Magnetic Resonance Imaging and Confocal Microscopy Studies of Magnetically Labeled Endothelial Progenitor Cells Trafficking to Sites of Tumor Angiogenesis



This article is corrected by:

  1. Errata: Magnetic resonance imaging and confocal microscopy studies of magnetically labeled endothelial progenitor cells trafficking to sites of tumor angiogenesis Volume 24, Issue 4, 1142, Article first published online: April 2006


AC133 cells, a subpopulation of CD34+ hematopoietic stem cells, can transform into endothelial cells that may integrate into the neovasculature of tumors or ischemic tissue. Most current imaging modalities do not allow monitoring of early migration and incorporation of endothelial progenitor cells (EPCs) into tumor neovasculature. The goals of this study were to use magnetic resonance imaging (MRI) to track the migration and incorporation of intravenously injected, magnetically labeled EPCs into the blood vessels in a rapidly growing flank tumor model and to determine whether the pattern of EPC incorporation is related to the time of injection or tumor size. Materials and Methods: EPCs labeled with ferumoxide–protamine sulfate (FePro) complexes were injected into mice bearing xenografted glioma, and MRI was obtained at different stages of tumor development and size. Results: Migration and incorporation of labeled EPCs into tumor neovasculature were detected as low signal intensity on MRI at the tumor periphery as early as 3 days after EPC administration in preformed tumors. However, low signal intensities were not observed in tumors implanted at the time of EPC administration until tumor size reached 1 cm at 12 to 14 days. Prussian blue staining showed iron-positive cells at the sites corresponding to low signal intensity on MRI. Confocal microcopy showed incorporation into the neovasculature, and immunohistochemistry clearly demonstrated the transformation of the administered EPCs into endothelial cells. Conclusion: MRI demonstrated the incorporation of FePro-labeled human CD34+/AC133+ EPCs into the neovasculature of implanted flank tumors.


Tumor growth and metastasis usually depend on formation of new blood vessels. Chemokines released by tumor cells promote activation, proliferation, and migration of endothelial cells to the tumor tissue [14], allowing rapid formation of functional neovasculature. Circulating endothelial cells contributing to tumor angiogenesis can originate from sprouting and co-option of neighboring pre-existing vessels [5, 6]. There is accumulating evidence that tumor angiogenesis may be supported by mobilization and functional incorporation of bone marrow– derived endothelial progenitor cells (EPCs) that promote the growth of certain tumors [4, 711]. EPCs have been detected in the circulation of lymphoma-bearing mice and patients with cancer [12]. When infused into immunocompromised mice, they incorporate into the vasculature of xenotransplanted tumors to a degree related to tumor volume and production of vascular endothelial growth factor (VEGF) [1214]. EPCs can be mobilized from the bone marrow by administering granulocyte colony-stimulating factor (G-CSF) to patients with cancer and healthy individuals [15, 16]. Expression of the cell surface marker CD133 (AC133) on a subpopulation of CD34+ human hematopoietic stem cells has been shown to indicate cells destined for endothelial differentiation and angiogenesis [17, 18].

The process of stimulation, release, and recruitment of EPCs from the bone marrow microenvironment to contribute to tumor angiogenesis is complex and multifactorial [14]. The relationship of the temporal and spatial distribution of EPCs recruited from the bone marrow to tumor growth is not completely understood. It is not clear whether EPCs are required for long-term integrity and establishment of tumor neovasculatures. EPCs are shown to contribute to vessel formation in the periphery and in the core of the tumor mass both experimentally and clinically [19]. Noninvasive imaging methods may provide for the localization of EPCs and hematopoietic stem cells within the tumor, furthering the understanding of extracellular and stromal components required for incorporation of these cells into the neovasculature and why certain tumors respond only partially to therapy. However, current imaging modalities do not allow the monitoring of the early migration and incorporation of EPCs into tumor neovasculature.

By complexing ferumoxides, a dextran-coated superpara-magnetic iron oxide (SPIO) nanoparticle approved by the U.S. Food and Drug Administration (FDA), to the transfection agent protamine sulfate, it is possible to magnetically label stem cells and detect them by magnetic resonance imaging (MRI) [20]. By magnetically labeling mouse Sca1+ hematopoietic stem cells, Anderson et al. used MRI to demonstrate the migration of SPIO-labeled Sca1+ cells into the surrounding neovasculature in an implanted glioma in the mouse brain [21]. Based on these results, by combining high-resolution MRI and magnetically labeled EPCs, it should be possible to detect the early migration and incorporation of these cells into the neovasculature in developing and invading tumors. Moreover, these results should provide insight into angiogenesis and possible strategies for intervention.

The primary purpose of this study was to track the migration and incorporation of intravenously injected, magnetically labeled human CD34+/AC133+ cells into the angiogenesis of rapidly growing flank tumor, and to determine whether patterns of EPC incorporation are related to time of injection or size of the tumor. The findings of MRI were correlated with histology, immunohistochemistry, and confocal microscopy of the tissues.

Materials and Methods

Preparation of EPCs (AC133+ Cells)

CD133+ (AC133) cells, a subpopulation of human CD34+ hematopoietic stem cells, were obtained from healthy volunteers entered on protocols approved by the Intramural Review Board at our institution. Each volunteer was mobilized with G-CSF (filgrastim; Amgen, Thousand Oaks, CA, 10 μg/kg per day for 5 days subcutaneously prior to automated apheresis on the Fenwal CS3000Plus cell separator (Baxter Healthcare, Deerfield, IL,, using peripheral venous access and anticoagulation with anticoagulant citrate dextrose solution formula A (Baxter Healthcare). The peripheral blood progenitor cell product was enriched for AC133+ cells by immunomagnetic selection on the CliniMacs system (Miltenyi Biotec Inc., Auburn, CA, Freshly prepared or cryopreserved/thawed AC133 cells were incubated in Stemspan media (StemCell Technologies, Vancouver, BC, Canada, containing 40 ng/ml stem cell factor, 40 ng/ml FMS-related tyrosine kinase-3, and 10 ng/ml thrombopoietin (all from Sigma-Aldrich, St. Louis, and incubated for at least 5 days, with the cell concentration kept at 1 × 106 per ml by adding fresh media on alternate days. On the day of cell labeling, cell concentration in the culture was determined and all cells were centrifuged, with the supernatant retained for future use (see below). Cells were analyzed for labeling efficiency, viability, and phenotypic markers as previously described [20].

Animal Model

All animal experiments were performed according to a protocol approved by our animal care and use committee (ACUC). For all studies, 1 × 106 cells from a C-6 rat glioma cell line (National Cancer Institute, Bethesda, MD, in 100 μl of media were implanted subcutaneously into the flanks of 6- to 8-week-old female BALB/c nude mice (Charles River Laboratories, Inc., Wilmington, MA, The rat glioma cell was chosen because the tumor was 100% reproducible, grows symmetrically in all animals, is hypervascular, and when approximately 1 cm in size is found to have little evidence of necrosis on pathology. An ectopic site was chosen for the glioma so that when the tumor grew to a predetermined size we could intervene with the infusion of stem cells intravenously or with MRI acquisition. Two sets of experiments were performed to determine the incorporation of labeled CD34+/AC133+ cells into the implanted tumor. In group 1, 3 × 106 labeled (n = 15) or unlabeled (n = 15) AC133+ cells were injected into the tail vein immediately after the subcutaneous implantation of tumor cells. In group 2, after the flank tumors were allowed to grow to approximately 0.2 cm in size (typically at day 3 after implantation), 3 × 106 labeled (n = 15) or unlabeled (n = 15) AC133+ cells were injected into the tail vein.

MRI was performed in group 1 and group 2 mice when tumors were approximately 0.5, 1 and 1.5 cm in size, corresponding to day 7, 14, and 21 after implantation. At least four mice from each group of animals receiving either ferumoxide–protamine sulfate (FePro)-labeled or -unlabeled AC133+ cells (day 7, 14, and 21 after implantation) at each time point underwent an MRI exam (including both in vivo and ex vivo MRI). Animals from each group receiving labeled or unlabeled cells at each time point were randomly selected for in vivo MRI. Because of restrictions in the approved ACUC protocol, serial MRI exams could not be performed in the same mouse to monitor disease progression over time. We compensated for this shortcoming of the study design by injecting/implanting multiple animals (at least six) at a time; we then randomly selected animals (two at each time point) for imaging and euthanized them at different time points (at tumor sizes of 0.5, 1, and 1.5 cm).

Preparation of FePro Complex and Labeling AC133+ Cells

The commercially available ferumoxide suspension (Feridex IV; Berlex, Wayne, NJ, contains particles approximately 80–150 nm in size and has a total iron content of 11.2 mg/ml (11.2 of iron μg/μl). Preservative-free protamine sulfate (American Pharmaceutical Partners, Inc., Schaumburg, IL,, 10 mg/ml, was prepared as a fresh stock solution of 1 mg/ml in distilled water at the time of use. Ferumoxides at a concentration of 100 μg/ml was put into a mixing flask or tube containing serum-free RPMI 1640 medium (Biosource, Camarillo, CA, containing 25 mM HEPES, minimum essential medium nonessential amino acids, sodium pyruvate, and l-glutamine. Protamine sulfate (6 μg/ml) was added, and the entire suspension was mixed for 5–10 minutes. The final FePro suspension was added directly to the cells at a concentration of 4 × 106 cells per ml and incubated for 2–3 hours, and then an equal volume of the originally collected supernatant was added to the cells [20]. The final concentrations of ferumoxides and protamine sulfate were 50 μg/ml and 3 μg/ml of medium, respectively. The cell suspension was then incubated overnight. After overnight incubation, cells were collected, washed twice with sterile phosphate-buffered saline (PBS), and resuspended at a concentration of 1.5 × 107 cells per ml in PBS for administration to the mice. Each mouse received either 3 × 106 labeled or unlabeled cells by tail vein injection. To investigate the contribution and incorporation of AC133+ cells into tumor neovasculature by confocal or fluorescent microscopy, FePro-labeled cells were also incubated with quantum dots (565 nm; Quantum Dot Corp., Hayward, CA, for 30–45 minutes, washed again with PBS, and resuspended at 1.5 × 107 cells per ml. Double-labeled (i.e., labeled with FePro and quantum dots) AC133 cells were injected intravenously in group 1 and group 2 mice (n = 3).

In Vivo MRI Protocol

Mice were imaged at 7-Tesla using a 22-cm horizontal bore MRI system (Bruker, Billerica, MA, with 39-Gauss/cm gradients and a 35-mm transmit-receive bird-cage volume coil. Mice were imaged in a head holder with 1.5%–2% isoflurane anesthesia by nosecone and physiological monitoring. Axial images over the flank were acquired with a 2.6-cm square field of view (FOV) and 0.5-mm slice thickness. MRI was performed using a two-dimensional, T2*-weighted gradient echo pulse sequence (TR [repetition time]/TE [echo time] = 500/4.3 milliseconds, number of excitations [NEX] = 8, 384 × 384 matrix zero-filled to 512 × 512, 70 × 70 × 500 μm resolution), and a two-dimensional, T2-weighted RARE (rapid acquisition with relaxation enhancement) sequence (TR/TE = 2400/7.0 milliseconds; RARE factor 8, NEX = 8, 256 × 256 matrix, 100 × 100 × 500 μm resolution).

Ex Vivo MRI Protocol

Mice were euthanized at selected time points for ex vivo MRI followed by histological analysis of the tumors. The mice were deeply sedated using CO2 inhalation and euthanized by cervical dislocation. Soon after cervical dislocation, the chest was opened quickly and intracardiac (through left ventricle) injection of PBS followed by 4% paraformaldehyde (PFA) was made to clear the blood from the vessels. Injected fluid mixed with blood was drained through a hole made in the right auricle. Tumors with surrounding tissues were removed immediately and put into 4% PFA and 3% sucrose. Fixed tumors were imaged in perfluoroether (Fomblin; Ausimont USA, Thorofare, NJ, and in groups of two to four in either a 7-Tesla, horizontal-bore MR imaging system (Bruker) with 39-Gauss/cm gradients, or a 7-Tesla vertical-bore micro-imaging system (Bruker) with 95-Gauss/cm gradients. Tumors were imaged with T2*-weighted, three-dimensional gradient echoes, at 60- to 90-μm isotropic resolution (TR/TE = 280/5.3 milliseconds, NEX = 4). FOV ranged from 2.3–2.7 cm in plane and 3.2 to 3.7 cm in the read dimension for multiple tumors imaged simultaneously. Total time required to obtain the images was more than 20 hours.

Confocal Microscopy

All mice injected with double-labeled (FePro plus quantum dots with green fluorescence 565 nm). AC133 cells were administered rhodamine-labeled (red color) lectin (Vector Laboratories, Burlingame, CA, intravenously through the tail vein 15 minutes before euthanasia to delineate the lumen of the vessels. Tumors were collected and then divided for frozen section at 50-μm thickness and routine histopathology. The sections were mounted on glass slides and analyzed by confocal microscopy (Leica, Heerbrugg, Switzerland, Some sections were made at 10-μm thickness and analyzed under a fluorescent microscope (Axioplan II; Carl Zeiss, Jena, Germany, followed by Prussian blue staining.

Immunohistochemistry and Prussian Blue Staining

After ex vivo MRI, tumors and surrounding tissues were divided in half for frozen sections, and PFA fixation with embedding in paraffin. Six- to 10-μm sections were cut from representative tumors (at least two) from each experimental setting. Consecutive slides from the paraffin block were evaluated by immunohistochemical techniques for the expression of endothelial cell markers using anti-human CD31 (platelet-endothelial cell adhesion molecule-1), von Willebrand Factor (vWF), and KDR1 (VEGF receptor 2) antibodies (all from Novocastra Ltd., New-castle upon Tyne, U.K., After sections were deparaffined and rehydrated, they were treated with 0.025% trypsin for 30 minutes at 37°C followed by a high-temperature unmasking process. Sections were incubated in 3% hydrogen peroxides in methanol for 5 minutes to block endogenous peroxidase in tissues. After blocking with horse serum, the sections were flooded with corresponding antibodies at 1:50 dilutions and incubated for 1 hour at room temperature. After thorough washing with running distilled water, the sections were flooded with horseradish peroxidase-labeled secondary antibody (DakoCytomation, Carpinteria, CA, for 30 minutes, washed thoroughly, and flooded with substrate solution (DAB [3,3′-diaminobenzidine tetrahydrochloride]) for 2–3 minutes. After thorough washing, the sections were counterstained with nuclear fast red, dehydrated, and coverslipped. Consecutive sections were stained with Perl's reagent for iron staining (Prussian blue).

To determine whether injected human AC133+ cells differentiated toward a macrophage lineage or whether host mouse macrophages phagocytosed the FePro-labeled cells, consecutive slides were stained for both mouse and human macrophages using anti-mouse CD68 (Serotec, Raleigh, NC, and anti-human CD68 antibodies (DakoCytomation), respectively. Slides from frozen sections were also processed using anti-human antibodies for endothelial markers (CD31, CD106, vascular endothelial cell adhesion molecule-1, and KDR1) as primary and fluorescent (fluorescein isothiocyanate)-labeled antibody as secondary and analyzed under a fluorescent microscope (Axioplan II; Zeiss, Oberkochen, Germany) using Axiovision 4 software (Zeiss). The images were processed using Adobe Photoshop 7.0 (Adobe Systems Inc., San Jose, CA,


Cell Labeling

Labeling efficiency with FePro complexes was approximately 100% for the human AC133+ cells. There were no significant differences in viability or proliferation capacity between labeled and unlabeled cells or in phenotypic markers between pre- and postlabeled AC133+ cells (data not shown).

MRI Tracking of FePro-Labeled AC133+ Cell Incorporation into Tumor

In vivo MRI did not differentiate between tumors at 0.5 cm in size in group 1 mice that received labeled or unlabeled AC133 cells at the time of flank tumor implantation. However, in vivo MRI did demonstrate linear hypointense regions within flank tumors that had grown to ≥1 cm (i.e., ≥day 7 after implantation) in animals receiving the i.v. injection of FePro-labeled cells, as compared with animals that received unlabeled cells. The linear (feathering-type pattern) hypointense areas observed in mice receiving labeled cells were easily differentiated as small and well defined dark features, compared with the changes in signal intensity in large tumors of either labeled or unlabeled mice, some of which developed amorphous dark areas in the tumor images indicative of necrosis (≥1.5 cm) (Fig. 1A–1D). The large dark area without the feathering pattern was due to necrosis and most often detected in the center of the tumor. Histology with Prussian blue staining showed a clear difference between the incorporated labeled cells and necrotic areas in larger tumors (1.5 cm; Fig. 1H)

Ex vivo MRI of mice in group 1 clearly demonstrated the incorporation of labeled AC133 cells into tumors of 0.5 cm in size or as early as 5–7 days after implantation of the C-6 glioma cells (Fig. 1E). Labeled AC133 cells were visible as linear hypointense regions at the periphery and center of the ≥1 cm tumors (i.e., ≥12 days after implantation and infusion of labeled cells). These linear hypointense regions within the tumor were due to infiltration of labeled cells and were clearly distinguishable from the changes in signal intensity on T2*-weighted imaging of tumor necrosis observed in the large tumors in mice injected with either labeled or unlabeled AC133 cells (Fig. 1). For group 2 mice receiving labeled cells when tumors were 0.2 cm in size, ex vivo MRI also demonstrated hypointense areas at the margin of the tumor as early as 3–5 days after i.v. administration of FePro-labeled AC133 cells. Compared with the mice in group 1, no differences were observed on in vivo or ex vivo MRI between group 1 and group 2 mice injected with labeled cells when tumors grew to ≥1 cm in size (Fig. 2). Hypointense areas at the tumor margins were not observed on MRI in group 2 mice receiving unlabeled cells (Fig. 2A).

Immunohistochemistry and Prussian Blue Staining

Frozen section (from group 1) and paraffin-embedded section (from group 2) revealed positive human endothelial markers (CD31, KDR1, vWF, and CD106), indicating endothelial transformation of the administered FePro-labeled or unlabeled cells and incorporation into the tumor neovasculature (Fig. 3). Corresponding consecutive sections revealed Prussian blue–positive cells in the same region as cells with positive markers for human endothelial cells (Fig. 3, arrows). None of the mouse macrophages (mCD68-positive cells) infiltrating into the large tumors was positive for iron oxide nanoparticles on Prussian blue staining. However, a few Prussian blue cells at the periphery of the tumor were positive for human macrophage marker (hCD68-positive cells), indicating that some of the administered FePro-labeled AC133 cells differentiated toward the macrophage lineage (Fig. 4).

Confocal and Fluorescent Photomicrography

Confocal microscopic images obtained at 565-nm emissions showed quantum dot-positive cells (green) incorporated in the tumor neovasculature, which was indicated by rhodamine lectin (red). Fluorescent microscopy of thin sections also showed quantum dot– and Prussian blue–positive cells along the endothelial lining of vessels (Fig. 5).


Endothelial progenitor cell incorporation into tumors or ischemic tissues depends on stimuli from the target tissue and the ability of the resident bone marrow population to respond [10, 11]. Ferrari et al. showed that there was increased incorporation of intravenously administered EPCs into the tumor neovasculature of mice after sublethal irradiation compared with animals that were not irradiated [22]. Others have shown the successful inclusion of AC133 cells in different disease models [19, 23], usually documented by immunochemistry or detection of a transfected reporter gene (i.e., green fluorescent protein) in targeted tissue. Although both in vivo bioluminescent and fluorescent imaging have high sensitivity to detect the migration of relatively few labeled cells, these images are acquired at low spatial resolution.

MRI is a noninvasive imaging technique with high spatial resolution, providing the delineation of morphology and pathology by adjusting the scan acquisition. At present, the degree of angiogenesis in a tumor is inferred from dynamic contrast-enhanced MRI in which the timing and signal intensity changes from first pass and washout of an MRI contrast agent are used as determinants in a model to estimate vascular permeability, density, and blood volume of the tumor [2426]. These estimates of tumor perfusion and vessel density cannot be used to determine whether EPCs are contributing to ongoing vasculogenesis in a growing tumor. Recent studies demonstrate in vivo MRI monitoring of the temporal spatial migration of stem cells and other cells labeled with ferumoxides complexed to various transfection agents [20, 21, 2731]. The straightforward method of magnetically labeling stem cells does not alter cell metabolism, function, proliferation, viability, or differentiation capacity and is not associated with short- or long-term toxicity [20]. FePro-labeled AC133 cells appear hypointense (dark) on MRI due to significant shortening of the T2 and T2* relaxation times by the iron oxide nanoparticles incorporated within the endosomes. Anderson et al. have shown that mouse bone marrow-derived Sca1+ cells labeled with ferumoxide–poly-l-lysine migrate and incorporate into the neovasculature of implanted RT2 glioma in brains of mice, based on a hypointense ring observed on T2- and T2*-weighted images surrounding the tumor [21]. The hypointense ring on MRI correlated to Prussian blue-positive cells and cells positive for endothelial markers at the leading edge of the tumor. In the current study, comparison of the MRI with histopathology demonstrated the infiltration and incorporation of FePro-labeled human AC133 cells into ongoing angiogenesis at the periphery of established growing tumors. The expression of endothelial markers (human KDR1, CD31, and vWF) in the iron-labeled cells suggests that the maturation of the EPCs toward endothelial cell phenotype was not interfered with by labeling with FePro. These double-labeled AC133+ cells incorporate into the neovasculature of the tumors as demonstrated by confocal microscopy (Fig. 5). It is important to note that at this time MRI cannot determine whether the FePro-labeled cells have differentiated into a specific cell phenotype, but cell function may be inferred from clinical course or from complementary imaging studies such as positron emission tomography, optical imaging, or possibly other biomarkers.

Our findings are consistent with other reports using confocal and fluorescent microscopy that demonstrated the migration and incorporation of up to 40% of transplanted EPCs into tumor neovasculature [32]. Recently, Hilbe et al. reported that AC133+ cells comprise 0%–50% of the neovasculature in patients with lung cancer, and 43 of 63 patients studied showed a higher number of AC133+ cells in the tumor compared with that of the control [19]. The homing of EPCs into neovasculature of tumors or ischemic tissue depends on the expression of local cytokines such as VEGF from the tissue in response to hypoxia and has been related to tumor size and distance from existing vessels [33]. The current study demonstrates that after injection, magnetically labeled EPCs could be detected by MRI in established tumors (i.e., group 2 mice) within 3–5 days, and in concurrently implanted tumors within 5–7 days when the tumor grew to 0.5 cm in diameter.

The goal of this study was to detect the migration of labeled cells into growing tumors; therefore, we did not quantify the total number of labeled cells within the implanted tumors. Once tumors were approximately 1 cm in size, there was no apparent difference on MRI or histopathology between group 1 and group 2 mice receiving labeled cells (i.e., the timing of labeled cell administration did not matter). At the early stages of tumor growth (0.5–1 cm), most of the labeled cells were found along the tumor margins and in between tumor and surrounding muscles or connective tissues. When tumors grew to 1.5 cm, the areas of low signal intensity detected on MRI as well as iron-positive cells on histology were no longer at the periphery of the tumor. This might be a result of the lack of availability of labeled AC133 cells for migration and homing from the bone marrow, lung liver, or spleen to the tumor due to apoptosis. Alternatively, the labeled cells may have undergone several cell divisions, thereby diluting the intracellular iron label to levels that could not be visualized on MRI or histopathology. In the larger tumors, there were fewer iron-positive cells and fewer cells with endothelial markers at the tumor periphery as compared with smaller tumors (data not shown). The purpose of this study was not to quantify the incorporated EPCs in the tumor. For quantification, a proper phantom study would be needed to normalize the tissue characteristics based on T2 or T2* maps on MRI. Investigations are under way to determine the optimal method for quantifying the number of labeled cells in tissues. Moreover, by creating and comparing T2 or T2* maps of the control and corresponding tumors with labeled cells, it might be possible to differentiate nonspecific susceptibility artifacts on MRI in tissue containing necrosis with hemorrhage from the areas with iron-labeled cells. Our preliminary laboratory data showed a significant difference in the slope of the maps created from labeled cells and a corresponding amount of free iron in phantoms (data not shown).

There has been concern that resident tissue macrophages would phagocytose apoptotic or dead FePro-labeled cells within tissues or the tumor, and thereby confound the interpretation of the decrease in signal intensity observed on MRI after administration of iron-labeled cells. We did not observed any uptake of FePro-labeled cells by mouse host macrophages. We assume that few, if any, injected iron-labeled dead cells after an i.v. injection would be expected to incorporate into the tumor, because these cells would be cleared by the liver and spleen. However, some FePro-labeled cells detected at the periphery of the tumors expressed CD68 human macrophage markers by immunohistochemisty. These iron-positive macrophages would contribute to the hypointense regions on MRI and could not be distinguished from FePro-labeled cells contributing to tumor angiogenesis. Increased metabolic activity of macrophages compared with stem cells would cause the iron oxide nanoparticles within endosomes to dissolve faster into elemental or chelated iron and therefore not contribute to the hypointense regions on MRI (data not shown).


We used two FDA-approved agents, ferumoxides and protamine sulfate, to label human AC133+ EPCs and have been able to demonstrate the incorporation of these cells into the neovasculature of implanted flank tumors. We noted differences in the MRI to detect cells from early to late stage of tumor growth as well as in mice in which labeled cells were infused simultaneously during implantation or after tumors were established. As the tumors progressed with time, the MRI differences resulting from the timing of the injection of labeled cells diminished; however, the MRI data suggest that there was an early incorporation of EPCs into tumor neovasculature in mice when tumors were approximately 0.2 cm in size. Magnetically labeled cells were detected earlier in tumor using microscopy and ex vivo MRI but were not clearly delineated on in vivo MRI, consistent with the partial volume effects and difference in image resolution. MRI tracking of FePro-labeled EPCs should allow the translation of this approach from bench-to-bedside for the detection of sites of active neovascularization in tumors, metastasis, or ischemic lesions and may facilitate the development of novel treatment strategies. Detecting the presence of magnetically labeled EPCs in growing tumors is an important first step to demonstrate that sufficient numbers of cells migrate and home to sites of active tumor growth and would potentially allow the development of genetically engineered EPCs that could be used in conjunction with a suicide gene to target tumor vasculature [22]. FePro labeling of EPCs and their detection by MRI may allow investigators to find the optimal quantity of stem cells to be infused for treatment of the tumor.


The authors indicate no potential conflicts of interest.

Figure Figure 1..

MRI of flank tumors at different stages of development. In vivo (A–D), ex vivo (E–G), and DAB-enhanced Prussian blue (counterstained with nuclear fast red) staining (H) of the section corresponding to (G) are shown. Labeled (A–C, E–G) or unlabeled (D) human AC133+ cells were injected intravenously at the time of tumor implantation in immunodeficient mice (group 1). In vivo MRI of tumor at 0.5 cm in size does not show any definite low signal intensity of labeled AC133+ cell migration and incorporation (A, ring encompass implanted tumor and adjacent muscle). However, multiple low signal intensity areas at tumor size 1–1.5 cm (B, C) are seen, and the areas of low signal intensity ([B, C], dotted rings) are definitely different from the low signal intensity seen in tumor with control cells ([D], dotted ring). On the other hand, ex vivo images show areas of low signal intensity in all tumors (E–G), indicated by arrows (E, F), and areas of linear low signal intensity on (G). Histology (×1 magnification) with Prussian blue staining (H) showed iron-positive cells at the corresponding site of areas of linear low signal intensity on (G). Whitish areas in the histology section represent necrotic areas. Abbreviations: DAB, 3,3′-diaminobenzidine tetrahydrochloride; MRI, magnetic resonance imaging.

Figure Figure 2..

In vivo (A–C) and ex vivo (D–F) MRI of flank tumors at different stages of development and corresponding histology with Prussian blue staining (G–I) are shown. Labeled human AC133+ cells were injected at least 3 days after tumor implantation into immunodeficient mice (group 2). In vivo MRI of tumor at 9 × 6 mm (B) and 13 × 9.5 mm (C) shows low signal intensities (arrows) at the margin of or inside the tumors, and the low signal intensity is different from that of control tumor (A) at 9 × 6 mm. All tumors show areas of low signal intensity on ex vivo MRI ([E, F], arrows). Prussian blue staining (DAB-enhanced) shows iron-positive cells at the periphery of the tumors. Scale bars = 1000 μm (G), 200 μm (I). Magnification, ×2 (E). Arrows indicate iron-positive cells (H). Abbreviations: DAB, 3,3′-diaminobenzidine tetrahydrochloride; MRI, magnetic resonance imaging.

Figure Figure 3..

Immunohistochemistry on formalin-fixed paraffin-embedded sections (A–D) and frozen sections (E–H) are shown, indicating transformation of injected labeled AC133+ cells into endothelial cells (CD31, KDR1, vWF, and CD106 markers positive). (A–D) represent consecutive slides. Prussian blue shows iron-labeled cells (D) at sites corresponding to the sites with endothelial cells (A–C). (E): Consecutive slide of (F–H) stained with Prussian blue shows iron-labeled cells at sites corresponding to the sites with endothelial cells. (A–D) are from one animal, and (E–H) are from another animal. Arrows indicate CD31 (A), KDR1 (B), and vWF (C), positive cells. Abbreviations: PB, Prussian blue; vWF, von Willebrand Factor.

Figure Figure 4..

The relationship of host macrophage (mouse CD68-positive cells) to iron-positive cells (Prussian blue [PB] staining). (A): Host macrophages (arrows) are observed at the margin of the tumor, where signs of tumor necrosis are seen. (B): However, the area of macrophage accumulation does not correspond to the site of incorporated iron-positive cells (arrows). No iron-positive cells are seen at the site of macrophage migration (B), and no host CD68-positive macrophages (C) are seen at the site of iron-positive cells (D). Slides are from different animals (top vs. bottom panel).

Figure Figure 5..

The incorporation of intravenously injected human AC133+ cells into the vasculature of the implanted tumor. Vessels are delineated by rhodamine-labeled lectin (red), and the quantum dot–labeled endothelial cells show green fluorescent. (A–C): Confocal microscopic images. (D–F): Fluorescence photomicrography. (A, D): Tumor vessels delineated by rhodamine-labeled lectin. (B, E): Quantum dot-positive cells (green). (C, F): Superimposed images. Prussian blue staining of the same slide shows iron-positive cells at sites corresponding to the sites of quantum dot–positive cells (G).


We acknowledge Charles Carter, Jr. and Vicki Fellows of the Cell Processing Section, Department of Transfusion Medicine, National Institutes of Health Clinical Center, for help in obtaining the CD34 cells. This research was supported by the Intramural Research Program of the Clinical Center at the National Institutes of Health.