Langmuir 2007, 23, 12329-12336
12329
Cellular Level Loading and Heating of Superparamagnetic Iron Oxide Nanoparticles Venkat S. Kalambur,† Ellen K. Longmire,‡ and John C. Bischof*,† Department of Mechanical Engineering, UniVersity of Minnesota, 111 Church Street SE, Minneapolis, Minnesota 55455, and Department of Aerospace Engineering and Mechanics, UniVersity of Minnesota, 110 Union Street SE, Minneapolis, Minnesota 55455 ReceiVed April 15, 2007. In Final Form: June 15, 2007 Superparamagnetic iron oxide nanoparticles (NPs) hold promise for a variety of biomedical applications due to their properties of Visualization using magnetic resonance imaging (MRI), heating with radio frequency (rf), and moVement in an external magnetic field. In this study, the cellular loading (uptake) mechanism of dextran- and surfactant-coated iron oxide NPs by malignant prostate tumor cells (LNCaP-Pro5) has been studied, and the feasibility of traditional rf treatment and a new laser heating method was evaluated. The kinetics of cell loading was quantified using magnetophoresis and a colorimetric assay. The results showed that loading of surfactant-coated iron oxide NPs with LNCaP-Pro5 was saturable with time (at 24 h) and extracellular concentration (11 pg Fe/cell at 0.5 mg Fe/mL), indicating that the particles are taken up by an “adsorptive endocytosis” pathway. Dextran-coated NPs, however, were taken up less efficiently (1 pg Fe/cell at 0.5 mg Fe/mL). Loading did not saturate with concentration suggesting uptake by fluid-phase endocytosis. Magnetophoresis suggests that NP-loaded cells can be held using external magnetic fields in microcirculatory flow velocities in vivo or in an appropriately designed extracorporeal circuit. Loaded cells were heated using traditional rf (260A, 357 kHz) and a new laser method (532 nm, 7 ns pulse duration, 0.03 J/pulse, 20 pulse/s). Iron oxide in water was found to absorb sufficiently strongly at 532 nm such that heating of individual NPs and thus loaded cells (1 pg Fe/cell) was effective (10 pg Fe/cell) and longer duration (30 min) when compared to laser to accomplish cell destruction (50% viability at 10 pg Fe/cell). Scaling calculations show that the pulsed laser method can lead to single-cell (loaded with NPs) treatments (200 °C temperature change at the surface of an individual NP) unlike traditional rf heating methods which can be used only for bulk tissue leVel treatments. In a mixture of normal and NP-loaded malignant tumor cells, the malignant cells were selectively destroyed after laser exposure leaving the unloaded normal cells intact. These studies hold promise for applications in cell purification and sorting and extracorporeal blood treatments in vitro.
1. Introduction Superparamagnetic iron oxide nanoparticles (NPs) hold promise for a variety of biomedical applications due to their properties of Visualization under magnetic resonance imaging (MRI), heating with radio frequency (rf), and moVement in an external magnetic field.1,2 Biomedical applications of these NPs include their use for magnetic separations of biomolecules in vitro,3 organ specific MR contrast agents in vivo,4 and fundamental biophysical studies.5 There is a renewed interest in using iron oxide NPs as multifunctional platforms for cell purifications in vitro, cellular and molecular imaging,6,7 and early (cellular level) tumor diagnosis and treatments in vivo.8 * To whom correspondence should be addressed. E-mail: bischof@ tc.umn.eduep. Phone: 612-625-5513. Fax: 612-625-4344. † Department of Mechanical Engineering. ‡ Department of Aerospace Engineering and Mechanics. (1) Kalambur, V. S.; Han, B.; Hammer, B. E.; Shield, T. W.; Bischof, J. C. Nanotechnology 2005, 16, 1221-1233. (2) Pankhurst, Q. A.; Connolly, J.; Jones, S. K.; Dobson, J. J. Phys. D 2003, 36, R167-R181. (3) Safarik, I.; Safarikova, M. J. Chromatogr., B 1999, 722, 33-53. (4) Weissleder, R.; Bogdanov, A.; Neuwelt, E. A.; Papisov, M. AdV. Drug DeliVery ReV. 1995, 16, 321-334. (5) Wilhelm, C.; Gazeau, F.; Bacri, J.-C. Phys. ReV. E 2003, 67, 6190861912. (6) Bulte, J. W.; Douglas, T.; Witwer, B.; Zhang, S. C.; Strable, E.; Lewis, B. K.; Zywicke, H.; Miller, B.; van Gelderen, P.; Moskowitz, B. M.; Duncan, I. D.; Frank, J. A. Nat. Biotechnol. 2001, 19, 1141-1147. (7) Perez, J. M.; Josephson, L.; Weissleder, R. ChemBioChem 2004, 5, 261264. (8) Mornet, S.; Vasseur, S.; Grasset, F.; Duguet, E. J. Mater. Chem. 2004, 14, 2161-2175.
In order to translate NP findings to applications for diagnosis and therapy, quantitative characterization of NP loading and selective destruction of NP-loaded cells by heat treatment are necessary. For this purpose assays to quantify cell loading need to be developed. These studies require comparison with kinetic models to help better predict loading and distribution behavior in cellular systems.9 Previous studies have used transmission electron microscopy (TEM) in cell mixtures to study the mechanism of loading and its dependence on NP coating.10-12 These studies showed that uncoated NPs were primarily localized in endosomes by a binding-internalization mechanism. In contrast, particles functionalized with specific membrane receptors (e.g., insulin,10 transferrin,11 and lactoferrin13) were found clustered on the cell membranes with no internalization. Other bulk methods used for quantifying cell loading include studying changes in T2 relaxation times in MR,14 radioactivity changes of labeled iron oxide NPs,15 atomic emission spectroscopy of elements in a sample,9 and electron spin resonance (detection of species with unpaired electrons like iron oxide) based detection (9) Chithrani, B. D.; Ghazani, A. A.; Chan, W. C. Nano Lett. 2006, 6, 662668. (10) Gupta, A. K.; Berry, C.; Gupta, M.; Curtis, A. IEEE Trans. NanoBiosci. 2003, 2, 255-261. (11) Berry, C. C.; Charles, S.; Wells, S.; Dalby, M. J.; Curtis, A. S. Int. J. Pharm. 2004, 269, 211-225. (12) Gupta, A. K.; Gupta, M. Biomaterials 2005, 26, 3995-4021. (13) Gupta, A. K.; Curtis, A. S. G. Biomaterials 2004, 25, 3029-3040. (14) Frank, J. A.; Miller, B. R.; Arbab, A. S.; Zywicke, H. A.; Jordan, E. K.; Lewis, B. K.; Bryant, L. H., Jr.; Bulte, J. W. Radiology 2003, 228, 480-487. (15) Moore, A.; Marecos, E. M.; Bogdanov, A., Jr.; Weissleder, R. Radiology 2000, 214, 568-574.
10.1021/la701100r CCC: $37.00 © 2007 American Chemical Society Published on Web 10/26/2007
12330 Langmuir, Vol. 23, No. 24, 2007
of iron oxide.16 Recently Wilhelm and co-workers16,17 reported on the efficient loading of anionic and albumin adsorbed iron oxide NPs into HeLa cells using magnetophoresis. Their studies on the effect of surface coating on the loading of NPs showed that anionic particles accumulate at a greater rate than albuminfunctionalized NPs. They also showed that negatively charged particles bind to the cell membrane and are then internalized by a process they refer to as “adsorptive endocytosis”. Following NP loading, heat treatment can be used for selective cell destruction. Traditionally, heating of iron oxide NPs has been accomplished using rf fields primarily for in vivo hyperthermia applications.18 The method requires concentrations of >1 mg Fe/g and rf fields applied continuously for time scales of 15-30 min.1,8 Additionally, the method has been shown in a number of theoretical studies to require NPs that are distributed over a millimeter-sized region19 with no effects at the cellular level. For a single cell loaded with NPs (gold), theoretical calculations show that “localization of temperature effects at the nanoscale” (or at the level of single NP or NP clusters) occurs only after heating by high-power optical pulses, typically of very short pulse duration (approximately nanoseconds).20 Although these scaling arguments20 suggest heating at the single NP or cluster level by short laser pulses, no analogous experimental studies have been done with iron oxide NPs. In this study the following questions were addressed: (i) What are the mechanisms by which tumor cells load dextran- and surfactant-coated iron oxide NPs? (ii) Can heating of iron oxide NPs be achieved at the cellular level? (iii) Does heat treatment result in selective destruction of NP-loaded cells? The kinetics of loading of dextran- and surfactant-coated iron oxide NP by prostate tumor cells (LNCaP-Pro5) were quantified by using magnetophoresis, a single-cell assay based on a balance of viscous and magnetic forces and verified by a bulk colorimetric assay. Furthermore, the feasibility of rf and laser heat treatment for selective destruction of loaded cells was evaluated. 2. Materials and Methods 2.1. Magnetic Nanoparticles. Two different iron oxide (Fe3O4, magnetite) NPs with the same mean diameter were used in this study. Aqueous suspensions of surfactant-coated (Ferrotec Inc., NH) and dextran-coated (Liquids Research Inc., U.K.) NPs with a mean diameter of ∼10 nm were used for this study. Surfactant-coated NPs are used for magnetic hyperthermia in vivo,1 while dextran-coated NPs are used as in vivo MR contrast agents.21 2.2. Cell Culture, Time-Points, and Concentrations. Prostate cancer cell lines (lymph node cancer of the prostate-Pro5, LNCaPPro5) were used for all loading, rf, and laser treatment studies. The cells were cultured in Dulbecco’s modified Eagle medium (DMEM), F12 supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin (P/S, 100 units/mL penicillin, and 100 µg/ mL streptomycin), 10-9 M DHT, and 1.2 g/L sodium bicarbonate. For all studies, cells were grown overnight as monolayers in 25 cm2 T-flasks at a concentration of 0.8-1.0 × 106 cells/flask in 5 mL of cell media, 37 °C, and 5% CO2. For the assay quantification, NPs were added the following day at various concentrations (0, 0.05, 0.1, 0.5, 1 mg Fe/mL; both NPs) and studied at time-points (1, 6, 24, 48, 72 h; surfactant-coated NP only). Monolayer cultures were washed (16) Wilhelm, C.; Gazeau, F.; Roger, J.; Pons, J. N.; Bacri, J.-C. Langmuir 2002, 18, 8148-8155. (17) Wilhelm, C.; Billotey, C.; Roger, J.; Pons, J. N.; Bacri, J. C.; Gazeau, F. Biomaterials 2003, 24, 1001-1011. (18) Jordan, A.; Scholz, R.; Wust, P.; Fahling, H.; Krause, J.; Wlodarczyk, W.; Sander, B.; Cogl, T.; Felix, R. Int. J. Hyperthermia 1997, 13, 587-605. (19) Rabin, Y. Int. J. Hyperthermia 2002, 18, 194-202. (20) Keblinski, P.; Cahill, D. G.; Bodapati, A.; Sullivan, C. R.; Taton, T. A. J. Appl. Phys. 2006, 100, 054305. (21) Shen, T.; Weissleder, R.; Papisov, M.; Bogdanov, A., Jr.; Brady, T. J. Magn. Reson. Med. 1993, 29, 599-604.
Kalambur et al.
Figure 1. Assays to study cell loading. (a) Magnetophoresis: magnetic force balances viscous drag on NP-loaded cells moving in an external gradient, dB/dx. The velocity of an individual cell is measured and used to calculate the number of NPs, N, loaded in a cell. (b) Colorimetric chemical assay: purple-colored complex obtained after reaction of ferrozine with Fe2+ in standards. Iron oxide NP loaded cells are dissolved in acid, complexed to ferrozine, assessed at 562 nm in a spectrophotometer, and compared to standards to determine cell loading. with HBSS, trypsinized (0.25 mL, 0.05% trypsin/0.53 mM EDTA), and maintained as suspensions in 5 mL of media before further experimentation. Studies on selective cell destruction using laser pulses were performed on a 1:1 mixture of LNCaP-Pro5 and smooth muscle cells (SMC). The SMC cell media contained DMEM and F12 with 10% FBS, 1% P/S, 0.5% L-glutamine, and 1.2 g/L sodium bicarbonate. 2.3. Assays to Study Cellular Loading. 2.3.1. Magnetophoresis. Magnetophoresis is the movement of a cell loaded with NPs through a fluid due to a magnetic force16,22 (Figure 1a). For a one-dimensional magnetic field gradient, the magnetic force balances the viscous drag for an NP-loaded cell moving in a fluid due to a magnetic field gradient. The force balance in an external field gradient,22 dB/dx (T/m), is given by mB
dB ) 6πηRcellV dx
(1)
where mB is the total magnetic moment of the cell loaded with NPs, Rcell is the radius of a single cell, and η is the dynamic viscosity of the cell media. For measuring the velocity V of cells, a rectangular well of dimensions (20 mm × 5 mm) was constructed on a glass microslide (76 mm × 25 mm, Becton Dickinson Labware, NJ) by marking the boundaries with nail polish. Around 40-45 µL aliquots of NP-loaded cells (2 × 106 cells in 5 mL) were placed in this well. A rectangular NdFeB permanent magnet (25 mm × 10 mm × 5 mm, N42 grade [Br ) 13 000 G], Indigo Instruments, ON) was placed such that the center-line of the permanent magnet coincided with the boundary of the well. The gradients were calculated using an analytical expression for the magnetic field from a permanent magnet.1,23 Upon magnet placement, the movement of individual cells was imaged in bright-field (optical magnification 10×) under (22) Zborowski, M.; Moore, L. R.; Williams, P. S.; Chalmers, J. J. Sep. Sci. Technol. 2002, 37, 3611-3633. (23) Hatch, G. P.; Stelter, R. E. J. Magn. Magn. Mater. 2001, 225, 262-276.
Cell Loading of Iron Oxide Nanoparticles
Langmuir, Vol. 23, No. 24, 2007 12331
Figure 2. Setup for rf and laser heating. (a) rf: the samples are placed at the center of a water-cooled 3-turn rf coil (20 mm diameter) connected to an ac power source. The field parameters are 260 A, 357 kHz, and 30 min. (b) Laser: cells are placed in a glass capillary tube before exposure to pulsed laser irradiation at 532 nm, 7 ns pulse duration, 20 pulse/s, 0.03 J/pulse. The total length of pulsing at each position was 30 s. a microscope (Olympus BX-50, Tokyo, Japan) and taped using a video recorder (JVC SR-VS10U, Japan). By a playback of tapes and measurements of times of individual cells to move a known distance, the velocity of cells was calculated. The number of NPs per cell, N, was then calculated as N)
36ηRcellV bMsdparticle3 dB/dx
(2)
where Ms is the saturation magnetization of iron oxide, dparticle is the diameter of a single NP, and the constant b represents the ratio of the net magnetization of the NP to its saturation magnetization in an external field. A value of 0.8 is used for b based on experimental values for Fe3O4 NPs obtained elsewhere.24 The mean radius of the cells, Rcell, was calculated from the playback of the video recordings. For this study, η and Ms are 0.001 Pa‚s (dynamic viscosity of water at 298 K) and 4.5 × 105 A/m (saturation magnetization of magnetite),25 respectively. Using a density value of magnetite, FFe3O4, to be 5170 kg/m3 and 72% iron in Fe3O4, single-cell loading is calculated as πdparticle3 6
mFe ) 0.72NFFe3O4
(3)
The concept of magnetophoresis can be extended for scaling the ability to manipulate and hold NP or NP-loaded cells within in vivo microcirculatory environments using external magnetic fields (see the Supporting Information, Figure 3). 2.3.2. Colorimetric Chemical Assay. The colorimetric assay is a bulk method that relies on the spectrophotometric detection of the colored complex formed after reaction of Fe2+ with ferrozine (Figure 1b).26 Cells, resuspended in 2 mL of distilled water, were incubated at 65-70 °C for around 2 h after addition of 0.5 mL of 1.2 M HCl and 0.2 mL of 2 M ascorbic acid. Following this, 0.2 mL of a reagent containing 6.5 mM ferrozine (Hach Company, CO), 13.1 mM neocuproine (Sigma Aldrich, MO), 2 M ascorbic acid (Alfa Aesar, MA), and 5 M ammonium acetate (Fisher Biotech, NY) was added. (24) Goya, G. F.; Berquo, T. S.; Fonseca, F. C.; Morales, M. P. J. Appl. Phys. 2003, 94, 3520-3528. (25) Rosensweig, R. E. J. Magn. Magn. Mater. 2002, 252, 370-374. (26) Fish, W. W. Methods Enzymol. 1988, 158, 357-364.
The resultant solution was left at room temperature (25 °C) for 30 min and centrifuged (1000×, 5-6 min). The total absorbance of the supernatant was assessed at 562 nm using a UV-vis spectrophotometer (path length ) 10 mm, Beckman DU64 spectrophotometer, Beckman Instruments, CA), and the concentration of iron oxide was calculated by comparing to standards. A standard curve was obtained using ferrous ammonium sulfate (Hach Company, CO) at final concentrations of 0.1, 0.2, 0.5, 1, 2, and 5 µg Fe/mL. Distilled water solution with all the above reagents served as a blank. 2.4. Heat Treatment of Cells. 2.4.1. Radio Frequency Heating of Cells. LNCaP-Pro5 cells incubated with surfactant-coated NP for 48 h at concentrations of 0, 0.05, 0.1, 0.5 mg Fe/mL were centrifuged (1000×, 10 min, Beckman GS-6R centrifuge, Beckman-Coulter Inc., CA) in microcentrifuge tubes (1.5 mL polypropylene tubes, 40 mm × 10 mm, Dot Scientific Inc., MI). The cell pellets were placed in an ac magnetic field (1 kW Hotshot, Ameritherm Inc., NY) at the center of a 3-loop rf coil (20 mm diameter, 3 turns, 7 mm distance between turns) and treated at 260 A, 357 kHz for 30 min (Figure 2a). After treatment they were immediately resuspended in 1 mL of media. Cell viability before and after treatments was measured using Hoechst and propidium iodide (PI) staining (see section 2.5). 2.4.2. Laser Heating. LNCaP-Pro5 cells (106 cells in 30 µL) were placed inside glass capillary tubes (5 cm length, i.d. 870 µm, volume 30 µL, Boreal Labs Inc., ON) (Figure 2b). Glass capillaries were chosen due to higher transmittance at 532 nm27 compared to plastic microcentrifuge tubes. The capillary tubes were filled with cells (2 × 106 cells/40 µL; cytocrit of 22% assuming a cell diameter of 20 µm) using a pipet (Eppendorf Repeater 4780, Eppendorf Inc., NY) with the tube ends press fitted to the pipet tips. Each tube was attached to a glass microslide so that individual sections were sequentially exposed to the laser (532 nm, Q-switched, Ultra CFR Nd:YAG laser system, Big-Sky Laser, MT) in 3 mm steps as shown in Figure 2b. The laser parameters were 3 mm beam diameter, 20 pulse/s, 7 ns pulse duration, 30 s irradiation time at each location, 0.03 J/pulse. Cell viabilities were assessed before and after laser exposure using a membrane permeability assay with Hoechst and PI (see section 2.5). 2.4.3. Laser: SelectiVe Cell Destruction. Studies of selective cell destruction after laser exposure were performed in a mixture of (27) Siegel, R.; Howell, J. R. Thermal Radiation Heat Transfer, 4th ed.; Taylor and Francis: New York, 2002.
12332 Langmuir, Vol. 23, No. 24, 2007
Kalambur et al.
malignant tumor (LNCaP-Pro5) and normal (SMC) cells. Two samples were prepared for this study. Sample 1 contained a 1:1 mixture of unloaded SMCs and LNCaP-Pro5 cells. Sample 2 contained both cells at a similar ratio with only the LNCaP-Pro5 cells loaded with surfactant-coated NP (0.5 mg Fe/mL extracellular concentration, 48 h). Before laser exposure, the SMCs were labeled with calcein AM (1 mg/mL, λexcitation ) 488 nm, λemission ) 530 nm, Molecular Probes, Eugene, OR) to distinguish them from LNCaPPro5 cells. SMCs were mixed with calcein at a final concentration of 1:1000 and incubated for 30 min at 37 °C and 5% CO2. After washing, calcein-labeled SMCs and LNCaP-Pro5 cells were mixed in suspension at a ratio of 1:1. The cell mixtures were then placed in capillary tubes and exposed to laser irradiation as before (see section 2.4.2). Cell viabilities before and after laser exposure were calculated using Hoechst and PI staining (see section 2.5). 2.4.4. Scaling Calculations of Heating: Radio Frequency Heating. Temperature changes after rf heating were scaled using the heat diffusion equation (see the Supporting Information, Figure 1a). The temperature change, ∆T1(r e R) inside a spherical region of radius R containing the NPs is given by ∆T1(r) )
(
(
))
R2SAR k1 1 4r2 + 1- 2 4k1 3k2 6 R
(4)
where k1 and k2 represent the thermal conductivity for the region containing the NPs and for the surrounding medium (water at 298 K), respectively, and SAR (specific absorption rate) is the heat generation (W/m3) due to the NPs. SAR (W/m3) is calculated as SAR ) VfSARmax
(5)
where Vf is the volume fraction of NPs. A theoretical value of 109 W/m3 is assumed for SARmax.25 The scaling calculations provide an upper limit for temperatures changes as typical experimental measurements of rf heating lead to SAR values that are at least 4-5 orders smaller than these theoretical vales.1,28 2.4.5. Scaling Calculations of Heating: Laser Heating. Estimates of the temperature rise of a single NP can be made based on measured temperature change of NP solutions after laser heating (Supporting Information, Figure 5, parts b and c). First, an isolated system of NPs in water is assumed. Second, heat diffusion from the NP to the solution during the pulse is neglected (diffusion length ∼ size of NP) as previously reported.29 Third, water does not absorb strongly at 532 nm; however, some container absorption led to a temperature increase in water (i.e., solution devoid of NPs). Thus the meaH2O+NP sured sensible heat rise of the NP solution (FeffCp,eff∆Teff,1pulse ) is the sum of the contributions of the estimated rise of the NPs (VfFNPCp,NP∆TNP,1pulse) and measured rise of water ((1 - Vf)FH2OCp,H2O∆TH2O,1pulse). This can be expressed as follows: H2O+NP FeffCP,eff∆Teff,1pulse ) (1 - Vf)FH2OCP,H2O∆TH2O,1pulse + VfFNPCP,NP∆TNP,1pulse (6)
The specific heat and density of the NP are CP,NP and FNP, respectively. CP,H2O and FH2O are the specific heat and density of water, respectively. The specific heats of iron oxide and gold are 670 and 129 J/kg K, while the densities are 5100 and 19 300 kg/m3, respectively. The effective densities (Feff) and specific heats (CP,eff) at different NP volume fractions are essentially that of water since NP volume fractions (Vf) are ,1. Having estimated the temperature rise with a given NP, we can use this to establish an absorption coefficient for the NP. The temperature change of a single NP after individual laser pulse is related to its absorption coefficient by the following equation:29 (28) Jordan, A.; Wust, P.; Fahling, H.; John, W.; Hinz, A.; Felix, R. Int. J. Hyperthermia 1993, 9, 51-68. (29) Zharov, V. P.; Mercer, K. E. y.; Galitovskaya, E. N.; Smeltzery, M. S. Biophys. J. 2006, 90, 619-627.
∆TNP,1pulse )
RI0 τ FNPCNP
(7)
where τ is the pulse duration (7 ns) and R is the absorption coefficient (m-1) of iron oxide at 532 nm. The laser is assumed to have a uniform intensity, I0 (W/m2) given by I0 )
Epulse 4 τ πd
2
(8)
beam
where Epulse is the energy of a single pulse (0.03 J) measured using a power meter and dbeam is the beam diameter 3 mm (as set on the instrument). This allows R estimated from eq 7 to be compared to R measured by a bulk absorption measurement with a UV-vis spectrophotometer (see the Supporting Information, section 1.2). 2.5. Cell Viability. The viability of the cells after cellular loading and heat treatment (rf or laser) was studied by staining with Hoechst 3342 and PI, a standard membrane permeability test.30 Hoechst 3342 (10 mM, MW ) 561.9, Sigma Inc., MO, λexcitation ) 350 nm, λemission ) 461 nm) stains the nuclei of all cells, whereas PI (1 mg/mL, MW ) 668.4, Molecular Probes, Eugene, OR, λexcitation ) 488 nm, λemission ) 617 nm) stains only dead cells. Cells in suspensions were mixed with a dye reagent containing Hoechst 3342 and PI at a final ratio of 1:1000 and 5:1000. The cells were incubated for 15 min at 37 °C and 5% CO2. The number of dead cells (staining for PI) versus the total number of cells (staining for Hoechst 3342) were counted under a fluorescence microscope (Olympus BX-50, Tokyo, Japan). Multiple fields (8-10) were counted for each NP concentration. A total of 200-300 cells were counted for each NP concentration. 2.6. Statistics. Each test was performed on three independent samples. The results are represented as mean ( standard deviation. Statistical significance between samples was performed using Student’s t test. A p value of less than 0.05 was considered to be significant.
3. Results 3.1. Cell Loading Assays. Nanoparticle loading at the cellular level was assessed by magnetophoresis. Figure 3 shows the velocities and corresponding NP loading distributions in LNCaPPro5 cells which were loaded with surfactant-coated NPs for 48 h at an extracellular concentration of 0.5 mg Fe/mL. The velocity of 386 cells was measured in this experiment. A log-normal distribution of velocity and cellular loading was observed. A similar distribution was observed for all experiments of magnetophoresis (data not shown). Interestingly these velocities were consistent with microcirculatory flows (1-100 mm/s), suggesting that it is feasible to manipulate and hold NP-loaded cells in vascular flow or appropriately designed extracorporeal flows using magnetic fields in vivo (see the Supporting Information, Figure 3). Cellular loading assessed by magnetophoresis was confirmed by a bulk colorimetric assay. Table 1 shows the loading of dextrancoated NP at 48 h using the two assays. Both assays show NP loading that increases in a dose-dependent manner with extracellular NP concentration. Since no statistically significant differences (p ) 0.21, 0.22, and 0.4 at 0.05, 0.1, and 1 mg Fe/ mL, respectively) between the assays were observed, we chose magnetophoresis for quantifying the kinetics and coating dependences as the assay is based on single-cell measurements. 3.2. Loading Dependence: Kinetics and Surface Coating. Figure 4a demonstrates that surfactant-coated NP loading on LNCaP-Pro5 cells saturates with time. The cellular loading increases up to 24 h at an extracellular concentration of 0.1 mg Fe/mL after which saturation is observed (mean cellular loadings 1.3, 1.4, and 1.3 pg Fe/cell at 24, 48, and 72 h, respectively). (30) He, X.; Bischof, J. C. Ann. Biomed. Eng. 2005, 33, 502-510.
Cell Loading of Iron Oxide Nanoparticles
Langmuir, Vol. 23, No. 24, 2007 12333
Figure 3. Cellular loading using magnetophoresis. Log-normal distribution of velocity and mass of iron in individual LNCaP-Pro5 cells for a single experiment (0.5 mg Fe/mL, 48 h, surfactant-coated NP, N ) 386 cells). Table 1. Comparison of NP Loading Measurements by Magnetophoresis and Colorimetric Assaya loading at 48 h (pg Fe/cell) extracellular concn (mg Fe/mL)
magnetophoresis
colorimetric assay
0.05 0.1 1
0.3 ( 0.03 0.56 ( 0.06 1.41 ( 0.13
0.73 ( 0.51 0.93 ( 0.44 2.02 ( 1.13
a LNCaP-Pro5 cells are loaded at 48 h with dextran-coated NPs. Results are presented as mean ( standard deviation of three replicate measurements.
This suggests that “steady-state” loading (saturation with time) is observed at around 24 h for this particle. Figure 4b indicates that NP loading on to LNCaP-Pro5 cells is strongly dependent on NP coating. At 48 h, surfactant-coated NP loading to LNCaP-Pro5 cells was saturable with extracellular concentration, whereas dextran-coated NP loading increased linearly with concentration (Figure 4b, inset). Saturation for surfactant-coated NPs was observed at 0.5 mg Fe/mL, with a maximum of 11.6 ( 2 to 11.6 ( 3.8 pg Fe/cell at 0.5-1 mg Fe/mL extracellular concentrations. Last, the loading seen for the surfactant-coated NP was an order of magnitude higher (11.6 pg/cell at 1 mg Fe/mL extracellular concentration) when compared to the dextran-coated NP (1.4 pg/cell at 1 mg Fe/mL extracellular concentration). 3.3. Radio Frequency versus Laser Heating of Cells after Iron Oxide NP Loading. Due to higher heating in solution (Table 2) and better cell loading (Figure 4b) by surfactant-coated NP compared to dextran-coated NP, we chose to focus on the former for cellular heating studies. Surfactant-coated NPs exhibit temperature changes that are 15-fold higher than dextran-coated NPs after rf heating (150 A, 175 kHz, 15 min, 3.7 mg Fe/mL) in water (Table 2). Surfactant-coated NP heating was higher (see the Supporting Information, Figure 5a) by 10 °C (0.4 mg Fe/mL) compared to dextran-coated NPs in water after laser exposure (45 s total heating). Figure 5a demonstrates the effectiveness of laser heating at 532 nm compared to rf at 260 A, 357 kHz for killing cells at lower levels of NP loadings and shorter times. Mean cellular loadings were obtained using magnetophoresis (Figure 4b). For NP-loaded cells (1.3 pg Fe/cell), the mean cell viabilities decreased significantly (p < 0.05) from 95.4 ( 3.1 to 13.7 ( 8% after laser exposure. At the same loading, changes in viability were smaller with rf treatment (95.4% ( 3.1% to 89.9% ( 3.1%). To achieve reductions in viability to 50% with rf heating, NP levels an order of magnitude higher (11 pg Fe/cell) were needed. In addition,
longer time periods (30 min) were needed for rf treatment to achieve this reduction in viability when compared to laser heating (30 s). 3.4. Selective Heat Treatment of Cells. Figure 5b shows that malignant cells loaded with NPs can be selectively destroyed leaving unloaded normal cells intact. Sample 1 containing unloaded LNCaP-Pro5 (malignant tumor) and SMC (normal) cells showed no significant change (p ) 0.74 for SMC and p ) 0.27 for LNCaP-Pro5) in viabilities after laser exposure. However, the viabilities of NP-loaded LNCaP-Pro5 (malignant) cells in sample 2, containing a mixture of unloaded SMCs and loaded LNCaP-Pro5 cells, decreased significantly (p < 0.05) from 92% to 14%. No significant differences were observed for the unloaded SMCs (p ) 0.94) in sample 2. These results suggest that heating maybe occurring at the cell or the nanoscale level. Estimates of temperature and absorption coefficients of individual NPs after a laser pulse are reported in Table 3 using eqs 6 and 7. The temperature change of an iron oxide NP after a single pulse is estimated as 200 °C with an absorption coefficient of 1.6 × 105 m-1. Similarly, the temperature rise estimate for a 33 nm gold NP is 900 °C giving rise to an absorption coefficient of 5.3 × 105 m-1. The ratio of the absorption coefficients of individual 33 nm gold to 10 nm iron oxide NPs at 532 nm is hence 3.3. Bulk UV-vis spectrophotometer results for the ratio of absorption coefficients of gold to iron oxide NPs in bulk solutions is 5 (Supporting Information, Figure 2).
4. Discussion 4.1. Loading Studies. The results showed that (i) loading of surfactant-coated NP saturates with time and concentration (Figure 4, parts a and b) and (ii) loading of dextran-coated NP increases linearly with concentration (Figure 4b, inset). 4.1.1. Mechanisms of Loading. The loading kinetics and the coating dependence suggest that surfactant-coated NPs associate by “adsorptive endocytosis” (analogous to a receptor-mediated endocytosis), whereas dextran-coated NPs load by a nonspecific fluid-phase endocytosis pathway. Saturation with time (24 h) and concentration was observed for surfactant-coated NP loading on tumor cells (Figure 4, parts a and b). The times for saturation for the surfactant NP are slow compared to typical values observed for various receptor-ligand kinetics.31 For example, time to achieve “95% of equilibrium receptor binding” is 2.5 min for fibronectin on fibroblasts, 7.5 min for insulin on fat-cells, and 15 min for transferrin binding on HepG2 cells.31 It is likely that saturation times observed in this study can be improved to these (31) Lauffenburger, D. A.; Linderman, J. J. Receptors: Models for binding, Trafficking and Signaling; Oxford University Press Inc.: New York, 1993.
12334 Langmuir, Vol. 23, No. 24, 2007
Kalambur et al.
Figure 4. Cell loading dependence on time and coating: (a) loading kinetics for surfactant-coated NPs on LNCaP-Pro5 cells at 0.1 mg Fe/mL; (b) coating dependence of loading on LNCaP-Pro5 cells for two different NPs (dextran- and surfactant-coated). The fit for surfactant NP and dextran NP in (b) are based on a simple receptorligand and linear model, respectively. The fit parameters for the surfactant NP are: Amax(particle1,magnetophoresis) ) 19.2 ( 7.4, K(particle1,magnetophoresis) ) 0.5 ( 0.4, R2 ) 0.93. The inset shows the linear fit for dextran NP with R2 ) 0.9. Table 2. Radio Frequency Heating of Iron Oxide Nanoparticles in Watera iron oxide nanoparticles (3.7 mg Fe/mL)
temp change (°C) after 15 min of rf heating
surfactant NP dextran NP
33.25 ( 3.50 1.94 ( 0.24
a Temperature change of 3.7 mg Fe/mL of dextran- and surfactantcoated NP in water after 15 min of rf heating. Field parameters were 175 kHz, 14 kA/m, 15 min. Results are presented as mean ( standard deviation.
orders (approximately minutes) by attaching specific receptors to these NPs.32 At saturation times up to 48 h, negatively charged33 (32) Lewin, M.; Carlesso, N.; Tung, C. H.; Tang, X. W.; Cory, D.; Scadden, D. T.; Weissleder, R. Nat. Biotechnol. 2000, 18, 410-414.
Figure 5. Heat treatment of cells after NP loading. (a) Cell viability (LNCaP-Pro5, 48 h) as a function of NP loading (surfactant-coated NP) and different heat treatments. The loading at 48 h was obtained using magnetophoresis. (b) Selective heat treatment of cells after laser irradiation. In sample 2, LNCaP-Pro5 cells loaded with surfactant NPs (0.5 mg Fe/mL, 48 h) are destroyed with unloaded SMCs remaining intact. No significant viability changes seen in sample 1 containing unloaded LNCaP-Pro5 and SMCs. The rf parameters in (a) are 260 A, 357 kHz, 30 min. The laser irradiation parameters for (a) and (b) are 532 nm, 20 pulse/s, 30 s, 0.03 J/pulse.
surfactant-coated NPs exhibit saturable and significantly higher loading with concentration than neutral34 dextran NPs (Figure 4b) suggesting that anionic (negative) NPs load more than neutral NPs to tumor cells. Similar charge dependence (anionic > neutral) has also been verified by other studies with HeLa cells using liposomes35 and other iron oxide NPs.16,17 These studies on (33) Rosensweig, R. E. Ferrohydrodynamics; Cambridge University Press: New York, 1985. (34) Moore, A.; Weissleder, R.; Bogdanov, A., Jr. J. Magn. Reson. Imaging 1997, 7, 1140-1145. (35) Miller, C. R.; Bondurant, B.; McLean, S. D.; McGovern, K. A.; O’Brien, D. F. Biochemistry 1998, 37, 12875-12883.
Cell Loading of Iron Oxide Nanoparticles
Langmuir, Vol. 23, No. 24, 2007 12335
Table 3. Temperature Changes of NPs after a Single Pulsea ∆THeff2O+NP (°C) NP surfactant Fe3O4 10 nm
gold NP 33 nm
∆TH2O (°C)
∆TNP (°C)
no. of pulses
multiple pulses
single pulse
multiple pulses
single pulse
single pulse
volume fraction (%)
200
6
0.03
3
0.015
184
0.01
600 900
16 25
0.027 0.028
6 11
0.01 0.012
204 191
600
26
0.043
6
0.01
899
0.005
a Experimental measurements of mean temperature changes of NP solutions after 45 (900 pulses), 30 (600 pulses), and 10 s (200 pulses) laser irradiation were measured using a thermocouple. NP temperatures are reported based on eq 6.
liposomes further showed that cationic (positive) NPs were loaded more effectively to cells than anionic (negative) NPs. On the basis of the saturable response of surfactant-coated NPs, a simple receptor-ligand equation given by
Aeq )
AmaxS K+S
(9)
can be used to model the process of its loading. Here, S is the extracellular NP concentration and Amax and K represent a maximum loading and an equilibrium dissociation constant, respectively. The model suggests that surfactant-coated NPs (similar to ligands) “adsorb” in a coating- or charge-dependent manner to specific “active” sites (which behave in a manner similar to receptors) on the cell membrane before internalization.16 Higher numbers of such sites lead to higher loading as shown by observations on fibroblasts (see the Supporting Information, Figure 4), larger cells with more “active sites (receptors)”. It should be noted that the ratio of surfactant-coated NPs bound to the membrane to that internalized is not known in this study. However, similar studies on HeLa cells with other anionic iron oxide NPs showed blockage of uptake at 4 °C (binding only) and saturation at 37 °C (binding and internalization) and concluded that the mechanism was “adsorptive endocytosis”,17 which supports observations made in this study. These and other studies have also shown that the NPs were sequestered in endosomes within cells after loading.9,16 The nonsaturable response and linear relationship between loading and concentration of dextran-coated NPs suggest that these NPs are internalized by fluid-phase endocytosis. Studies on other dextran-coated NPs have confirmed that these particles are taken up fluid-phase endocytosis.17,34 4.2. Heating Studies. The results of heating on NP-loaded cells showed that (i) pulsed laser illumination at 532 nm is more effective than rf at 260A, 357 kHz in destroying cells at lower loading concentrations and times (Figure 5a) and (ii) laser irradiation can destroy malignant tumor cells loaded with NPs in a population while leaving unloaded normal cells intact (Figure 5b). 4.2.1. Radio Frequency Versus Laser Heating. Conventional methods of heating iron oxide NPs rely on the use of rf fields.1,36 The mechanism of rf heating involves the combined relaxation of the magnetic moment (Ne´el) and thermal motion (Brownian) of NPs in an alternating magnetic field.25 The method is used for the hyperthermic treatment of tumors in vivo (magnetic fluid hyperthermia).8,18 This method requires local or interstitial delivery of NPs37 at doses of g 1 mg Fe/(g tumor) with the NPs distributed uniformly over regions of one or more millimeter in characteristic dimension (Figure 6).19 Radio frequency heating (36) Jordan, A.; Wust, P.; Scholz, R.; Tesche, B.; Fahling, H.; Mitrovics, T.; Vogl, T.; Cervos-Navarro, J.; Felix, R. Int. J. Hyperthermia 1996, 12, 705-722. (37) Moroz, P.; Jones, S. K.; Gray, B. N. Int. J. Hyperthermia 2002, 18, 267284.
Figure 6. Scaling analysis of rf heating: rf heating at various NP volume fractions and control volumes (2R ) 10 µm vs 1 mm, see eq 4). Calculations use a theoretical heating absorption of 109 W/m3 for Fe3O4.
cannot be used to effectively heat (>10 K) a single cell (Figure 6). The decreases in cell viability observed with rf heating (Figure 5a) are probably a combined effect of both extracellular and intracellular NPs within the cell pellet. To improve in vivo applications of iron oxide NPs for cellular level imaging and therapies, intravenous injections will be necessary.8 As systemic delivery will likely result in lower NP concentrations at the site of interest in vivo, new approaches for heating cells with smaller NP concentrations are needed.37 Scaling (Figure 6) and theoretical calculations20 show that high-intensity pulsed laser methods can lead to single-cell (loaded with iron oxide NPs) treatments unlike rf heating methods which can be used only for bulk tissue leVel treatments. Our data show for the first time that targeted destruction can be achieved after iron oxide NP loading with laser exposure and that with laser treatment significantly lower NP concentrations is needed compared to rf heating (Figure 5a). The mechanism of laser heat treatment involves the absorption of light by NPs and conversion of this energy to heat.38 Temperature change of a 10 nm iron oxide NP after a single laser pulse is estimated to be 200 °C (Table 3). For comparison, temperature change of a single gold NP is 900 °C. Scaling calculations elsewhere with similar laser parameters on 40 nm gold NPs estimate temperature change of 1200 °C.29 The heat diffusion lengths during a single pulse (7 ns) are 50 nm (x4Rt where R is 10-7 m2/s). Since membranes are typically 5 nm in (38) Welch, A. J.; Van Gemert, M. J. C. Optical-Thermal Response of LaserIrradiated Tissue; Plenum Press: New York, 1995.
12336 Langmuir, Vol. 23, No. 24, 2007
thickness39 the heating of a 10 nm NP may lead to membrane melting or poration (i.e., disruption). It has also been suggested that rapid temperature increases around gold NPs of 1000 °C after high-energy laser exposure may lead to bubble formation and collapse.29,40 Further studies are warranted to more completely define the mechanism of cell killing by NP laser heating. The advantages of using laser treatments over rf include both shorter times and lower loading concentrations for effective heat killing of cells. The shorter times (30 s) make laser treatment amenable to continuous flow and batch processing applications in vitro, unlike rf, which requires heating to be continuous for 15-30 min. Though the drop in viability at 1 pg Fe/cell was < 10% after laser exposure, complete cell kill
