Abstract
Cellular therapies require methods for noninvasive visualization of transplanted cells. Micron-sized iron oxide particles (MPIOs) generate a strong contrast in magnetic resonance imaging (MRI) and are therefore ideally suited as an intracellular contrast agent to image cells under clinical conditions. However, MPIOs were previously not applicable for clinical use. Here, we present the development and evaluation of silica-based micron-sized iron oxide particles (sMPIOs) with a functionalizable particle surface. Particles with magnetite content of >40% were composed using the sol-gel process. The particle surfaces were covered with COOH groups. Fluorescein, poly-l-lysine (PLL), and streptavidin (SA) were covalently attached. Monodisperse sMPIOs had an average size of 1.18 μm and an iron content of about 1.0 pg Fe/particle. Particle uptake, toxicity, and imaging studies were performed using HuH7 cells and human and rat hepatocytes. sMPIOs enabled rapid cellular labeling within 4 h of incubation; PLL-modified particles had the highest uptake. In T2*-weighted 3.0 T MRI, the detection threshold in agarose was 1,000 labeled cells, whereas in T1-weighted LAVA sequences, at least 10,000 cells were necessary to induce sufficient contrast. Labeling was stable and had no adverse effects on labeled cells. Silica is a biocompatible material that has been approved for clinical use. sMPIOs could therefore be suitable for future clinical applications in cellular MRI, especially in settings that require strong cellular contrast. Moreover, the particle surface provides the opportunity to create multifunctional particles for targeted delivery and diagnostics.
Keywords
Introduction
Cell-based therapeutic strategies are a major focus for regenerative medicine (38). The transplantation of adult cells, stem cells, and induced pluripotent stem cells is promising for the treatment of acute and degenerative diseases. Clinical trials of cell-based therapeutic strategies show promising results for the treatment of heart failure (1), neurological disorders such as ischemic stroke (5) or multiple sclerosis (29), and metabolic disorders such as diabetes mellitus (31) and metabolic liver disease (11,35). Cell transplantation offers the chance to support or modulate regeneration of the patient's own diseased organs (22). Cell transplantation could improve the cost effectiveness and feasibility of treating large numbers of patients (19).
Challenging the use of regenerative medicine, however, is the fact that conventional medical imaging techniques such as ultrasound, X-ray, or computer tomography are not suitable for visualizing transplanted cells. This limitation of imaging technology precludes the noninvasive localization of transplanted cells following administration (36). Moreover, taking serial samples for histological or molecular analyses is not possible in humans. Factors that could have negative effects on the outcome of cellular therapies, for example, poor engraftment or rapid degradation of transplanted cells, are therefore difficult to determine. As an example, liver cell transplantation (LCT) has been successfully evaluated in small animal models, and in rodents transplanted cells can correct the metabolic deficiencies of the liver or reverse hepatic failure (12). However, clinical studies have so far been unable to reproduce these results, precluding the substitution of LCT for organ replacement (35). These clinical results may in part be caused by complications of cellular therapies, such as the microembolization or systemic delivery of transplanted cells, that cannot be detected in time without methods for noninvasively tracking transplanted cells (4,24). Moreover, engraftment and proliferation of transplanted liver cells is challenging and clinically applicable strategies for selective stimulation of transplanted cells are not available.
Labeling cells with superparamagnetic iron oxide particles as an intracellular contrast agent and using magnetic resonance imaging (MRI) to detect the labeled cells is considered a clinically applicable strategy for the noninvasive visualization of transplanted cells (36). Nanometersized iron oxide particles (SPIOs) are clinically approved and have already been used to image dendritic cells, islet cells, and stem cells in a clinical setting (6). However, to visualize transplanted cells where enhanced contrast for sufficient detectability is needed, such as in visualization by abdominal MRI, larger particles with higher relaxivity must be used (27,34). Commercially available polymer-encapsulated micron-sized iron oxide particles (MPIOs) have been evaluated for MRI monitoring of stem cell and adult cell transplantation in large animal models (23,28); however, these particles are not clinical grade and thereby not clinically applicable. Moreover, the currently commercially available MPIOs are not suited for theranostic applications such as drug delivery or particle-based stimulation of transplanted cells (13).
Among other biocompatible materials, silica is widely used to coat nanometer-sized iron oxide particles (33). Silica is an inert coating material, preventing the aggregation of the superparamagnetic particle core. Moreover, silica enables surface modifications that provide the opportunity for combined diagnostics and therapy (7,18). We developed silica-based micron-sized iron oxide particles (sMPIOs) and used HuH7 cells and adult liver cells to investigate the detectability of sMPIO-labeled cells with a clinical 3.0 T MR scanner. We analyzed the cellular particle internalization and evaluated the possible adverse effects of sMPIOs in vitro. Our goal was to synthesize functionalizable micron-sized superparamagnetic particles that are applicable as an intracellular contrast agent for MRI in a clinical setting and suitable for theranostic modifications.
Materials and Methods
Silica-Based Micron-Sized Iron Oxide Particles
Superparamagnetic sMPIOs were custom-made and kindly provided by microparticles GmbH (Berlin, Germany). sMPIOs with a total iron oxide content of >40% were synthesized using the sol-gel process. The particle core consisted of a silica-based polymer matrix with homogenous distribution of nanometer-scaled iron oxide particles (2). The particle surface was covered with a silica-based polymer coating containing carboxylic acid (COOH) groups (>50 μmol/g). The silica coating provided hydrophilic surface properties and colloid stability and prevented iron leakage. Additionally, fluorescein was attached to the particle surface. For further modification of the particle surface, sMPIOs in 300 μl of 0.1 N 2-(N-morpholino)ethanesulfonic acid (MES) buffer (pH 5.0) were incubated for 3 h with 1 mg N′-(3-dimethylaminopropyl)-N-ethylcarbodiimid HCl (EDC) and 1 mg of either streptavidin (SA) or poly-l-lysine (PLL). Thereafter, particles were washed with deio nized water and phosphate-buffered saline (PBS; pH 7.4).
sMPIOs with COOH groups (COOH-sMPIOs) had a negatively charged particle surface, PLL-modified particles (PLL-sMPIOs) had a positively charged particle surface, and SA-modified particles (SA-sMPIOs) had a nearly neutral surface charge under physiological pH conditions. These three particle types were used for further investigation.
Particle Characterization
The particle surface properties were investigated using scanning electron microscopy (Carl Zeiss Nano Technology Systems, Oberkochen, Germany). Impedance measurements were performed using a Coulter Multisizer II (Beckman Coulter, Inc., Fullerton, CA, USA) to verify the particle size. To estimate the iron content of sMPIOs, particle dilutions were evaluated using continuum source atomic absorption spectrometry (CSAAS). Measurements were performed as described previously using a prototype CSAAS at the Institute for Analytical Sciences (ISAS, Berlin, Germany) (26). For investigation of linear dynamic range, samples with 3, 15, and 30 × 106 particles/ml were taken from the particle stock solution, which contained 7.1 × 109 particles/ml.
Cell Isolation and Culture
HuH7 Cells
HuH7 cells (derived from a male hepatocellular carcinoma) were obtained from the JCRB Cell Bank (Shinjuku, Japan, JCRB0403) and grown in standard culture flasks in Dulbecco's modified Eagle's medium (DMEM; Biochrom, Berlin, Germany) supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate, 4 mM l-alanyl-l-glutamine, and 10% fetal calf serum (FCS) at 37°C (all from Biochrom, Berlin, Germany).
Primary Human Hepatocytes
Primary human hepatocytes were isolated as previously described (27) from the liver tissue of n = 4 patients (age: 59±16 years; gender: three females, one male) undergoing partial hepatectomy following ethical approval (EA2/137/09) and after obtaining informed consent. Cells had a mean viability of 86.0±7.2% after isolation as determined by the trypan blue exclusion test (Sigma Aldrich, St. Louis, MO, USA). Hepatocytes were seeded on collagen-coated 6-well and 12-well culture plates (Sarstedt, Nurnbrecht, Germany) and cultivated for up to 7 days in Williams' medium E with Glutamax (Invitrogen, Karlsruhe, Germany), supplemented with 4 mM l-alanyl-l-glutamine, 1 μM insulin (Biochrom), 1 μM dexamethasone/fortecortin (Merck, Darmstadt, Germany), 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate, 15 mM HEPES (Biochrom), and 10% FCS. The medium was changed every 24 h, and supernatant samples were collected and stored (–80°C) for further analysis.
Primary Rat Hepatocytes
Primary rat hepatocytes were isolated from n = 5 male Lewis rats (Harlan-Winkelmann, Borchen, Germany) with a modification of the two-step perfusion protocol described by Seglen et al. (30). All experiments were performed in accordance with federal law and were approved by the local authorities for animal research (O 0459/09). Briefly, livers were flushed via the portal vein with Leffert's buffer [consisting of HEPES, KCl (Merck), NaCl (Roth, Karlsruhe, Germany), NaH2PO4 (Serva Electrophoresis, Heidelberg, Germany), and glucose (Merck)] containing EGTA (Fluka Chemie, Neu-Ulm, Germany) prior to digestion with collagenase (Biochrom) in CaCl buffer (Merck). Hepatocytes were purified using density gradient centrifugation (50% percoll; Biochrom). Cells with a mean viability of 68.5±8.5% were obtained from percoll purification. Hepatocytes were cultured for up to 7 days in DMEM supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, 4 mM l-alanyl-l-glutamine, 2.5 ng epidermal growth factor (EGF; Sigma Aldrich), 7 μl glucagon (Biochrom), 1 μM insulin, 1 μM dexamethasone/fortecortin, and 10% FCS at 37°C on standard 12-well plates. Supernatant samples were taken as described above.
Cell Labeling with sMPIOs
HuH7 cells and hepatocytes were labeled in adhesion culture. Eighteen hours prior to labeling, 1 × 106 cells were seeded in six-well culture plates or proportionally fewer cells were seeded in smaller plates. Cells were washed with PBS, replaced with culture medium containing the final particle concentration, and incubated with sMPIOs for 4 h at 37°C. Control groups were replaced with medium without particles. Thereafter, cells were gently washed three times with PBS to remove free and loosely bound particles and were replaced with fresh medium.
Experimental Design and Groups
For the initial evaluation of the particle uptake and definition of the MRI threshold, HuH7 cells were incubated with COOH-sMPIOs, PLL-sMPIOs, and SA-sMPIOs at a concentration of 100 particles/cell. Toxicity studies were performed with HuH7 cells labeled with 40, 60, 80, and 100 COOH-sMPIOs/cell. Cells incubated with iron citrate (Merck; 30 μg/ml) and Triton X (Sigma Aldrich; 1.7 mol/ml) were used as control groups. To investigate the particle uptake mechanisms, HuH7 cells were additionally incubated with 100 particles/cell under one of two conditions: hypothermia (4°C) or following a 30-min pretreatment with sodium acid phosphate (Merck; 3 mM) (39). Primary human hepatocytes were incubated with increasing particle concentrations (10–100 particles/cell) to identify the optimal concentration for labeling. To evaluate possible adverse effects of sMPIO labeling on liver cells, primary rat hepatocytes were incubated with increasing concentrations of PLL-sMPIOs (20–60 particles/cell), COOH-sMPIOs (40–80 particles/cell), and SA-sMPIOs (60–100 particles/cell). Primary human hepatocytes were incubated with COOH-sMPIOs at a final concentration of 60 particles/cell. Control groups consisted of cells cultured under standard conditions and cells cultivated in medium with a ferric iron concentration of 30 μg/ml to induce iron overload (27).
Investigation of the Particle Uptake and Toxicity Studies
Quantification of the Particle Uptake
The particle uptake was counted on micrographs. Pictures were taken by phase contrast and transmitted light microscopy at 200-fold magnification (Axiovert 40CFL; Zeiss) from culture days 1 to 7. Particles were counted in randomly chosen fields. The particle uptake and labeling efficiency (number of labeled cells×100/total number of cells) were calculated.
Verification of the Intracellular Localization of sMPIOs
The intracellular localization of sMPIOs in labeled cells was investigated by fluorescence microscopy. Human hepatocytes were fixed using 4% paraformaldehyde (Serva Electrophoresis) and permeabilized using ice-cold methanol (Roth). The cells were stained for cytoplasmic proteins and for nuclei by sequential 1-h incubations with a primary antibody against human cytokeratin 18 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), a secondary antibody indocarbocyanine (JacksonImmunoResearch, Newmarket, Suffolk, UK), and 4′,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, Steinheim, Germany). The analyses were performed using a filter set and a fluorescence lamp.
The intracellular localization of sMPIOs was further investigated using transmission electron microscopy (TEM). HuH7 cells cultivated on glass cover slips (Chance proper, Smethwick, Warley, UK) were used for TEM studies. Cells were fixed in 2.5% glutaraldehyde (Serva Electrophoresis), postfixed using 2% osmium tetroxide (ChemPur, Karlsruhe, Germany), dehydrated in ascending alcohol concentrations, and then infiltrated overnight in an Epon-containing solution (Serva Electrophoresis). The cover slips were then placed in plastic capsules (Plano, Wetzlar, Germany) for polymerization (48 h at 60°C). Then the cover slips were removed, and ultrathin sections were transferred onto 200-mesh copper grids (Plano). Samples were stained with 4% aqueous uranyl acetate (Merck) and Reynolds lead citrate [consisting of methylene blue (Merck), azure II, and borax (Sigma Aldrich)] and imaged using an electron microscope (EM 906; Zeiss).
Toxicity Studies
The cell titer glow luminescent assay (Promega Corporation, Madison, WI, USA) was used for toxicity studies with sMPIOs. The assay is based on the conversion of luciferin by a recombinant luciferase in the presence of ATP. Measurements were performed 18 h after sMPIO labeling or following 18 h of iron citrate or Triton X exposure using a Fuostar Optima plate reader (BMG Labtech, Ortenberg, Germany).
Magnetic Resonance Imaging and T2 Maps
An agarose block (1% agarose supplemented with 0.5 mM/ml gadolinium; Dotarem; Guerbet, France) with five rows of 12 circular pockets (depth: 10 mm; diameter: 5 mm) was prepared for in vitro MRI using a custom-made molding tool. Labeled and native cells were resuspended from the culture plates with trypsin/EDTA (Biochrom). A series of samples with increasing numbers of cells (500– 100,000 cells) diluted in 100 μl PBS was placed into the pockets of the agarose block. Thereafter, the pockets were completely filled with PBS.
MRI was performed with a clinical 3.0 T MR scanner (Signa 3T94, GE Healthcare, Milwaukee, WI, USA) and a four-channel phased array torso coil using a T2-weighted fast spin echo sequence (FSE), a T2*-weighted sequence, and a 3D T1-weighted LAVA-sequence (Liver Acquisition with Volume Acceleration) (Table 1). Details of the LAVA sequence are the following: repetition time (TR) 3.75 ms, echo time (TE) 1.756 ms, inversion time (IR) 5 ms, flip angle (FA) 12°, field of view (FOV) 296 mm, acquired slice thickness (ST) 3 mm, reconstructed ST 1.5 mm, 72.9% sampling, 90% phase FOV, reconstruction matrix 512 × 512.
MRI Sequences and Parameters
Magnetic resonance imaging (MRI) was performed with a 3.0 T MR scanner using a four-channel phased array torso coil. PHH, primary human hepatocytes; FSE, fast spin echo; TR, repetition time; TE, echo time; LAVA, liver acquisition with volume acceleration.
To quantify the T2 relaxation time, resp. relaxation rate R2 = 1/T2, 2D fast spin echo image acquisitions were performed using repeated measurements holding all acquisition parameters constant except for TE, which varied: TR 3,260 ms, TE 14.7/44.2/83.6/122.9/157.3 ms, FA 90°, echo train length (ETL) 16, FOV 200 mm, ST 1 mm, 100% sampling, 60% phase FOV, reconstruction matrix 512 × 512, number of acquisitions (NEX) 6. T2 maps were calculated voxel-wise using a linear regression of the logarithmized signal intensities. The calculation module was implemented in the visualization software Amira, version 5.22 (Visage Imaging, Berlin, Germany). The average T2 relaxation time was calculated using a volume of interest (VOI) covering the total pocket taking partial volume effects into account.
Establishment of the MRI Detection Limit
The establishment of the MRI detection limit was primarily based on radiological evaluation of images from sMPIO-labeled HuH7 cells. Image analysis was performed in consensus with clinical and experimental radiologists. T2 map measurements were used to confirm the morphologically observed differences. A R2 relaxation rate of 11/s or greater correlated to the morphological defined detection limit for MRI.
Evaluation of Cell Damage and Synthesis Parameters of sMPIO-Labeled Cells
Biochemical parameters were analyzed from the supernatants of cultivated primary human and rat hepatocytes daily from culture days 2–6 or 7. Aspartate aminotransferase (AST) and alanine aminotransferase (ALT) enzyme activities were measured (NobiFlow GOT-IFCC, GPT-IFCC). Urea formation was detected with an enzyme-based detection kit (NobiFlow Harnstoff-UV, all HITADO, Mohnesee, Germany).
Statistical Analysis
Data are expressed as mean±standard deviation (SD). Statistical analysis (one-way ANOVA with subsequent post hoc multiple comparisons, Wilcoxon test) was performed using Microsoft Excel for Mac 2004 (V11.3.6, Microsoft Corporation, Redmond, WA, USA) and SPSS version 13 for Windows (SPSS Inc., Chicago, IL, USA). A value of p ≤ 0.05 was considered statistically significant. Each experiment was performed at least in triplicate and measurements were repeated at least twice.
Results
sMPIOs Are Monodisperse and Show Narrow Size Distribution
Scanning electron microscopy showed highly mono-dispersed particles of approximately uniform size and spherical shape (Fig. 1A, B). Multisizer measurements revealed a mean size of 1.18 μm and coefficient of variation values below 5% for COOH-sMPIOs, PLL-sMPIOs, and SA-sMPIOs (data not shown). CSAAS measurements showed a linear increase of the iron content of the particle dilutions (R2 = 0.98) (Fig. 1C). The mean iron content of COOH-sMPIOs was 1.13±0.02 pg Fe per particle. The mean iron content of PLL-sMPIOs and SA-sMPIOs was 0.86±0.08 pg Fe and 0.82±0.02 pg Fe per particle, respectively.
sMPIO surface properties, iron content, and toxicity studies. Scanning electron microscopy of carboxylic acid-silica-based micron-sized iron oxide particles (COOH-sMPIOs) revealed monodisperse particles with a spherical shape (A, B). Iron concentration was measured using continuum source atomic absorption spectrometry (CSAAS). The iron content of dilutions of the particles with all three surface modifications increased linearly with the particle concentration (C). Toxicity studies showed no significant differences of the ATP content of HuH7 cells incubated with increasing concentrations of COOH-sMPIOs compared to native cells. Cells exposed to Triton X had no detectable ATP levels (*p < 0.05 compared to native cells; n = 3, one-way ANOVA with subsequent post hoc multiple comparisons) (D). Scale bars: 2 μm (A) and 200 nm (B). PLL, poly-l-lysine; SA, streptavidin.
sMPIOs Have No Dose-Dependent Toxicity to Labeled Cells
Toxicity studies with HuH7 cells showed no dose-dependent toxicity of sMPIOs (Fig. 1D). Native cells had a relative luminescence unit (RLU) of 149±17, while the RLU of cells incubated with 40 or 100 COOH-sMPIOs/cell was 118±21 and 127±6 (n.s. compared to native cells). Iron citrate lead to a slightly stronger decrease of the ATP content (RLU: 115±23), while Triton X caused a complete depletion of the ATP content (p < 0.05 compared to native cells).
Cellular Labeling with sMPIOs Is Feasible and Energy Independent
HuH7 cells were incubated with 100 particles/cell to investigate the feasibility of cellular labeling with sMPIOs. Following the 4-h incubation period, particles were microscopically detectable inside the cytoplasm of the incubated cells (Fig. 2A–C). PLL-sMPIOs tended to cluster inside the cells and on the culture plate whereas COOH-sMPIOs and SA-sMPIOs did not form clusters. Transmission electron microscopy (Fig. 2D–F) showed the intracellular localization of sMPIOs in adherent HuH7 cells. Particles were loosely distributed in the cytoplasm and were not enclosed by cellular membranes. No differences were observed in the intracellular localization among particles with the three different surface modifications. Fluorescence microscopy demonstrated the intracellular localization of sMPIOs in labeled primary human hepatocytes (Fig. 2G).
Particle uptake studies. HuH7 cells were incubated with 100 particles/cell to investigate the feasibility of cellular labeling with sMPIO (COOH-sMPIOs: A, D; PLL-sMPIOs: B, E; SA-sMPIOs: C, F). Under light microscopy, particles were clearly visible in the cytoplasm of labeled cells (A–C). Electron microscopy demonstrated the intracellular localization of sMPIOs in adherent labeled cells (D–F). Fluorescence microscopy was performed with primary human hepatocytes (G) (red: CK18; blue: nuclei; green: COOH-sMPIOs with covalently attached fluorescein groups). The particle uptake was not influenced by labeling under hypothermia or following blockade of the cellular respiratory chain using sodium acid phosphate (n = 3, Wilcoxon test, n.s.) (H). Scale bars: 10 μm (A–C), 2.5 μm (D–F), and 5 μm (G).
The quantification of the particle uptake revealed differences among the three particle types. Incubation with COOH-sMPIOs resulted in an uptake of 15±4 particles/cell and a labeling efficiency of 94±4%. Incubation with the same amount of PLL-sMPIOs led to a significantly higher particle uptake (23±6 particles/cell, p = 0.006) and a significantly higher labeling efficiency (98±3%, p = 0.03). The SA-sMPIO uptake (5±1 particles/cell) and labeling efficiency (72±11%) were significantly lower compared with COOH-sMPIOs and PLL-sMPIOs (p < 0.005).
To investigate the particle uptake mechanism, HuH7 cells were incubated with sMPIOs under one of two conditions: hypothermia (4°C) or blockage of the cellular respiratory chain with sodium acid. The particle uptake and labeling efficiency under these two conditions were not significantly different from the uptake and labeling under normal conditions (Fig. 2H), indicating that particle internalization was mediated by energy-independent mechanisms.
sMPIO-Labeled Cells Induce Signal Voids in T1- and T2-Weighted Sequences at 3.0 T MRI
In vitro MRI was performed to investigate the detectability of sMPIO-labeled cells and to define a detectability threshold for 3.0 T MRI. Pure particles without cells induced homogeneous signal extinctions. No differences were observed among the three particle types. HuH7 cells incubated with 100 particles/cell induced signal voids in T1- and T2*-weighted MRI, with the strongest signal disruption in the T2*-weighted sequence whereas native, nonlabeled cells did not induce detectable signal changes. Signal voids increased in correlation to the number of incorporated particles/cell and the number of labeled cells in the agarose pockets (data not shown). T2 maps generated from 2D fast spin echo images confirmed the morphologically observed differences of the signal voids. PLL-sMPIO-labeled cells, which contained the highest number of particles, induced the largest voxel volumes at each investigated cell dilution. The R2 relaxation rate increased nonlinearly in correlation to the increasing numbers of cells per agarose pocket (Table 2). A R2 relaxation rate of 11/s or greater was induced by 5,000 PLL-sMPIO-labeled cells, 10,000 COOH-labeled cells, and 25,000 SA-sMPIO-labeled cells, which correlates with the morphologically defined signal threshold.
R2 Relaxation Rate of sMPIO-Labeled HuH7 Cells
R2 relaxation rates were calculated voxel-wise from 2D fast spin echo images to quantify the T2 relaxation time of silica-based micron-sized iron oxide particle (sMPIO)-labeled cells. R2 relaxation rates increased in correlation to the increasing number of labeled cells per pocket in the agarose phantom. Poly-l-lysine (PLL)-sMPIOs, which showed the strongest cellular uptake, displayed the largest change in relaxation rate at each cell dilution. A R2 relaxation rate of at least 11/s was achieved at the morphologically defined detection threshold (*). COOH, carboxy lic acid; SA, streptavidin.
Incubation of Human Liver Cells with 60 sMPIOs/Cell Enables the Detection of Cell Cluster with MRI
Human hepatocytes were incubated with increasing numbers of the three particle types (10–100 sMPIOs/cell) to investigate the dependence of particle uptake on incubation concentration. The results from the initial study with HuH7 cells demonstrated that an uptake of at least 20 particles/cell was the threshold to guarantee detectability of the T2* effect in a cell cluster with a T1-weighted MRI. As in HuH7 cells, sMPIO uptake depended on the particle surface. The particle uptake increased in linear correlation to the incubation concentration (Table 3). MRI was performed with cells incubated with 60 particles/cell. In measurements using the T1-weighted LAVA sequence, hepatocytes labeled with PLL-sMPIOs (particle load: 41±8 particles/cell) and SA-sMPIOs (19±1) were detectable from a minimum of 7,500 cells whereas COOH-sMPIOs containing 24±12 particles/cell were detectable from a minimum of 10,000 cells (Fig. 3A–D). In T2*-weighted sequences, cells labeled with particles with all three surface modifications were detectable down to a minimum of 1,000 cells (Fig. 3E–H).
Labeling and imaging of human liver cells using sMPIOs. sMPIO-labeled human hepatocytes were investigated using T1-weighted 3D liver acquisition with volume acceleration (LAVA) (A–D) and T2*-weighted sequences (E–H). Decreasing numbers of labeled cells were investigated using the agarose phantom (T1: 10,000–2,500 cells; T2*: 7,500–1,000 cells). PLL-sMPIO- and SA-sMPIO-labeled cells were detectable from 7,500 cells, whereas COOH-sMPIO-labeled cells were detectable from 10,000 cells. Human hepatocytes labeled with COOH-sMPIOs were cultivated for 6 days in adhesion culture to evaluate morphological changes, transaminase leakage, and urea synthesis (I: day 1; J: day 2; K: day 4; L: day 6). Pictures were taken using phase contrast and permitted light microscopy at 200-fold magnification. Particles are clearly visible inside the cytoplasm of the cells. Morphological changes of sMPIO-labeled cells compared to native cells were not observed throughout the culture period. Scale bar: 10 μm.
Particle Uptake of Primary Human Hepatocytes
Primary human hepatocytes were incubated with increasing particle concentrations to investigate the incubation concentration dependence of the particle uptake. Particle load increased linearly with the incubation concentration. PLL-sMPIOs showed the greatest particle uptake, whereas SA-sMPIOs showed the lowest particle uptake.
Labeling with sMPIOs Is Stable and Has No Adverse Effects on Liver Cells
Primary rat and human hepatocytes were used to investigate whether sMPIOs have adverse effects on labeled cells. Primary rat hepatocytes were incubated with increasing concentrations of the three particle types. As expected, the particle uptake increased with increasing incubation concentration. The uptake of PLL-sMPIOs was increased compared with the other groups. The particle load tended to increase over the culture period. Transaminase leakage was low and not significantly different between native and labeled cells at the same culture days. Urea secretion decreased over time but showed no differences between native and labeled cells (data not shown). Thus, incubation with increasing concentrations of particles with any of the three surface modifications in ranges needed to obtain sufficient contrast on MRI had no specific adverse effects on transaminase leakage and urea synthesis of rat hepatocytes.
To confirm these results with human hepatocytes, cells incubated with COOH-sMPIOs at a final concentration of 60 particles/cell were used. Labeled human hepatocytes presented a polygonal shape, granular cytoplasm with vesicular inclusions, and one or more nuclei (Fig. 3I–L). Labeling with COOH-sMPIOs caused no alterations in the morphology of the hepatocytes. Over the entire culture period, no morphological differences were observed between sMPIO-labeled hepatocytes and native cells. Labeling was stable throughout the culture period (Fig. 4A). AST and ALT leakage of both sMPIO-labeled and native hepatocytes was slightly increased at culture day 2 (Fig. 4B, C). Thereafter, the levels of sMPIO-la-beled cells were low and not significantly different from native cells at the same culture days. Moreover, urea formation was similar to the native control group (Fig. 4D). In contrast, the AST leakage of iron-oversupplied cells was significantly higher than that of native and sMPIO-labeled cells at culture days 3–5. The ALT leakage of iron-oversupplied cells was significantly higher than that of native and sMPIO-labeled cells at culture day 3.
In vitro characterization of sMPIO-labeled primary human liver cells. Particle retention (A) and biochemical parameters [B: aspartate aminotransferase (AST); C: alanine aminotransferase (ALT); D: urea] of COOH-sMPIO-labeled primary human hepatocytes. Cells were labeled at a concentration of 60 particles/cell. Iron-stimulated cells were cultivated in the presence of iron citrate. Data are expressed as mean±SD. Significant differences between the groups at the same culture day: *p < 0.05 (n = 3, one-way ANOVA with subsequent post hoc multiple comparisons).
Discussion
We developed and evaluated silica-based micron-sized iron oxide particles with a COOH group coating that was suitable for surface modifications. Fluorochromes as well as PLL and SA were attached to demonstrate the functionalizability of sMPIOs. COOH-sMPIOs enabled the rapid labeling of HuH7 cells and adult hepatocytes within 4 h through passive particle uptake. Covalently adding PLL to the sMPIO surface significantly increased the particle uptake while SA-modified particles showed a reduced particle uptake and labeling efficiency. Cellular labeling proved to be stable and had no adverse effects on cultivated human and rat hepatocytes. sMPIOs enabled the detection of 1,000 cells with T2*-weighted 3.0 T MRI and the detection of at least 10,000 labeled cells using the T1-weighted LAVA sequence, which is applicable for imaging of the liver under clinical conditions.
Labeling of cells with superparamagnetic iron oxide particles is a promising strategy for the noninvasive imaging of cellular therapies using magnetic resonance imaging (36). Zhu et al. demonstrated the feasibility of MRI tracking of neural stem cells in patients with brain trauma (41). De Vries et al. detected misinjection after ultrasound-guided needle injection of dendritic cells in lymph nodes of melanoma patients by using MRI monitoring (10). In MRI of the central nervous system, the detection limits of cells labeled with iron oxide particles are as low as a few hundred cells (9). In the abdomen, the detection of transplanted cells is much more challenging: the quality of MR images can be influenced by moving artifacts of the intestines, intraluminal air as well as the patient's breathing, and the patient's general cooperation. Moreover, larger coils are required to image the abdomen, which increases the field of view but decreases image resolution. MRI with 3.0 T or higher can be used to improve abdominal diagnosis by increasing the signal-to-noise ratio, contrast, and resolution of the images (16,25). T2* sequences especially profit under these conditions from the increased magnetic susceptibility induced by iron oxide particles.
Two strategies are possible to increase the detectability of cells with MRI. Labeling with nanometer-sized particles can be enforced using transfection agents such as PLL or protamine sulfate, which allow the incorporation of millions of ferumoxide particles into cells (20). Alternatively, micron-sized iron oxide particles have been used for magnetic cell labeling. Commercially available Bangs® MPIOs contain much more magnetite than nanometer-sized ferumoxides [1–10 pg vs. 10−6–10−7 pg of iron (32)], have significantly higher R2/R2* molar relaxivities, and require only a few particles/cell to induce sufficient detectability (34). We previously investigated the feasibility of tracking transplanted cells in a preclinical swine model using Bangs® MPIO-labeled liver cells and 3.0 T MRI. We achieved a detection threshold of 10,000 cells using the T1-weighted 3D LAVA sequence. This enabled us to detect microthrombi caused by transplanted cells, which is a common unwanted event following LCT that cannot be visualized using conventional imaging techniques (24,28). The T2*-weighted sequence is very sensitive to iron oxide particle-induced magnetic susceptibility effects, but it proved not to be applicable in the clinical setting, as lengthy breath-holding periods were needed to achieve satisfactory image quality. Thus, the T1-weighted LAVA sequence was designated for clinical imaging of the T2* effect of iron oxide particle labeled liver cells in the abdomen with 3.0 T MRI. However, since the Bangs® MPIOs are composed of polystyrene/divinylbenzene, an inert, nondegradable polymer that has not been FDA approved for use in humans, our results with these particles cannot be directly translated to clinical use.
Silica, chosen for the fabrication of sMPIOs, is considered to be biologically inert, biocompatible, and nontoxic (33). Silica has already been used to encapsulate and stabilize nanometer-sized iron oxide particles (8,17,37). Recently, ultrasmall silica-based nanoparticles have been approved by the FDA for targeted molecular imaging of cancer (3,15). The sMPIOs presented in our study have a diameter of 1.18 μm and an iron content of about 1 pg Fe/particle, similar to the commercially available Bangs® MPIOs. sMPIOs are highly monodisperse, which should guarantee comparable signal from all particles and reproducible imaging characteristics of labeled cells. The COOH particle surface enables further surface modifications through covalent attachment of the amine groups of proteins or antibodies, which is the requirement for modification of particles for theranostic applications (13). Tumor cells and adult liver cells were labeled with sMPIOs within 4 h of incubation time. Rapid labeling is important for cells that have to be cultivated in adherence, such as liver cells; in our experience, these cells cannot be efficiently resuspended for transplantation following prolonged adherence periods. sMPIO uptake was increased by modifying the particle surface with positively charged PLL and was decreased by modification with SA. These results agree with reports that positively charged particle surfaces are attracted to negatively charged cell plasma membranes (40). However, PLL-sMPIOs clustered inside and outside of the cells, which could limit the further application of these particles.
Since the particle uptake kinetics were not affected by hypothermia or the blockage of the cellular respiratory chain, we assume that the internalization of sMPIOs is mediated by energy-independent mechanisms. In contrast, Kunzmann et al. reported that nanometer-sized silica particles were incorporated through an active cytoskeleton-dependent process by macrophages and dendritic cells (14). Presumably, in our results, the larger sMPIOs attach to the cell surface and passively penetrate the cellular membrane. In support of this hypothesis, we did not find membranous encapsulation of the internalized sMPIOs within the cytoplasm of the cells.
Intracellular contrast agents need to be retained over a distinct period of time and should not have adverse effects on the labeled cells. Initial toxicity studies with HuH7 cells showed no dose-dependent toxicity of COOH-sMPIOs under incubation concentrations necessary for sufficient cellular labeling. Primary hepatocytes, which are a more accurate model of in vivo conditions, were used for further in vitro studies (3). Moreover, primary hepatocytes were investigated as an example of cells that need to be labeled in adherent culture and resuspended for administration. Labeling of rat hepatocytes with increasing concentrations of the three particle types did not determine specific toxicities of sMPIOs. These results were confirmed in human hepatocytes using COOH-sMPIOs. Labeling also proved to be stable: no release of the particles into the culture medium was observed. Furthermore, particle load increased over time, possibly because of the advancing confluence of hepatocytes under in vitro conditions. Moreover, particles released from apoptotic cells may have been incorporated from adherent cells. Although further studies are necessary, especially addressing long-term stability and toxicity of sMPIOs, our data indicate that sMPIOs are biologically inert and have no short-term adverse effects on labeled cells, which should qualify them for further clinical consideration.
The most important point to address, however, is the imaging capacity of sMPIOs under conditions applicable for clinical MRI. We therefore used a 3.0 T MR scanner, which is routinely available in most large centers and clinics and enables high-quality MRI of the abdomen. No special hardware modifications or gradient inserts were applied to investigate imaging capacities under conditions similar to the clinical setting. Using the T1-weighted LAVA sequence, COOH-sMPIO-labeled cells were detectable from clusters of at least 10,000 labeled cells using the T2* effect, which is comparable to the results obtained with Bangs® MPIOs under similar conditions. PLL- and SA-sMPIOs were even detectable at 7,500 cells, although the particle loads were rather different. A possible explanation for this effect might be that clustering of PLL-modified sMPIOs interfered with the MR field, which affected the detectability levels. Moreover, the accuracy of microscopic particle counting might be affected by particle clustering, which correlates with the higher standard deviations observed with PLL-sMPIOs. The T2*-weighted sequence had an even lower detection threshold of the iron oxide susceptibility effect and enabled the detection of 1,000 labeled cells independent of the particle surface modification. Since the detectability of sMPIO-labeled cells depends on the number of incorporated particles, the detection threshold could theoretically be further decreased down to hundreds of cells by enhancing labeling efficiencies using longer incubation periods or higher incubation concentrations. The R2 relaxation rate of sMPIO-labeled cells increased nonlinearly with the number of cells. Usually, a linear correlation between change in relaxation rate and contrast medium concentration is expected. The nonlinear relationship was presumably caused by agglomeration of the sMPIOs incubated cells. The inhomogeneous intracellular distribution of the sMPIOs cells unfortunately increases the difficulty to quantify the iron loading by MRI.
In conclusion, silica-based micrometer-sized iron oxide particles are nontoxic, enable rapid cellular labeling, and generate sufficient contrast on MRI in vitro. These characteristics make sMPIOs attractive for in vivo use and further clinical application. However, further studies are necessary to investigate possible long-term toxicity or degradation of sMPIOs. For example, Nkansah et al. recently reported on specific short- and long-term degradation properties of poly(lactide-co-glycolide) and cellulose MPIOs (21). Animal studies are also necessary to investigate in vivo imaging of sMPIO-labeled cells. However, as silica is FDA-approved and our in vitro studies with different cell types have demonstrated the harmlessness of sMPIOs in concentrations needed for sufficient contrast for cellular MRI, this new class of micron-sized iron oxide particles is clearly suitable for further clinical applications. Moreover, the ability to make different particle surface modifications provides an opportunity to create multifunctional theranostic agents for diagnostic imaging, drug delivery, or selective particle-based stimulation of transplanted cells.
Footnotes
Acknowledgments
The authors are grateful to Peter Behringer, Suanne Kolano, Steffen Lippert, and Ann Skiba for expert technical assistance, Virginia Ding-Reinelt for MRI operational assistance, and Petra Schrade for performing transmission electron microscopy. This study was funded by the European Regional Development Fund (EFRE 10144342). The authors have no conflicts of interest.