Can Magnetic Targeting of Magnetically Labeled Circulating Cells Optimize Intramyocardial Cell Retention?

Abstract

Therapeutic intracavitary stem cell infusion currently suffers from poor myocardial homing. We examined whether cardiac cell retention could be enhanced by magnetic targeting of endothelial progenitor cells (EPCs) loaded with iron oxide nanoparticles. EPCs were magnetically labeled with citrate-coated iron oxide nanoparticles. Cell proliferation, migration, and CXCR4 chemokine receptor expression were assessed in different labeling conditions and no adverse effects of the magnetic label were observed. The magnetophoretic mobility of labeled EPCs was determined in vitro, with the same magnet as that subsequently used in vivo. Coronary artery occlusion was induced for 30 min in 36 rats (31 survivors), followed by 20 min of reperfusion. The rats were randomized to receive, during brief aortic cross-clamping, direct intraventricular injection of culture medium (n = 7) or magnetically labeled EPCs (n = 24), with (n = 14) or without (n = 10) subcutaneous insertion of a magnet over the chest cavity (n = 14). The hearts were explanted 24 h later and engrafted cells were visualized by magnetic resonance imaging (MRI) of the heart at 1.5 T. Their abundance in the myocardium was also analyzed semiquantitatively by immunofluorescence, and quantitatively by real-time polymerase chain reaction (RT-PCR). Although differences in cell retention between groups failed to be statistically significant using RT-PCR quantification, due to the variability of the animal model, immunostaining showed that the average number of engrafted EPCs was significantly ten times higher with than without magnetic targeting. There was thus a consistent trend favoring the magnet-treated hearts, thereby suggesting magnetic targeting as a potentially new mean of enhancing myocardial homing of intravascularly delivered stem cells. Magnetic targeting has the potential to enhance myocardial retention of intravascularly delivered endothelial progenitor cells.

Introduction

Cell therapy is a promising approach to myocardial regeneration and neovascularization but currently suffers from inefficient homing after intracavitary infusion (21). Here we tested a physical approach whereby in which we loaded cells with iron nanoparticles to retain them in the heart with a subcutaneously implanted magnet.

Biodegradable superparamagnetic iron oxide nanoparticles are efficiently internalized by a wide variety of cells, including stem cells, with few impairment of cell functions (3,43). The “magnetic” cells thus obtained, and their migration to various body tissues, can be visualized by means of magnetic resonance imaging (MRI) (7,37). In addition, these cells are sufficiently magnetic to be remotely manipulated. Magnetic control of cell motion has already been used in basic studies of cell mechanics (40), cell sorting applications (29), and tissue engineering (14,22). In the field of cell therapy, magnetic targeting was first used to enhance liver cell transplantation (4,26) and was recently shown to facilitate endothelialization of stented vascular grafts in blood flow conditions (30–32). Magnetic positioning of endothelial cells at sites of arterial injury has also been achieved in perfused vessels, allowing endothelial lentiviral gene transfer (20). Gene transfer has also been enhanced by placing magnetic nanobeads carriers under an external epicardial magnet (24).

Here we examined whether magnetically assisted cell delivery could enhance myocardial retention of human cord blood-derived endothelial progenitor cells (EPCs) after intracoronary injection in a rat model of myocardial infarction. We first determined the optimal conditions of EPC loading with anionic superparamagnetic nanoparticles, in terms of cell viability and iron uptake. The magnetic properties and magnetophoretic mobility of the loaded cells were then determined in vitro, with the same magnet as that subsequently used in vivo. Finally, the efficiency of cell retention in vivo was assessed by MRI, immunohistochemistry, and quantitative real-time polymerase chain peaction (RT-PCR).

Materials and Methods

Isolation and Culture Conditions of Cord Blood Endothelial Progenitor Cells (CB-EPCs) and EPC-Derived Cells

Ethical approval and informed consent from patients was obtained prior to the study. Mononuclear cells were isolated from human umbilical cord blood by centrifugation, then plated in collagen type I-coated wells (Becton Dickinson) and maintained in endothelial basal medium-2 (EBM-2) supplemented with complete endothelial growth media (EGM2-MV SingleQuots; Lonza) at 37°C with 5% CO₂. Outgrowth of endothelial colony-forming cells (ECFCs), characterized by the formation of a cell cluster with a cobblestone aspect, was monitored daily. To obtain EPC-derived cells, ECFCs were trypsinized, seeded at 10,000 cells/cm² and expanded in EGM2-MV medium. The cells derived from the ECFCs were used after one or two passages and can be considered phenotypically as late outgrowth EPCs. Their in vitro functionality was demonstrated by the formation of capillary-like structures after an 18-h culture in Matrigel, the upregulation of intercellular adhesion molecule (ICAM)-1 and vascular cell adhesion molecule (VCAM)-1 upon stimulation by tumor necrosis factor-α (10 ng/ml), and the uptake of acetylated low-density lipoprotein (1c-LDL). Finally, EPC-derived cells were immunophenotyped with the following monoclonal antibodies: anti-CD31 fluorescein isothiocyanate (FITC), anti-CD144 phycoerythrin (PE; Becton Dickinson), anti-vascular endothelial growth factor receptor 2 allophycocyanin (VEGFR2-APC; Sigma), and anti-von Willebrand factor (vWF) (Dako). Data were acquired and analyzed with a FACStar flow cytometer (Becton Dickinson).

Preparation and Characterization of Magnetically Labeled Cells

Cells were labeled with 8-nm-diameter anionic citrate-coated maghemite nanoparticles (AMNPs) provided by PECSA (CNRS/UPMC, Paris, France) and prepared as previously described (42). Adherent EPCs were incubated at 37°C with AMNPs in RPMI culture medium (Sigma, France) supplemented with 5 mM sodium citrate, at iron concentrations of 0.5, 2, 5, and 10 mM for 30 min.

Transmission electron microscopy was used to locate the nanoparticles within the cells. Iron-loaded EPCs were cultured overnight or for 10 days in EBM-2 culture medium, then fixed and embedded in Epon, using standard methods. Ultrathin sections of 70 nm were examined with a Zeiss EM 902 transmission electron microscope operating at 80 kV (Plateforme Mima2, INRA, Jouy en Josas, France).

For proliferation assays, EPCs incubated with 0, 2, 5, or 10 mM iron for 30 min were seeded in 24-well plates at a density of 2 × 10⁴ cells/well, then harvested and counted in triplicate 1, 2, 3, 4, and 5 days later.

Cell migration was measured with modified Boyden chambers (6.5-mm Costar Transwells, Corning, NY) consisting of two medium-filled compartments separated by a microporous membrane (5 μm pore size). EPCs (2 × 10⁴ cells) incubated with 0, 2, 5, or 10 mM iron for 30 min were placed in the upper compartment and allowed to migrate through the membrane into the lower compartment containing EBM-2 and human recombinant VEGF (40 ng/ml). After 24 h of incubation at 37°C, the lower side of the filter was washed and fixed with 2% paraformaldehyde. Cells were stained with Giemsa and counted in triplicate.

C-X-C chemokine receptor type 4 (CXCR4) expression was determined by fluorescence-activated cell sorting (FACS) analysis. EPCs incubated with 0 or 5 mM iron for 30 min were analyzed immediately and 1, 2, and 4 days later. CXCR4 was labeled with an APC-labeled anti-human-CXCR4 antibody (BD Pharmingen) and with an antibody against epitope 6H8 of the N-terminal extracellular domain of the CXCR4 receptor. Cells were fixed with 2% formaldehyde and analyzed with a FACStar flow cytometer (Becton Dickinson).

Annexin V-FITC (green fluorescent dye) and propidium iodide (red fluorescent dye) were combined (kit Sigma APO-AF) to determine the effect of magnetic labeling on cell death. Magnetically labeled cells and control untreated cells were detached and 0.5 million cells were transferred in a staining tube and washed twice with PBS and once with staining buffer (10 mM HEPES, 140 mM NaCl, 2.5 mM CaCl₂, pH 7.5). They were then dispersed in 500 μl of staining buffer with 1 g/ml annexin V-FITC and 2 g/ml propidium iodide and incubated for 10 min in the dark, at room temperature. The fluorescence intensities of 1,000 cells were then quantified using fluorescence microscopy (DMIRB Leica) and image acquisition (Micromax, Princeton Instruments). Each determination was performed in triplicate.

For in vivo experiments, EPCs were incubated for 30 min with AMNPs at an iron concentration of 5 mM and then cultured overnight in EBM-2 culture medium before injection.

Magnetophoretic Mobility of Labeled EPCs

The magnetophoretic mobility of iron-loaded cells was measured by means of single-cell magnetophoresis, using a neodymium-iron-boron cylindrical permanent magnet (diameter 20 mm, height 8 mm) to create a magnetic field gradient. The same magnet was used for in vivo experiments (see below). The magnetic field lines around the magnet were mapped by point-by-point magnetic field measurements with a Hall effect gaussmeter. Iron-loaded cells were trypsinized and suspended in 200 μl of medium in a glass chamber placed 6 mm from the magnet surface. The experimental setup was adapted to a bright-field microscope equipped with a 10x objective. Cell displacement along the magnetic field gradient was monitored with a CCD video camera and analyzed with dedicated software (NIH, Bethesda, USA). The diameter D and velocity v of 100 different cells were analyzed for each labeling condition (0.5, 2, 5, and 10 mM extracellular iron for 30 min).

Cell velocity was constant in steady-state conditions in a region with a uniform magnetic field gradient, reflecting the balance between the friction drag force F_D = 3πη Dν (where η = 10⁻³ Pa•s is the medium viscosity) and the magnetic force experienced by the cell, F_m = M = dB/dz, where Mis the cell magnetic moment in the field B(z), and dB/dz the magnetic field gradient. Knowing the magnetic field and magnetic field gradient, the magnetic moment of a cell can be deduced from its measured velocity and diameter, yielding the iron mass per cell (42).

Animal Model, Cell Transfer, and Magnetic Targeting

Animal experiments were performed in compliance with the French law and under the supervision of a investigator approved for animal experiments. Forty immunocompetent female Wistar rats were used (mean weight 250 g, Charles River, France). They were anesthetized with isoflurane (3% at induction, 2% for maintenance; Baxter) and tracheally ventilated at a rate of 70/min with a 0.2 ml average insufflate volume (Alphalab, Minerve). Analgesia was induced with 10 mg/kg ketoprofen (Merial) subcutaneously. The heart was exposed by sternotomy and the left coronary artery was occluded for 30 min, between the pulmonary artery trunk and the left atrial appendage, followed by 20 min of reperfusion. The aorta was then briefly cross-clamped and the rats were randomly divided into three treatment groups (Table 1). Group 1 received 500,000 iron-loaded EPCs and magnetic guidance (see below); group 2 received 500,000 iron-loaded EPCs, without magnetic guidance; and group 3 received culture medium alone.

Table 1.

Numbers of Animals in the Two Cell Transfer Groups and in the Different Analyses (MRI, Immunohistochemistry, and RT-PCR)

	Group 1: Magnet (n = 14)	Group 2: No Magnet (n = 10)	Group 3: Control (n = 7)
MRI	n = 4	n = 4
Immunohistology	n = 4 *	n = 3 †
RT-PCR	n = 10 ‡	n = 7 §	n = 7

*

Two of the four rats were also imaged in MRI.

†

One of the three rats was also imaged in MRI.

‡‡

Twoof the 10 rats were also imaged in MRI.

§

Three of the seven rats were also imaged in MRI.

In group 1, the same cylindrical magnet as that used for in vitro studies was placed anterior to the heart for 1 min immediately after EPC injection. Then, at the end of the surgical procedure, a smaller circular neodymium magnet (7 mm diameter, 5 mm thickness), providing a field of 0.1 Tesla and a gradient of 11 T/m at 4 mm from its surface, was implanted subcutaneously for 24 h.

Assessment of Cardiac Function

Pre- and postinfarct cardiac function was evaluated by transthoracic echocardiography (Sequoia 516, Siemens, equipped with a 15-MHz transducer) in animals sedated with 2% isoflurane (Baxter) to confirm myocardial injury. Parasternal long- and short-axis views were obtained with both M-mode and two-dimensional images. The left ventricular end-diastolic surface area (LVEDS), LV end-systolic surface area (LVESS), LV end-diastolic length (LVEDL), and LV end-systolic length (LVESL) were measured on parasternal long axis views with two-dimensional images. Volumes were calculated as (8/3Π) x (surface²/length). The ejection fraction was calculated as (LVEDV – LVESV)/LVESV x 100.

EPC Monitoring by MRI

Four animals each from groups 1 and 2 were imaged on a 1.5 T clinical whole-body magnetic resonance (MR) scanner using a 47-mm surface coil (Philips) under general anesthesia (ketamine and chlorpromazine) the day after surgery. The magnet was removed from group 1 animals before MRI via a lateral excision, and the surgical incision was sutured. Partial cardiac gating was used to optimize image acquisition, resulting in synchronization on about one half of heart beats. Images were acquired in strict axial, two-cavity, and four-cavity views, using CINE gradient echo sequences with T1 weighting [repetition time TR = 348 ms, echo time TE = 19 ms, tilt angle α = 18°, field of view (FOV): 110 × 87 mm, matrix: 276 × 224, slice thickness: 2 mm, number of slices: 12, number of scan acquisitions (NSA): 2, acquisition time: 4 min 50 s] and T2* weighting (TR = 25 ms, TE = 4.6 ms, α = 35°, FOV: 100 × 100 mm, matrix: 248 × 248, slice thickness: 1 mm, number of slices: 25, NSA: 3, acquisition time 4 min 15 s). After in vivo imaging the animals were killed and the hearts were explanted and imaged inside a tube containing CryoMatrix (Thermoscientific), using the same parameters as in vivo. The hearts were then placed in liquid nitrogen and stored at −80°C for further histological and RT-PCR analysis.

Histological and Immunohistochemical Analyses

Twenty-four hours after EPC injection, the hearts were explanted and cut into two halves that were immediately fixed in Optimal Temperature Cutting Medium (Tissutec) and placed in liquid nitrogen. Blocks were sliced into 7-μm-thick cryosections using an ultramicrotome (LM 1850, Leica) and examined with a light microscope (Leica DMIL, Wetzlar, Germany) equipped with a digital camera (Qicam, Qimaging, Burnaby, BC, Canada). The infarct area was identified by standard hematoxylin and eosin staining. EPCs were detected by immunofluorescence with an antibody directed against a specific human nuclear protein (lamin A/C, Novocastra). Pearls iron stain was used to verify the colocalization of the iron label with EPCs. For this purpose, slices were incubated for 20 min with 2% potassium ferrocyanide in 3.7% hypochloric acid, then washed and counterstained with nuclear fast red before observation.

Polymerase Chain Reaction (PCR) Analysis and Real-Time RT-PCR

RNA was extracted from whole hearts with the RNA Mini kit from Quiagen. One microgram of RNA was reverse-transcribed with Moloney murine leukemia virus (Mu-MLV) reverse transcriptase (Invitrogen, Cergy, France) and oligo(16)dT. Real-time quantitative PCR was performed with a Light Cycler (Roche Diagnostics, Mannheim, Germany) or a Chromo4 thermal cycler (Biorad). The reaction mixture contained 12 μl of Roche SYBR Green I mix (including Taq DNA polymerase, reaction buffer, deoxynucleoside trisphosphate mix, SYBR Green I dye, 3 mM MgCl₂), 0.25 μM appropriate forward and reverse primers, and 2 μl of cDNA. The amplification program included initial denaturation at 95°C for 15 or 8 min, and 40 cycles of denaturation at 95°C for 10 s, annealing at 65°C for 8 s (Light Cycler), and extension at 72°C for 8 or 30 s. The temperature transition rate was 20°C/s (Light Cycler). Fluorescence was measured at the end of each extension step. Gene expression was determined by comparing the fluorescence intensity of experimental samples with that of a control series containing known concentrations of total human cDNA (ng/μl).

Statistical Analysis

Mice were classified according to their order of arrival using a randomized list. The nonparametric Wilcoxon and Kruskall-Wallis tests were used. Statistical analyses were done with SAS software v8.2. The continuous data are presented as mean values and range (min to max). All tests were one-sided. Statistical significance was assumed at values of p < 0.05.

Results

Magnetic Labeling of EPCs

Transmission electron microscopy images of magnetically labeled EPCs (Fig. 1A) showed intraendosomal confinement of magnetic nanoparticles on days 1 and 10. Anionic citrate-coated maghemite nanoparticles are spontaneously adsorbed to the plasma membrane before being internalized by the endocytosis pathway (41). They are then sequestered in lysosome-like vesicles, which distribute among dividing cells. Cell proliferation was not adversely affected by magnetic labeling with iron concentrations up to 10 mM (Fig. 1B). EPC migration along a VEGF gradient (0–40 ng/ml) was similar to that of untreated cells after labeling with 2 mM (p = 0.23) and 5 mM (p = 0.14) extracellular iron, but was significantly reduced at 10 mM iron (p = 0.017) (Fig. 1C). CXCR4 expression was not affected by labeling with 5 mM iron, as measured on days 0, 1, 2, and 4 postlabeling.

Figure 1.

Endothelial progenitor cell (EPC) magnetic labeling. (A) Transmission electron microscopy showed efficient cell capture and intracellular confinement of anionic nanoparticles inside endosomal compartments (labeling conditions: 5 mM, 30 min). On day 10 postlabeling, nanoparticles were still concentrated inside lysosome-like intracellular structures, which become scarcer as the cells divided. (B) EPC proliferation for up to 5 days of culture was not impaired by magnetic labeling with up to 10 mM extracellular iron. Bars represent the standard deviation of cell numbers in triplicate experiments. (C) EPC migration along a vascular endothelial growth factor (VEGF) gradient, measured in Boyden chambers, was not significantly affected by labeling with 2 mM and 5 mM extracellular iron, compared to untreated cells (100% migration), but was reduced after labeling with 10 mM iron.

To evaluate the cytotoxicity of the magnetic labeling, we compared the amount of apoptotic cells and necrotic cells for control EPCs and magnetically labeled EPCs, by staining the cells with annexin V and propidium iodide. In control experiments (untreated cells), 4.9 ± 2.8% of the cells were stained by annexin V and 4.2 ± 2.3% by propidium iodide. Magnetic labeling failed to increase these percentages, which averaged 5.4 ± 2.8% and 4.0 ± 1.7%, respectively.

Magnetic Mobility of Labeled EPCs

The magnet used for EPC targeting was chosen for its dimensions and magnetic properties. The magnetic field strength along the magnet axis (B) fell from 320 mT at the surface to 100 mT at a distance of 11 mm. The corresponding magnetic field gradient (grad B) ranged from 32 to 15 T/m (Fig. 2A). The magnetophoretic mobility of iron-loaded cells was assessed via an observation window located 6 mm from the magnet, with a field strength of 180 mT and a field gradient of 19 T/m. In this magnetic field, loaded cells moved along the magnetic field gradient (Fig. 2B). After loading with 5 mM extracellular iron for 30 min, cell velocities ranged from 40 to 140 μm/s (Fig. 2C), with a mean of 77 ± 20 μm/s. This distribution reflected the heterogeneous iron uptake by the cell population. The mean magnetic mobility rose from 77 ± 12 to 90 ± 25 μm/s when the iron concentration rose from 5 to 10 mM, with saturation occurring at 10 mM (Fig. 2D). Assuming a balance between viscous drag (proportional to cell velocity) and the magnetic force exerted on the cell, we deduced that the range of magnetic force was 4–15 pN. Taking into account the impaired migration capacity after incubation with 10 mM iron and the small increase in magnetization between 5 and 10 mM iron, we chose to use 5 mM iron labeling for 30 min for in vivo experiments. In these conditions the force experienced by the cell was 11.3 ± 3.4 pN and the iron load per cell 10.1 ± 3.1 pg.

Figure 2.

Magnetic properties of labeled cells. (A) Quantification of the magnetic field (B) and the magnetic field gradient (grad B) generated along the axis of the magnet used for cell targeting. The window indicates the area in which magnetic cell mobility was measured. (B) Successive shots (1-s intervals) of cells migrating along the magnetic field gradient. (C) Histogram of magnetic mobility of cells labeled with 5 mM iron for 30 min. (D) Mean magnetic mobility as a function of the extracellular iron concentration. Bars represent the dispersion of mobility values within the cell population for each condition, as represented in the histogram. Magnetic mobility began to saturate at an iron concentration of 10 mM. v: velocity.

Surgical Model

Ligation of the anterior coronary artery generated an apical infarct demonstrated by macroscopic blanching. To validate the procedure of brief aortic cross-clamping for ensuring perfusion of the coronary arteries, preliminary studies were performed with methylene blue injected into the left ventricle while the aorta was cross-clamped. This resulted in the entire heart becoming blue and dilated, with no apparent passage of the dye into the ascending or descending aorta. Furthermore, the presence of a myocardial infarct was systematically controlled by functional measurements. Echocardiography performed before the operation and for up to 2 h after reperfusion revealed absolute reductions of 14% in the shortening fraction and 24% in the ejection fraction in the nonmagnet groups, indicating the efficiency of the surgical procedure to induce infarction. In the magnet groups, post surgery echocardiography was made impossible due to the position of the magnet, but hematoxylin and eosin staining confirmed the presence of a subendocardial myocardial infarct in each rat (Fig. 3D).

Figure 3.

(A) Ligation of the anterior coronary artery. Following median sternotomy a finger's breadth above the costal margin, the pericardium was opened and the anterior coronary artery was circled with a 7-0 suture, 3–5 mm distal to its origin, between the pulmonary artery and the left atrium, and then transiently occluded by means of a tourniquet. (B) Aortic clamping. EPCs were injected into the left ventricle during the acute phase of myocardial infarction, 20 min after the onset of reperfusion. The ascending aorta was occluded with a vessel loop for the 20 s required to inject the cells. (C) Implantation of the second magnet. A few seconds after EPC injection a circular magnet 20 mm in diameter and 7 mm thick was placed above the heart for 30 s. Then, when the sternotomy incision had been closed, a hollow-centered circular magnet 7 mm in diameter and 5 mm thick, with a field strength of 0.3 T and a gradient of 11 T/m, was implanted subcutaneously above the cardiac cavity for 24 h. (D) Subendocardial ischemia. Histologic studies showed subendocardial hemorrhagic necrosis (hematoxylin and eosin staining), reflecting the presence of a subendocardial infarct (circle).

Assessment of Magnetic Targeting

Immunolabeling with primary antibody against a human protein (A/C lamin) was used to identify injected EPCs. Lamin A/C-positive cells also stained positively for Pearl's reagent, thereby demonstrating the presence of iron (Fig. 4A). In the two groups of EPC-treated rats (groups 1 and 2), 30% of the hearts were analyzed by immunofluorescence. In group 1 (magnet), three of the four hearts contained a few EPCs, mainly located near the base of the heart, while the fourth contained a very large number of cells. In contrast, in group 2 (no magnet), the three examined hearts contained very few widely scattered cells, located at the apex or the base. The numbers of cells detected in each heart are shown in Figure 4B, in the form of a histogram representing the number of immunofluorescence sections (n_slides) in which a given range of cells (n_cells) was observed. In group 2 (no magnet), most of the sections contained fewer than 20 EPCs (Fig. 4C), while in group 1 (magnet), most sections contained 20–40 cells, with some harboring 100 or more cells (Fig. 4D). On average, the mean number of human cells per sections [mean value (range)] was 10 times higher in the magnetic targeting group than in the EPC control group [23 (0–198) vs. 2 (0–24)], p < 0.0001.

Figure 4.

(A) Correlation between immunolabeling (human cells revealed by green fluorescence) and iron labeling (blue color with Pearl's stain). The fluorescence image corresponds to a section from a group 1 heart containing 20–40 cells (magnet). A good colocalization is found. (B) Histogram of immunofluorescence intensity (n_slides) according to the number of cells (n_cells) present in each image. In the magnet group (dark bars), more labeled cells are seen, and the median is shifted to the right compared to the no magnet group. (C) Typical group 2 section (no magnet), containing only a few cells (0–20). (D) Typical group 1 section, containing abundant cells (80–100) (magnet). Scale bar: 100 μm (A–D).

RNA was extracted from 10 hearts in the magnet group, 7 hearts in the EPC control group, and 2 hearts in the medium-only control group. The amount of human cDNA tended to be higher, although not significantly, in the rats with an implanted magnet, as shown in Figure 5A. In the no-magnet group, cDNA values between 0 and 20 ng/μl were most frequent, while values between 20 and 40 ng/μl were prevalent in the magnet group. However, the data were not statistically significant (p = 0.107).

Figure 5.

(A) Histogram of the numbers of rat hearts containing given amounts of human cDNA (ng/μl), showing higher values in group 1 (magnet) than in group 2 (no magnet), indicating that more cells were grafted in the magnet group. (B) MRI T2* (top) and T1 (bottom) gradient echo images of the explanted heart of a rat in group 2 (no magnet). No hyposignals were visible in the ventricles. The cDNA value in this rat was within the range 0–20 ng/μl. (C) T2* (top) and T1 (bottom) axial (left) and transverse (right) MRI sections of an explanted heart from group 1 (magnet) with a cDNA value in the range 40–60 ng/μl. Note the marked hyposignal in the left ventricle induced by magnetic cells (arrow). (D) In vivo T2* MRI (left) of a rat from group 2 (no magnet). Note the clear hyposignal at the base of the left ventricle (arrow). T2* images of the explanted heart (right) showed the same hyposignal in the left ventricle (arrow). The cDNA value in this rat was within the range 20–40 ng/μl. This indicates partial homing of labeled cells, even without magnet.

MRI Detection of Injected Cells

Four rats in group 1 (magnet) and four rats in group 2 (no magnet) were examined by MRI (Table 1). Hypo-intensities representing labeled EPCs were seen on T2*-weighted contrast-enhanced images of two rats, at the base and lower wall of the LV (Fig. 5C, D) in the infarcted tissue and also remote in normal myocardium. The MRI findings correlated well with the quantitative PCR results. The MRI findings were consistent with the quantitative PCR results and the histologically identified cell clusters.

Discussion

The major finding of this study is that magnetic targeting may be an efficient means of enhancing homing of systemically delivered cells to sites of myocardial injury. This conclusion is based on immunofluorescence quantification of the injected cells, quantitative PCR of human cDNA, and MRI cell imaging. In this xenogeneic setting, the use of human-specific markers avoided to mistakenly conclude that the grafted cells are still present whereas iron released upon their death has been engulfed in macrophages and subsequently gives rises to false positive signals (2). Although between-group differences failed to reach statistical significance using RT-PCR, quantification of cell numbers per histological section indicated a statistically relevant difference in favor of the magnet-treated hearts.

Over the past years, early reperfusion of myocardial infarctions has resulted in significant improvements in outcomes. However, some patients still experience a poor prognosis, particularly those presenting late or with a large anterior infarction. Under these circumstances, the major complication is the late development of left ventricular remodeling with its attendant risk of heart failure. This mandates the search for additional therapeutic interventions, among which cell therapy has gained an increased interest. Primarily for practicality reasons (easiness of harvest, immediate availability, autologous origin, possibility of use without major processing, and small size), bone marrow-derived mononuclear cells have been the most widely investigated in this setting under the form of intracavitary infusions into the reopened infarct-related vessel (5,15,18). Although the initial premise was that these cells would effect some regeneration of the infarcted myocardium (28), the paradigm has now shifted towards predominant structural (increased wall thickness) and paracrine (activation of host-associated cytoprotective signaling pathways by cell-released mediators) effects (17). While these effects have been extensively documented in animal studies, clinical results have yielded less optimistic results with only modest improvements in left ventricular ejection fractions and volumes (25). Several factors may account for these suboptimal outcomes, among which a low myocardial retention rate of intracoronarily infused cells (21). Consequently, interventions that would increase this retention rate are now recognized important for improving the overall efficacy of the procedure. These interventions basically fall into three main categories. A first option is to rethink the design of the delivery devices, which has been done with a recently introduced catheter fitted with an expandable transendothelial needle, which allows a perivascular transfer of cells, directly into the myocardial tissue (39). A second category of interventions targets the cells themselves and entails preimplantation manipulations aimed at overexpression of cell surface adhesion molecules like E-selectin (35), β-2 integrins (9), or chemokines like CXCR4, which is the cognate receptor for stromal-derived factor (SDF)-1, a chemokine required for homing of progenitor cells to ischemic tissues (11). From a clinical standpoint, however, these manipulations are fraught with several limitations because they may require a time period not compatible with urgent treatments or involve genetic engineering, which carries its own safety issues. The third category of interventions targets the myocardial tissue directly. While biologically-oriented treatments such as intramyocardial injections of SDF-1 (36) or similar chemokines like the stem cell factor (27) look difficult to implement clinically in the setting of an acute infarct, such may be not the case for much less invasive physically oriented treatments. The latter basically comprise low-energy shock wave (1), ultrasound-mediated destruction of microbubbles, which induces a transient premeabilization of the endothelial wall facilitating cell trafficking (16), or magnetic targeting. This strategy has recently been shown effective in a rat model of carotid artery injury (23) and, more recently, of myocardial infarction (10) where cardiospheres were injected directly into the left ventricular wall under magnet application. However, to the best of our knowledge, it has not yet been tested in the more clinically relevant and less invasive setting of an intravascular approach simulating intracoronary injections.

A prerequisite for an objective assessment of techniques designed to enhance homing of circulating stem cells is the ability to track their fate in vivo. Among the various imaging modalities, cell loading with magnetic nanoparticles and their subsequent visualization by MRI has gained a growing interest (6). Recently, to assess cell localization in cardiac therapy, bone marrow-derived stromal cells were labeled with iron oxide nanoparticles for in vivo cell detection by MRI (8,38). Cells could be monitored for 10 weeks after infarction and cell therapy, demonstrating that MRI cell imaging is a valuable technique to measure the time course of cell migration and the degree of cell homing during cardiac cell therapy. Here, using a rat model of myocardial infarction and intracavitary injection of endothelial progenitor cells labeled with iron oxide nanoparticles, we demonstrate a good correlation between MRI noninvasive follow-up of the injected cells and immunofluorescence or quantitative PCR data.

Besides, the ‘nanomagnetic” behavior of the nanoparticles in the presence of a magnetic field opens up other applications, such as magnetic vectorization. We have previously succeeded in guiding synthetic magnetic vectors such as magnetoliposomes to tumor sites (13) and to the cerebral microcirculation (34). Other recent studies have shown that iron-loaded hepatic cells and stem cells can be retained in the liver with an external magnet after intrasplenic and intravenous injection, respectively. Very recently, magnetically labeled endothelial progenitor cells were successfully attracted and retained at sites of carotid artery injury by using an external magnet (23). Likewise, Cheng et al. have reported that cell retention could be markedly improved when intramyocardial injections of cardiospheres were combined with magnet application over the apex of the heart. It is noteworthy that although this study differs from ours by the type and number of cells, the magnetic targeting procedure and the route of cell delivery, it reported that external magnetic field application tripled the number of engrafted cells, an increase quite similar to that found in the present study. The clinical relevance of these observations is illustrated by the good correlation between higher cell retention rates and improved left ventricular function (10). The use of external magnets will necessarily be limited to superficial body sites, but implanted magnets could be used for deeper sites. Recently, magnetic stents were developed to attract and retain therapeutic agents (magnetic cells or microspheres). They are composed either of permanent magnetic material (31) or material that can be magnetized by a uniform external field (12,32). They allow the target zone to be treated repeatedly by simple intravenous injection of magnetic vectors and exposure to a uniform field. Furthermore, the development of electromagnetic catheters might represent an alternate approach for driving cells towards discrete myocardial areas of focused magnetic field. Here we show that an external magnet can be used to improve cell retention in cardiac muscle in blood flow conditions.

The first limitation of this study pertains to the surgical model. The requirement for a simultaneous intraventricular injection and aortic cross-clamping results in an unavoidable variability in intracavitary cell distribution but our dye experiments as well as previous studies (19,33) using a similar protocol have nevertheless documented the ability of such intracavitary injections under transient aortic cross-clamping to result in effective intracavitary cell delivery. Based on the current results, statistical calculations indicate that a threefold higher number of animals would have been required to demonstrate statistical significance. Nevertheless, the present data on intramyocardial cell homing in the magnet-treated group of animals yet provide an encouraging efficacy signal. The 24-h time point after cell injection was chosen in order to reduce animal mortality. Further studies are warranted to follow cell homing at later time points. It should also be noted that the magnitude of homing may also have been reduced by the relatively small number of cells, a decision dictated by their large size (13 μm) and the resulting risk of coronary thrombosis. Furthermore, the magnet was chosen for its weight and shape compatible with subcutaneous implantation in rats. The magnetic field gradient, which largely determines the efficiency of vectorisation, could be further optimized. Finally, cardiac MR imaging was performed with a 1.5-T device equipped with a standard antenna, whereas spatial resolution could be markedly increased by using a stronger field (4.7 T for example) and a cooled surface antenna.

In conclusion, in a rat model of myocardial infarction, this pilot proof-of-concept study suggests that homing of circulating stem cells can be improved by magnetic targeting and warrants additional benchwork to confirm the validity of the concept, optimize parameters of targeting and assess the relevance of this approach in a clinically relevant large animal model.