; Perchant, A.; Le Gargasson, J. F.; Vicaut, E. J. Vasc. Res. 2004, 41, 400-411.

imaging using magnetic and fluorescent nanoparticles. First, cells may be labeled with these nanoparticles in vitro before in vivo transplantation. In this case, surface properties of the particles play a crucial role in the efficiency of cell internalization. After transplantation of the labeled cells, in vivo imaging monitors cell migration throughout the body. Cellular monitoring is particularly important for evaluating cell-based therapy. Second, magnetic nanoparticles may be used in conjugation with a drug carrier, for example, by encapsulation into liposomes,9,10 allowing the use of magnetic guidance to target the carrier to specific sites. The drug itself may be linked to the magnetic particles or to the membrane of the vesicles. In a recent study, Lu et al.11 point out that there are few reports in the literature describing the preparation of magnetic and fluorescent nanoparticles. There are many publications describing micrometric magnetic particles12 that exhibit fluorescence properties, but only a few relate the preparation of particles on the nanometer scale. Hatanaka et al.13 describe the immobilization of FITC-avidin onto the surface of iron oxide particles during the synthesis of the nanoparticles. These particles are used to observe magnetic patterns on floppy disks. The synthesis of nanocomposite particles with a superparamagnetic core and a shell of quantum dots is described by Wang et al.14 The quantum dots are linked to the surface of the nanoparticles via a thiolmetal bond. Synthesis takes place in a biphasic medium, and the (9) Lesieur, S.; Grabielle-Madelmont, C.; Me´nager, C.; Cabuil, V.; Dahdi, D.; Pierrot, P.; Edwards, K. J. Am. Chem. Soc. 2003, 125, 5266-5267. (10) Martina, M. S.; Fortin, J. P.; Me´nager, C.; Cle´ment, O.; Barratt, G.; Grabielle-Madelmont, C.; Gazeau, F.; Cabuil, V.; Lesieur, S. J. Am. Chem. Soc. 2005, 127, 10676-10685. (11) Lu, H.; Yi, G.; Zhao, S.; Chen, D.; Guo, L. H.; Cheng, J. J. Mater. Chem. 2004, 14, 1336-1341. (12) Mokari, T.; Sertchook, H.; Aharoni, A.; Ebenstein, Y.; Avnir, D.; Banin, U. Chem. Mater. 2005, 17, 258-263. (13) Hatanaka, S.; Matsushita, N.; Abe, M.; Nishimura, K.; Hasegawa, M.; Handa, H. J. Appl. Phys. 2003, 93, 7569-7570. (14) Wang, D.; He, J.; Rosenzweig, N.; Rosenzweig, Z. Nano Lett. 2004, 4, 409-413.

10.1021/la052710u CCC: $33.50 © 2006 American Chemical Society Published on Web 05/11/2006

5386 Langmuir, Vol. 22, No. 12, 2006

Bertorelle et al.

Figure 1. Procedure for the preparation of fluorescent magnetic nanoparticles.

2. Experimental Section

particles are dispersed in the organic phase. Further treatment is necessary to make them soluble in the aqueous phase for biological applications, and the particles must be vortexed to avoid aggregation due to their low solubility and critical size (30 nm). Fluorescence labeling of magnetic particles using citric acid has recently been described.15Citrate stabilization of magnetic nanoparticles in aqueous medium is well known.16 Citrate is adsorbed on the ferric oxide surface via one or two of the carboxylate moieties, allowing the use of one of the ungrafted carboxylic acid groups to bind a dye. However, in this case dye binding results in the partial annihilation of surface charges, which is demonstrated by electrophoresis measurements. Confocal fluorescence microscopy provides additional evidence of ferrofluid destabilization under a magnetic field. Several studies describe the immobilization of fluorescent markers in a layer of silica around the magnetic particles. For example, Lu et al.17 describe the incorporation of organic fluorescent dyes into silica shells by covalent coupling of the dye with the sol-gel precursor. The advantage of this synthesis is the presence of silanol groups that can interact with various coupling agents; however, the coating considerably increases the size of the nanoparticles (around 50 nm). In another study,11 the dye is an inorganic fluorophore that is introduced into a layer of silica. However, the saturation magnetization value of these nanocomposites (8.4 emu/g) dramatically decreases compared with that of uncoated particles (65 emu/g). In other words, the nanoparticles lose their magnetic properties. This work describes the synthesis of bifunctional particles that exhibit superparamagnetic and fluorescence properties. The superparamagnetic core of ferric oxide is covered by fluorescent dyes. The chemical synthesis is based on the covalent coupling of modified organic fluorophores with m-2,3-dimercaptosuccinic acid (DMSA), which strongly interact with the surface of the ferric oxide nanoparticles. Negative surface charges due to the acid-base behavior of grafted DMSA cause repulsive particle interactions and prevent aggregation. A stable colloidal dispersion of magnetic fluorescent nanoparticles (fluorescent ferrofluid) is obtained. The synthesis is described here for the common organic dyes rhodamine and fluorescein, but it may be extended to other fluorophores.

2.1. Synthesis and Physicochemical Characterization of Bifunctional Nanoparticles. All chemicals and reagents were purchased from Sigma. 2.1.1. Synthesis. The magnetic nanoparticles used in the present work were prepared according to Massart’s method.18,19 Colloidal magnetite was chemically oxidized to maghemite particles (γFe2O3) that have long-term stability in either alkaline or acidic media. The surface charge of the particles depends on the pH of the solution. Dispersion in water yields a stable colloidal solution called an ionic ferrofluid. The coupling of the fluorescent dye was performed on particles that were dispersed in an acidic medium. The process is based on the strong interaction between dimercaptosuccinic acid (DMSA) and the positively charged surface of γFe2O3 particles.20,21 DMSA is coupled to the particle by at least one carboxyl and one thiol. Ungrafted thiols are subsequently available for coupling, and unadsorbed carboxylates ensure surface charges. In the synthesis process described in Figure 1, the coupling between DMSA and the fluorescent dye can be made before or simultaneously with the one between DMSA and the magnetic particles. Two common fluorescent dyes were selected: rhodamine B and a fluorescein derivative, fluorescein diacetate maleimide. The molar ratios for the process were 10% DMSA/Fe and 25% dye/DMSA for rhodamine B or 20% for fluorescein. These ratios, corresponding to the maximum rate of coupling, were experimentally determined. Figure 1 summarizes the coupling procedure for rhodamine B and cystamine through amide bond formation. Seventeen milligrams of rhodamine B, 7 mg of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC), and 4 mg of cystamine were stirred in 25 mL of water for 3 h at pH 7. Then, 25 mg of DMSA and 1.25 mL of an acidic ferrofluid (1 M Fe) were added, and the mixture was stirred for 2 h. The resulting hybrid magnetic particles formed a flocculate, which was rapidly separated from the supernatant with a permanent magnet and washed twice with water. Next, 25 mL of water was added, the pH was increased to 9 by the addition of 1 M TMAOH, and the hybrid particles were stirred in this alkaline medium for 45 min. This step allows the redispersion of the magnetic particles in water. Then, the pH was adjusted to 7 with 1 M HCl. Finally, the particles were submitted to successive precipitation. NaCl was added to the solution until the destabilization of the ferrofluid occurred (∼0.1 g of NaCl powder for 10 mL of ferrofluid). At this time, turbidity appeared, and the precipitation of the particles was

(15) Sahoo, Y.; Goodarzi, A.; Swihart, M. T.; Ohulchanskyy, T. Y.; Kaur, N.; Furlani, E. P.; Prasad, P. N. J. Phys. Chem. B 2005, 109, 3879-3885. (16) Fauconnier, N.; Bee, A.; Roger, J.; Pons, J. N. Prog. Colloid Polym. Sci. 1996, 100, 212-216.

(17) Lu, Y.; Yin, Y.; Mayers, B. T.; Xia, Y. Nano Lett. 2002, 2, 183-186. (18) Massart, R. IEEE Trans. Magn. 1981, 17, 131. (19) Lefe´bure, S.; Dubois, E.; Cabuil, V.; Neveu, S.; Massart, R. J. Mater. Res. 1998, 13, 2975-2981.

Fluorescent Superparamagnetic Nanoparticles accelerated under a magnetic field gradient (a strong Nd-Fe-B magnet). Magnetic separation steps were made until the supernatant was colorless. The last dispersion step was made in Hepes buffer. For the fluorescein derivative, the process was slightly different. We used commercial fluorescein diacetate maleimide that reacts with the thiol groups of DMSA (Figure 1). Twenty-four milligrams of fluorescein diacetate maleimide was dissolved in 12 mL of acetone and added to 25 mg of DMSA (dissolved in 12 mL of water at neutral pH). The mixture was stirred for 4 h at room temperature. Acetone was then removed by evaporation, the volume made up to 25 mL, and the pH was adjusted to 11 with KOH. The solution was stirred for 2 h to allow for the hydrolysis of the acetate protecting group. The pH was lowered to 4 with HCl, and 1.25 mL of acidic ferrofluid (1 M Fe) was added. The remaining steps followed the rhodamine B procedure. 2.1.2. Determination of the [Dye]/[Fe] Ratio. The molar ratio of the dye and iron of the magnetic particles was determined after chemical decomposition of the coupling between DMSA and the particles in alkaline medium (pH 12, 4 h). The magnetic particles were then separated from the fluorescent supernatant by magnetic decantation. The dye concentration was deduced from ultravioletvisible absorption (Hitachi U2000) after dilution in phosphate buffer (pH 7 for rhodamine B, pH 10 for fluorescein). After chemical decomposition of the magnetic particles in acidic medium, the iron concentration was determined by atomic absorption spectrophotometry (Pelkin Elmer Analyst 100). 2.1.3. Analysis of the Size and Surface Charges of the Nanoparticles. The morphology of the nanoparticles was investigated by transmission electron microscopy (TEM, JEOL 100CXII). Colloid suspensions were deposited directly onto a carbon-coated copper grid. The mean particle size was determined by dynamic light scattering (Macrotron spectrometer Amtec, Malvern 7132 correlator card). Hydrodynamic diameter measurements were performed at 20 °C and angles of 45, 60, 90° after dilution in Hepes buffer. Each sample was analyzed for 1800 s. The surface charge of the nanoparticles was investigated through ζ-potential measurements (Zetasizer 4, Malvern Instruments). The ζ-potential measurement was performed in Hepes buffer so that the pH was 7.5. Because of the precipitation-dispersion cycles, the quantity of salt in the ferrofluid was low. The conductivity was low enough to allow good measurements of the ζ potential. 2.1.4. Quantification of the Fluorescence Properties. A steadyfluorescence study was performed on a Varian Cary-Eclipse spectrofluorometer. The emission spectrum of the magnetic fluorescent particles was compared to an equal molarity of unadsorbed dye in Hepes buffer. 2.1.5. Quantification of the Magnetic Properties. The magnetic properties of the fluorescent ferrofluid were determined using a vibrating magnetometer.22 The applied magnetic field was increased gradually up to 80 × 104 A/m. The magnetic measurement allowed the determination of two characteristics of the magnetic fluid: the volume fraction of magnetic particles φ (Ms ) msφ) and the parameters of the size distribution (d0, σ). The magnetic and fluorescence properties of the ferrofluid were observed by optical microscopy under a magnetic field. The magnetic water-in-oil emulsion was made with dioleoyl-sn-glycero-3-phosphocholine (DOPC) as the surfactant. 2.2. Cell Labeling. 2.2.1. Intracellular Uptake of Magnetic Nanoparticles. Our model cell system consists of human cervical cancer cells (HeLa) grown in DMEM culture medium supplemented with 10% inactivated fetal bovine serum, 50 units/mL penicillin, 40 mg/mL streptomycin, and 0.3 mg/mL L-glutamine. The mean cell diameter was 14.6 ( 0.8 µm. For fluorescence microscopy, cells were grown on glass coverslips for 2 days before incubation with (20) Fauconnier, N.; Pons, J. N.; Roger, J.; Bee, A. J. Colloid Interface Sci. 1997, 194, 427-433. (21) Halbreich, A.; Sabolovic, D.; Sestier, C.; Geldwerth, D.; Pons, J. N.; Roger, J. WO 97/01760 1997, European Patent EP 847528A1. Halbreich, A.; Sabolovic, D.; Sestier, C.; Geldwerth, D.; Pons, J. N.; Roger, J. U.S. Patent 6,150,181, 2000. (22) Foner, S.; McNiff, E. J. ReV. Sci. Instrum. 1968, 39, 171-170.

Langmuir, Vol. 22, No. 12, 2006 5387 rhodamine nanoparticles (5 mM Fe) for 1 h at 4 °C. Fluorescence images were captured using a Leica DMIRB microscope with a 63× oil-immersion lens and digital camera and were processed using Metaview software. To investigate membrane localization, the first sample was fixed immediately after the 4 °C incubation with 3% paraformaldehyde in phosphate-buffered saline (PBS). To restore the internalization pathway, cells of the second sample were fixed with 3% paraformaldehyde in PBS after the chase process: 2-h chase at 37 °C in RPMI culture medium followed by three washes with RPMI. To access the magnetic properties, the cells were submitted to a uniform field before and during fixation. For this third sample (magnetization of intracellular fluorescent endosomes), cells were fixed when submitted to an uniform magnetic field, B ) 80 × 103 A/m, after both the 1-h incubation at 4 °C and the 2-h chase at 37 °C. 2.2.2. Magnetophoresis Measurements. The migration of magnetic cells under a controlled magnetic field gradient provides access to the determination of the magnetic load and the determination of the cells’ mean velocity. Briefly, the magnetophoresis assay consists of following the movement of magnetically labeled cells toward a permanent magnet. In the observation window (6 mm from the magnet), the magnetic field (central value of 139 × 103 A/m) was directed toward the magnet and increased with a constant magnetic field gradient GradB ) 14.8 × 103 A/m/mm. Consequently, the magnetic force experienced by each cell had a constant value of Fm) NµGradB, where N is the number of nanoparticles attached to the cell and µ is the magnetic moment of each nanoparticle. When the permanent regime was reached, the magnetic force was exactly balanced by the viscous one: Fm ) Fv ) 6πηRV, where η is the viscosity of water, R is the cell radius, and V is the cell velocity. Practically, we measured the velocity of a hundred individual cells (directly proportional to the number of particles associated with the cell), resulting in the distribution of the cellular magnetic uptake.23

3. Results and Discussion 3.1. Physicochemical Characterization of the Bifunctional Nanoparticles. Using 0.05 M Fe, we found a [dye]/[Fe] ratio of 0.1% for rhodamine B and 1.2% for fluorescein, which corresponds to 10 and 120 dye molecules per particle, respectively. These results show that the coupling between the fluorescein derivative and the particles is more efficient than the two-stage rhodamine B procedure. However, the quantity of the rhodamine B ferrofluid is sufficient in terms of fluorescence, and this method of synthesis has the advantage of low cost. We have verified that the presence of the dye was due to a strong complexation and not to weak or electrostatic interactions. Test reactions (simply mixing dye and ferrofluid) showed that the unadsorbed dye was washed away by the flocculationredispersion cycles, and the washed sample no longer fluoresced. A transmission electron microscopy image of the fluorescent particles is shown in Figure 2 with a graph of their size distribution deduced from dynamic light scattering. The mean hydrodynamic diameter was 30 nm and represents 95% of the population. Very few aggregates were found. The zeta potential measurements were the same value (-36 mV) as for the particles coated with DMSA alone. Covalent coupling does not modify the electrostatic surface charges because the ratio of dye to iron is low and the dye is coupled via thiol groups, which do not contribute to surface charges. The emission spectrum of the magnetic fluorescent particles compared to an equal molarity of unadsorbed dye in Hepes buffer is shown in Figure 3. The spectra are similar, with a small red (23) Wilhelm, C.; Gazeau, F.; Roger, J.; Pons, J. N.; Bacri, J. C. Langmuir 2002, 18, 8148-8155.

5388 Langmuir, Vol. 22, No. 12, 2006

Bertorelle et al.

Figure 2. TEM image of rhodamine hybrid particles and their size distribution deduced from dynamic light scattering.

Figure 3. Excitation and emission spectra of the free dye (1) and the magnetic fluorescent particles (2) for fluorescein (A) and rhodamine B (B) in Hepes buffer.

shift for rhodamine (4 nm) due to the amide groups or to the interaction between the dye and the oxide nanoparticles. Peaks of emissions bands occurred at 516 nm for fluorescein and 578 nm for rhodamine B. The fluorescence intensity of the hybrid nanoparticles was 3.5 times lower than for fluorescein and two times lower than for rhodamine. Quenching occurs when fluorophores contact a metal surface, and it is possible that the same energy transfer occurs with metal oxide particles. Nevertheless, there is still sufficient emission for biological imaging. The magnetization curves (Figure 4A) were similar to those obtained for the initial dispersion of magnetic particles. The measurement was made in the fluid phase (ferrofluid); each magnetic particle bears a magnetic moment µ ) msV, where V is the volume of the particle and ms is the saturation magnetization of the grain material. The particles align themselves in the direction of the applied magnetic field. When the magnetic field is removed, the particles are subject to entropic energy. This magnetic behavior corresponds to a superparamagnetic comportment; there is no hysteresis loop. The magnetization (M) increases rapidly and

reaches saturation (Ms) at low magnetic field. The saturation magnetization values were 190 A/m (46 emu/g) for the dispersion of rhodamine-coated particles and 140 A/m (34 emu/g) for the dispersion of fluorescein-coated particles, corresponding to 0.05 and 0.07% volume fractions of magnetic particles, respectively. Experimental magnetization curves are correctly described assuming a log-normal distribution of particle size with a characteristic magnetic diameter d0 and a polydispersity index σ. The parameters of the particle size distribution were determined by an analysis of the shape of the magnetization curve. d0 and σ were 7.5 nm and 0.35, respectively, and were the same as for the initial ferrofluid. Moreover, the shape of the magnetization curve shows a high stability of the particles in the medium and under a magnetic field. To illustrate the magnetic and fluorescence properties of this fluid, a water-in-oil emulsion was prepared, with phospholipids (DOPC) as the surfactant. Fluorescence microscopy showed that the aqueous ferrofluid was localized inside the droplets and that the magnetic droplets aligned in the direction of the magnetic field (Figure 4B). The bifunctional nanoparticles exhibit a very high stability in aqueous solution, are superparamagnetic, and exhibit fluorescence properties. The particle size and surface charges are not modified compared to those of the initial ferrofluid, which is particularly important with regard to the internalization of the particles in cells. 3.2. Labeling of Living Cells. To probe the capacity of these hybrid nanoparticles for combined fluorescent and magnetic labeling of living cells, we investigated their interactions with cells in culture. At 4 °C, the dynamic internalization process of the cell is inhibited so that only interactions between nanoparticles and the cell membrane occur. The fluorescence pattern shown in Figure 5A demonstrates the binding of fluorescent nanoparticles onto the cell membrane. The whole membrane was labeled, with sites of enhanced fluorescence presumably due to clusters of nanoparticles. Warming the cells to 37 °C restored the internalization pathway, and the nanoparticles adsorbed on the cell membrane were chased to the cell interior as illustrated in Figure 5B. Internalization from their location on the membrane occurs via small vesicles, and the nanoparticles are further confined inside endosomes, appearing as fluorescent spots in the cell cytoplasm. Hence, binding a fluorescent dye onto magnetic nanoparticles enables their direct imaging and localization in living cells. Our observations using fluorescence microscopy are consistent with

Fluorescent Superparamagnetic Nanoparticles

Langmuir, Vol. 22, No. 12, 2006 5389

Figure 4. (A) Magnetization curve of rhodamine-coated particles dispersed in Hepes buffer. Ms is the saturation magnetization, and M is the magnetization. (B) Fluorescence microscopy image of a water-in-oil emulsion with the fluorescent ferrofluid. Phospholipids (DOPC) were used as the surfactant. Magnetic droplets form chains in the direction of the magnetic field. Bar ) 20 µm.

Figure 5. Use of fluorescence microscopy to monitor the localization patterns of magnetic nanoparticles in living cells. (A) First sample: cells were fixed immediately after 1 h of incubation at 4 °C with fluorescent magnetic nanoparticles. Fluorescence is visible on the cell membrane. (In this image, the middle of the cell is in focus.) (B) Second sample: cells were fixed after 1 h of incubation at 4 °C and 2 h of chase at 37 °C. The fluorescence now appears in small, localized spots in the cell interior. Top, fluorescence image; center, transmission image of the same cells; bottom, overlay of the top and central images, with the fluorescence colored red for clarity. Bar ) 10 µm; magnification ) 63×.

a previous study of the interactions of HeLa cells with DMSA nanoparticles (free of fluorescent dye) by transmission electron microscopy.23 DMSA-coated nanoparticles, with their tiny volume and negative surface charge, show a high electrostatic affinity for the cell membrane that in turn triggers a massive capture by cells (up to 107 nanoparticles per cell). Membrane adsorption initiates the internalization of nanoparticles within 100-nmdiameter vesicles, which is the first step of the endocytotic pathway that is the usual mechanism for cells to ingest

extracellular materials. Nanoparticles are then delivered to larger 0.6-µm-diameter membrane organelles or endosomes. Our current study with fluorescent nanoparticles allows the visualization of the first step (membrane adsorption) and the final step (nanoparticles densely confined within endosomes) of this pathway. To assess the magnetic properties of these fluorescent endosomes, cells were submitted to a uniform field before and during fixation (Figure 6). The magnetic nanoparticles enclosed in the endosomes confer a magnetic moment to these biological

5390 Langmuir, Vol. 22, No. 12, 2006

Bertorelle et al.

Figure 6. Observation of chains of magnetic fluorescent endosomes inside the cell. Third sample: cells were fixed when submitted to a magnetic field after 1 h of incubation at 4 °C and 2 h of chase at 37 °C. Endosomes forming chains were observed in both the fluorescence image (top) and the transmission image (center). Fluorescent chains (red) co-localize with chains observed in transmission in the overlay image (bottom). Bar ) 10 µm; magnification ) 100×.

vesicles when submitted to a magnetic field. Magnetic endosomes behave as small magnets and attract each other via dipoledipole interactions, forming small chains in the cell cytoplasm. The formation of chains of fluorescent endosomes demonstrates the association of both magnetic and fluorescence properties within the endosome and emphasizes the association of the fluorophore with the nanoparticle magnetic core. These magnetic fluorescent endosomes are new relevant biological tools designed to track the dynamics of endocytotic vesicles in the presence of magnetic forces and are a continuation of the study described by Wilhelm et al.24 We have estimated the average magnetic load by measuring the velocity of magnetic cells when submitted to a magnetic field gradient.25 This velocity results from the balance between the magnetic and viscous forces and can be directly converted to the number of nanoparticles N attached to the cell. We find a mean load of N ) (4.7 ( 0.8) × 106 particles per cell; that is, 5.8 ( 0.9 pg of Fe per cell (Figure 7A), the same as obtained for DMSA particles.23 Furthermore, cell migration triggered by a magnetic field gradient was observed by fluorescence microscopy (Figure 7B). For a short exposure time (0.2 s), the cells appear rounded and move with a mean velocity of 49 ( 9 µm/s, which is measurable between two images (time interval of 2 s). For a longer exposure time (1 s), we observed the cell trajectory, appearing as a comet with a length directly reflecting cell velocity. (24) Wilhelm, C.; Browaeys, J.; Ponton, A.; Bacri, J. C. Phys. ReV. E 2003, 67, 011504. (25) Wilhelm, C.; Gazeau, F.; Bacri, J. C. Eur. Biophys. J. 2002, 31, 18-125.

Figure 7. (A) Distribution of the cellular magnetic load quantified by magnetophoresis. The number of particles per cell was deduced from the velocity of individual magnetic cells driven into movement in the direction of a magnetic field gradient. Cells were loaded on average with millions of fluorescent magnetic nanoparticles. (B) Observation by fluorescence microscopy of the cell migration toward high magnetic field, with two different exposure times: 0.2 s (top) and 1 s (bottom). Magnification ) 20×.

This magnetophoretic mobility that is controlled by external gradients can be designed for in vitro cell sorting, immunomagnetic separation, or in vivo targeting of cells (e.g., stem cells) to organs of interest (e.g., in the framework of cellular grafts or cell-based therapy). In conclusion, the anionic bifunctional nanoparticles described here are very efficient candidates for mixed fluorescent and magnetic cell labeling because of their electrostatic interactions with the cell membrane. This nonspecific interaction would allow localized labeling of a variety of cells including lymphocytes and stem cells. In addition to in vitro applications for cell biology, the combined properties of fluorescence and magnetism associated with nanoparticles offer new opportunities for in vivo imaging.

Fluorescent Superparamagnetic Nanoparticles

Magnetic labeling has recently been used to noninvasively monitor cells injected into live animals by MRI.26 Magnetic fluorescent nanoparticles serve both as magnetic resonance contrast agents for MRI and optical probes for intravital fluorescence micros(26) Rivie`re, C.; Boudghene, F.; Gazeau, F.; Roger, J.; Pons, J. N.; Laissy, J. P.; Allaire, E.; Michel, J. B.; Letourneur, D.; Deux, J. F. Radiology 2005, 235, 959-967.

Langmuir, Vol. 22, No. 12, 2006 5391

copy.8 Moreover, the development of optical tomographic imaging techniques in vivo (for example, fibered confocal fluorescence microscopy) offers the prospect of a second imaging modality with bifunctional nanoparticles, providing a higher spatial resolution and specificity, and MRI is used to visualize the anatomical background. LA052710U