Tracking Transplanted Stem Cell Migration Using Bifunctional, Contrast Agent-Enhanced, Magnetic Resonance Imaging

NeuroImage 17, 803– 811 (2002) doi:10.1006/nimg.2002.1194

Tracking Transplanted Stem Cell Migration Using Bifunctional, Contrast Agent-Enhanced, Magnetic Resonance Imaging Michel Modo,* ,† ,1 Diana Cash,† Karen Mellodew,‡ Steven C. R. Williams,† Scott E. Fraser,§ Thomas J. Meade,§ Jack Price,‡ and Helen Hodges* ,¶ *Department of Psychology, †Neuroimaging Research Group, ‡Department of Neurosciences, Institute of Psychiatry, De Crespigny Park, London SE5 8AF, United Kingdom; §Beckman Institute, California Institute of Technology, Pasadena, California; and ¶ReNeuron Ltd., Guilford GU2 7AF, United Kingdom Received January 17, 2002

The ability to track stem cell transplants in the brain by in vivo neuroimaging will undoubtedly aid our understanding of how these cells mediate functional recovery after neural transplantation. One major challenge for the development and refinement of stem cell transplantation is to map the spatial distribution and rate of migration in situ. Here we report a method for tracking transplanted stem cells in the ischemia-damaged rat hippocampus by magnetic resonance imaging (MRI). Before transplantation, stem cells were labeled in vitro either with a novel bifunctional contrast agent, gadolinium rhodamine dextran (GRID), identifiable by both MRI and fluorescence microscopy, or with PKH26, visible exclusively under fluorescence microscopy. At different time points following engraftment, the brains were evaluated by both histology and ex vivo MR imaging. Transplanted stem cells were identified by MRI only if prelabeled with GRID, whereas fluorescence microscopy detected transplanted cells using either label. The distribution of GRID-labeled stem cells identified by MRI corresponded to those detected using fluorescence microscopy. These results demonstrate that GRID-enhanced MRI can reliably identify transplanted stem cells and their migration in the brain. © 2002 Elsevier Science (USA) Key Words: neural transplantation; magnetic resonance imaging; gadolinium rhodamine dextran; cerebral ischemia; rat; brain repair; molecular imaging.

INTRODUCTION The clinical prospects of neural transplantation to treat brain damage are greatly dependent on the thorough preclinical investigation of functional and ana1 To whom correspondence should be addressed at Institute of Psychiatry (http://www.iop.kcl.ac.uk), De Crespigny Park, London SE5 8AF, United Kingdom. Fax: ⫹44 (0) 207 848 0055. E-mail: [email protected].

tomical recovery. In turn, animal models of brain damage enhance our understanding of several factors involved in successful transplantation while also allowing the technological development necessary to monitor the survival of these grafts. Furthermore, monitoring of grafts noninvasively is an important aspect of the ongoing safety assessment of the procedure. Animal models of global ischemic brain damage that produce selective and well-defined lesions provide the opportunity to investigate the migration and integration of stem cells within a well-characterized environment. In rats, global ischemic brain damage resulting from a transient loss of blood supply to the brain (Pulisnelli and Brierly, 1979) produces a selective depletion of pyramidal cells in the CA1 field of the hippocampus and causes impairments of spatial memory (Hodges et al., 1997) akin to the cognitive deficits seen in patients recovering from cardiac arrest (Petito et al., 1987; Squire, 1992). Such impairments in spatial memory are partially restored after transplantation into the hippocampus of either fetal tissue or conditionally immortalized stem cells following global ischemia in rats (Netto et al., 1993; Sinden et al., 1997). These temperature-sensitive conditionally immortalized stem cells (from the Maudsley hippocampal cell line clone 36, MHP36) proliferate at the permissive temperature of 33°C to provide a large number of undifferentiated stem cells that can be transplanted to repair brain damage (Price et al., 2001). At the nonpermissive brain temperature, MHP36 cells stop proliferating and a large number (⬎70%) differentiate into either neurons or astrocytes, with approximately 25% of cells remaining undifferentiated in a quiescent state (Sinden et al., 1997). The extensive migration of stem cells away from the injection site, their ability to intermingle with host cells and integrate seamlessly into the host parenchyma, as well as the practicality of being grown in vitro for transplantation, are some of the advantages of conditionally immortalized stem cells (Gray et al.,

803 ©

1053-8119/02 $35.00 2002 Elsevier Science (USA) All rights reserved.

804

MODO ET AL.

2000). However, because they migrate and integrate so successfully into the host parenchyma, it is very difficult to differentiate host and grafted cells by magnetic resonance imaging (MRI). This differs from fetal grafts, which can be recognized by either the macroscopic appearance as a separate mass or by contrast enhancement using relaxation agents that traverse the compromised blood– brain barrier of the graft mass (Norman et al., 1989). To follow transplanted stem cells by MRI, therefore, requires a labelling method that will make grafted cells distinguishable from host cells. Moreover, the contrast produced by the label must be sufficient to permit detection of small cell clusters using high-resolution MRI. Previously, Jacobs and Fraser (1994) described a method using microscopic resolution MRI to detect embryonic cells prelabeled with a contrast agent (dextranlinked gadolinium) at different phases of frog embryonic development. We have further modified this reagent, incorporating a second functional group, the red fluorescent dye rhodamine, to create a bifunctional reagent, GRID, detectable by both MRI (via gadolinium) and fluorescence microscopy (via rhodamine) (Hu¨ ber et al., 1998). Stem cells loaded with this compound prior to transplantation therefore become visible using either imaging technique. Here we report the application of this method to track the migration of transplanted stem cells in the global ischemia-damaged brain. MATERIALS AND METHODS In Vitro Experiments GRID labeling. The contrast agent, gadolinium rhodamine dextran (GRID), consists of a dextran polymer backbone (⬃10 kDa) amino-modified (CH 2) 4™NH 2 with covalently attached Gd 3⫹ chelates and tetramethylrhodamine (total molecular weight ⬃16.6 kD). The chelate is a derivative of diethylenetriaminepentaactic acid (DTPA) and renders the Gd 3⫹ ion inert (Gd-DTPA). There are between 9 and 12 Gd 3⫹ chelates per dextran (Hu¨ ber et al., 1998). The paramagnetic properties of Gd 3⫹ ions in GRID (T 1 molar relaxivity, 17 mM s ⫺1) decrease the local T 1 and T 2 of water protons to produce additional contrast in the acquired MR image. Incorporating Gd 3⫹ into cells for transplantation will therefore change their relaxation characteristics, leading to a differentiation between labeled and nonlabeled cells. GRID was therefore used to label cells in vitro before intracerebral implantation. GRID was diluted (1:1) with distilled water at least 24 h before being used for cell labeling. Frozen MHP36 cells (passage 42) were thawed and cultured for at least one passage before being used for cell labeling according to a standard protocol (Sinden et al., 1997). During labeling, MHP36 cells were grown under proliferative

conditions with GRID added at a final concentration of 45 ␮M Gd 3⫹ and 2.73 ␮M tetramethylrhodamine per milliliter of medium. After 6 h of incubation with GRID, the medium was discarded and cells were removed from the flask by adding Hanks’ balanced salt solution (HBSS, Gibco, UK) without Ca 2⫹ and Mg 2⫹. The suspension was centrifuged (1500 rpm for 5 min) and cells were either seeded in new medium for in vitro experiments or resuspended for transplantation (25,000 cells/␮l) with 1 mM N-acetyl-L-cysteine (NAC, Sigma) in HBSS (Gibco, UK). Cell viability for transplantation, as determined by trypan blue (Sigma, UK) exclusion, was ⬎90% after GRID labeling. At the end of grafting cell viability was 85% as determined from the same vial used for transplantation by trypan blue exclusion. To investigate how proliferation of MHP36 cells affects the detection of GRID-labeled cells, MHP36 cells labeled with GRID were cultured under proliferative (33°C with bFGF) or nonproliferative (37°C without bFGF) conditions. After 7 days, the medium was discarded and cells were fixed with 4% paraformaldehyde (in 0.2 M phosphate-buffered saline, PBS) for 15 min before being washed three times with HBSS. Fixed cultures were counterstained with blue fluorescent Hoechst (Sigma, UK) at a 1:1000 dilution (in PBS) for 1 min and washed three times with HBSS. A coculture assay using primary cells from rat E15 cerebral cortex was used to determine if GRID leaks from the transplanted cells. For this experiment, MHP36 cells were labeled with GRID and primary cells were labeled with the green fluorescent dye PKH67 (Sigma, UK). PKH67 labels dissociated rat embryonic cortex without any signs of leakage (K. Mellodew, unpublished observations). Cells were grown in medium consisting of 10% fetal calf serum (FCS, Gibco, UK) in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, UK) at 37°C (Williams and Price, 1995). If leakage and reuptake of GRID occurred, some cells would show both red and green fluorescence. Cells were fixed after 7 days of coculturing. PKH26 labeling. To provide a control for the contribution of GRID to detect transplanted cells by MRI, a non-contrast-enhancing red fluorescent dye, PHK26 (Sigma, UK), was used to prelabel MHP36 cells and to identify transplanted cells histologically. The cells were prelabeled by incubation for 4 min with PKH26, an aliphathic reporter molecule that integrates into the cell membrane by selective partitioning. Hass et al., (2000) have shown the stability and reliability of this marker for neural transplantation. After PKH26 labeling, cells were suspended (25,000 cells/␮l) with NAC in HBSS. Viability of cells was determined by trypan blue exclusion, ranging from 83 to 92% before grafting and 73 to 85% after grafting.

MR TRACKING OF TRANSPLANTED STEM CELLS

805

Animal Experiments On arrival all 25 Wistar rats (Charles River, UK) weighed between 230 and 245 g. Animals were acclimated and handled for at least 1 week prior to surgery. Before surgery food and water were available ad libitum. Cells were grafted into a stable nonprogressive lesion, 2–3 months after global ischemia with either GRID (n ⫽ 12)- or PKH26 (n ⫽ 12)-labeled cells. A control group (n ⫽ 3) consisted of global ischemic animals with a vehicle (NAC) infusion. Perfusion time points for MRI were 1, 3, 7, and 14 days postgrafting. All procedures were in accordance with the UK Animals (Scientific) Procedures Act of 1986 and the ethical review process of the Institute of Psychiatry, University of London. Global ischemia. To induce global ischemia, the four-vessel occlusion (4VO) method of Pulsinelli and Brierly (1979) was used. Under halothane anesthesia (4% induction, 2% maintenance in 70% nitrous oxide and 30% oxygen) the vertebral arteries were electrocoagulated through the alar foraminae on the first cervical vertebra and the common carotids were brought to the surface. Elastic ties were inserted around both carotids. The carotid arteries were occluded 24 h later. Occlusion of the carotid arteries was achieved in awake animals by tightening of the snares for 15 min. Global ischemia, i.e., reduction of 95% of blood flow to the brain, was indicated by loss of the righting reflex within 2 min of occlusion. Head and rectal temperatures were monitored and maintained (rectal temperature at 37 ⫾ 0.5°C, head temperature at 36 ⫾ 0.5°C). Grafting. Anesthesia was induced by intramuscular (im) injection of Hypnovel (midazolam 0.01– 0.03 ml per 100 g body wt, Roche, UK) and Immobilon (etorphine hydrochloride, 0.04 mg/ml, and methotrimeprazine 18 mg/ml; 0.01 ml per 100 g body wt, C-Vet Veterinary Products, UK). Animals were placed in a stereotaxic frame and an incision was made along the midline before the skin was retracted to expose the skull. Two deposits of 2 ␮l of cell suspension were implanted in the left hemisphere (Site 1: AP ⫽ ⫺3.3, L ⫽ ⫺1.3, V ⫽ ⫺2.8; Site 2: AP ⫽ ⫺4.2, L ⫽ ⫺3.4, V ⫽ ⫺3.1). The suspensions were delivered at a speed of 1 ␮l/min. The syringe was left in place for 2 min after each deposit to allow the cells to disperse. For sham grafting an equal volume of NAC was injected. A total of ⬃100,000 cells per animal were implanted. After surgery, animals were injected subcutaneously with the immunosuppressant cyclosporin A (CsA, Sandimmun, 10 mg per 100 g body wt, Sandoz, Basel, CH) in Cremophor EL (1:3, Sigma, UK), and Revivon (diprenorphine 0.272 mg/ml; 0.01 mg per 100 g body wt, C-Vet Veterinary Products, UK) was injected intramuscularly to reverse anesthesia. Immunosuppression was given on alternate days until perfusion.

FIG. 1. GRID-labeled MHP36 cell in vitro. (A) More than 98% of MHP36 cells were labeled with GRID (in red) and GRID fluorescence was limited to the cells in vitro (Hoechst staining in blue). After cells were labeled and washed no GRID was observed outside of the cells. (B) After 7 days of coculturing of E15 cerebral cortex primaries labeled with PKH67 (green fluorescent) and MHP36 cells labeled with GRID (red fluorescent) no cell containing both fluorescent markers was detected, suggesting that there was no leakage or reuptake of either marker. (C) GRID-labeled cells grown in the proliferative condition at 33°C show a diluting out of the label due to cell division and, after 7 days in culture, no longer allow cell detection based on GRID fluorescence. (D) MHP36 cells labelled with GRID grown in the nonproliferative condition at 37°C for 7 days still allow detection of cells based on GRID fluorescence despite three or four further cell divisions. Bar ⫽ approximately 40 ␮m.

Perfusion. For perfusion, animals were overdosed with Pentoject (pentobarbitone sodium, Animal Care Ltd., UK). The descending aorta was clamped to restrict blood flow to the upper body half. Rats were prewashed transcardially with heparinized saline (0.9% NaCl in distilled H 2O containing 5000 units of heparin, CP Pharmaceuticals, UK, per liter) followed by 4% paraformaldehyde (PFA) in 0.2 M phosphatebuffered saline (PBS). Magnetic resonance imaging. Fixed heads were stored and scanned in 4% paraformaldehyde in 0.2 M PBS. All MRI was performed on a small animal, horizontal-bore, 4.7-T NMR system (Oxford Systems, UK) controlled by a UNITY Inova-200 imaging console (Varian, USA). A quadrature birdcage radiofrequency coil with 63-mm internal diameter (Varian, USA) was used for signal transmission and reception. The raw data were Fourier transformed using in-house developed software. The choice of the imaging protocol was determined based on a need for a high spatial resolution as well as the distinction between the host and the gadolinium-bearing graft. After an initial evaluation of different scanning parameters, we adopted an imaging protocol consisting of coronal T 1-weighted (TR ⫽ 500,

806

MODO ET AL.

TE ⫽ 15), T 2-weigted (TR ⫽ 4000, TE ⫽ 45), and proton density-weighted (TR ⫽ 4000, TE ⫽ 15) scans with a 256 ⫻ 256 matrix size, 4 averages per phase encoding step, 20 contiguous 500-␮m-thick slices, and a field of view of 2.5 ⫻ 2.5 cm. This protocol afforded an acquisition of images with an in-plane resolution of 124 ⫻ 124 ␮m within 7 min for T 1-weighted images and 35 min for T 2- and PD-weighted images. Immunohistochemistry. After scanning was completed, the brains were removed and placed into 30% sucrose for at least 1 week prior to cutting. Sections of 50-␮m thickness were cut on a cryostat microtome (Leica, Germany) at ⫺24°C and collected free-floating in sucrose. For immunohistochemical staining, sections were washed 3 ⫻ 5 min in 0.1 M PBS prior to application of the primary antibody used for identification of neurons (monoclonal mouse anti-NeuN, 1:1000, Chemicon, UK) or astrocytes (polyclonal goat anti-GFAP, 1:4000, Chemicon, UK). Primary antibodies were applied overnight at 4°C. Sections were washed 3 ⫻ 5 min in PBS and then incubated with a cross-adsorbed secondary antibody (goat anti-mouse Alexa488 for NeuN, and rabbit anti-goat Alexa488 for GFAP, 1:500, Molecular Probes, UK) for 45 min at room temperature. Sections were washed 3 ⫻ 5 min in PBS before coverslips were applied with Vectashield for immunofluorescence (Vector, UK). RESULTS In Vitro Labeling with GRID To identify transplanted stem cells in the brain by MRI, grafted cells need to produce a characteristic MR signal that differs from that of the host cells. Incubation of MHP36 cells with the bifunctional contrast agent GRID for 6 h resulted in incorporation of the marker into ⬎98% of the cells (as determined by cell counting under the fluorescence microscope), producing clearly visible fluorescence signals (Fig. 1A). However, after 7 days of culture under conditions promoting proliferation of the cells (33°C with bFGF), GRID fluorescence had become so weak as to render the cells undetectable (Fig. 1C). The loss of GRID fluorescence can be explained by progressive dilution of the signal as the cells divide. After the same period of culturing under non-proliferative conditions (37°C without bFGF), GRID-labeled cells remained clearly visible under fluorescence microscopy (Fig. 1D). To determine if GRID would leak out of the labeled MHP36 cells, GRID-labeled MHP36 cells (red fluorescent) were cocultured with primary E15 cerebral cortex cells labeled with the green fluorescent transmembrane dye PKH67. Occurrence of cells containing both fluorescent markers would indicate leakage and reuptake of one or both agents. After 7 days of coculture, under a non-proliferative condition for the MHP36

cells, we did not detect any cells incorporating both labels, confirming that there was no leakage and reuptake of GRID in vitro at this time point (Fig. 1B). MRI Detection of Grafted Cells MHP36 cells prelabeled with GRID were transplanted into rats with global ischemic brain damage just above the alveus of the hippocampus. GRID-labeled transplants clearly demarcated the injection tract on T 1-, T 2-, and PD-weighted MR images (Fig. 2). The injection tract was located at the stereotaxic coordinates used for transplantation, reaching from the dura through the cerebral cortex to the hippocampus. The high density of labeled cells within the injection site resulted in a strong signal loss and dark appearance of the injection tract in all scans. Surrounding the injection tract, in the T 1-weighted scans, a hyperintense border was apparent around the dark transplant region. This hyperintensity is presumably due to the weaker T 1-shortening effect of the small number of GRID-bearing cells that migrated out from the injection tract. These hyperintense borders were not detected in either the PD- or T 2-weighted scans, as these imaging protocols are not sufficiently sensitive to detect such low concentrations of the contrast agent. On the other hand, the anatomical resolution and the distinction between grafted and host tissue was superior in T 2-weighted scans compared with T 1- and PDweighted scans. As graft survival varied between animals (as shown by histology), the contrast revealed by MRI was more marked in rats with greater graft survival. Little difference in the appearance of the grafts was evident between 1 and 7 days postgrafting. The GRIDlabeled transplants were detected on T 2- and T 1weighted images up to 7 days postgrafting as a delineated injection tract with a band of additional cells along the corpus callosum and the CA1 field (Fig. 3). At 14 days, most cells had migrated away from the injection tract (as indicated by histology) and transplants could no longer be reliably detected in the region of the injection tract itself using current imaging parameters. However, failure to detect cells by MRI at 14 days is also likely to be due to ex vivo scanning of brains in paraformaldehyde which yields a higher image resolution, but reduces exchange of water with the contrast agent and possibly attenuates the contrast-enhancing effects of small amounts of GRID. Minor differences in graft location (e.g., injection tract closer to the midline than expected from the stereotaxic coordinates) between grafted animals could be explained by slight differences in graft placement due to interanimal variability (e.g., differences in skull size), but this did not affect the appearance of the grafts. To determine the contribution of GRID to the detection of transplants, animals were grafted with PKH26-

MR TRACKING OF TRANSPLANTED STEM CELLS

807

FIG. 2. MR images of GRID-labeled transplants 7 days postgrafting. GRID-labeled transplants were visible in T 1-, proton density (PD)-, and T 2-weighted scans. On the T 1-weighted scans a region of signal loss (densely packed injection tract produced a dark signal on all images) is surrounded by a hyperintense area attributed to the dispersion of grafted cells (clearly visualized in the first and second images).

labeled cells. In scans from rats with PKH26-labeled cells, grafts were not visible (Fig. 3) except for the small indentations (“dips”) at the site of injection. However, in one animal, perfused 1 day postgrafting, a small deposit in the hippocampus was visible. This change in signal apparently was due to the extremely densely packed deposit of transplanted cells which differed markedly in consistency from the surrounding tissue. Otherwise, there were no discernible differences in the MRI scans over time in this group. The

MRI signal change detected in animals with GRIDlabeled grafts can therefore be attributed to the prelabeling of MHP36 cells with this bifunctional contrast agent. Detection of Grafted Cells by Histology The bifunctional nature of the GRID label made it possible to corroborate MRI visualization of transplants using histological methods. Sectioned brains

FIG. 3. MR (T 2-weighted) images of the time course of GRID- and PKH26-labeled MHP36 cells. GRID labeling allowed detection of the transplant by MRI (arrows), whereas PKH26-labeled cells did not produce a contrast detectable by MRI. However, in the brains with PKH26-labeled transplants, a dip in the cortex at the injection site was detectable (arrows). By 14 days after transplantation the dispersion of GRID-labeled cells away from the injection tract no longer allowed reliable detection of grafts at the spatial resolution of the present experiment.

808

MODO ET AL.

FIG. 4. Histological correlates of GRID-labeled transplants. (A) Detection of GRID-labeled transplants by fluorescence microscopy. At a macroscopic level (a), the GRID-labeled transplant can be detected in the injection tract (b) and migrating across the corpus callosum (c). A higher magnification of cells in corpus callosum (d) shows that it is possible to identify single cells within the GRID-labeled transplant that are migrating along the corpus callosum. Bar ⫽ approximately 3.5 mm (a) 1 mm (b) 300 ␮m (c), and 50 ␮m (d). (B) As GRID is detectable indirectly by MRI (T 1-weighted image used here) and directly by fluorescence microscopy, a direct overlay of both images (after multimodal image coregistration) affords identification and corroboration of the transplant by both techniques. I, injection tract; cc, corpus callosum, lv, lateral ventricle.

from rats used in earlier MRI scanning were examined under fluorescence microscopy to reveal the rhodamine functionality of GRID. At low magnification, GRIDlabeled transplants revealed the distribution and survival of transplanted cells (Fig. 4A). An overlay image consolidating images produced in fluorescence microscopy and MRI is shown in Fig. 4B. This revealed an excellent match, indicating that both techniques (fluorescence microscopy and MRI) can detect the location and migration of transplanted cells. Transplanted cells close to the injection tract were densely packed with GRID (Fig. 4A) and consequently produced the signal loss apparent on the MR images. Histological detection of PKH26-labeled cells showed a distribution similar to that of GRID-labeled cells, but produced no coincident signal in the MR images. Transplanted GRID- and PKH26-labeled cells that colonized the hippocampus showed differentiation into

both astrocytes and neurons, indicating that labeling did not prevent these cells from acquiring a mature phenotype (Fig. 5). GRID labeling of MHP36 cells showed some dense fluorescent clustering of the label on the cell, whereas PKH26 labeling showed a more distributed pattern with less intense fluorescence. Only a few transplanted cells differentiated into neurons or astrocytes at the early time points (1 and 3 days postgrafting), but by 7 and 14 days after grafting a few GRID-labeled cells that “lined up” in the CA1 and CA2 fields in the hippocampus of the ischemia-damaged brain differentiated into neurons. DISCUSSION Stem cell grafting offers great promise to treat neurological disorders (Gray et al., 2000). To realize stem cell grafting as a treatment approach, a noninvasive

MR TRACKING OF TRANSPLANTED STEM CELLS

809

FIG. 5. Neuronal and astrocytic differentiation of GRID- and PKH26-labeled transplants. GRID-labeled cells showed an integration into the ischemia-lesioned hippocampus and integrated cells developed a neuronal phenotype as detected by colocalization with the neuronal marker NeuN (in green) and GRID (in red). Likewise, GRID-labeled cells differentiated into astrocytes as detected by GFAP (in green). GRID-labeled cells showed few punctate immunofluorescent clusters compared with the PKH26, which showed a more dispersed, punctate appearance.

810

MODO ET AL.

method for the in vivo assessment of graft survival is required to monitor patients. MRI methods are potentially well suited for this use. Due to their seamless integration into the host parenchyma and migration over long distances in the brain, stem cell grafts cannot be detected like fetal grafts based on their mass morphology. It is therefore important to develop neuroimaging technology that is able to distinguish between the host and the graft with a sufficiently high sensitivity and specificity to identify and map the fate of transplanted cells. Labeling of MHP36 cells with GRID allowed reliable detection of cells in vitro for up to 7 days and cocultures did not show any leakage of GRID from the MHP36 cells by 7 days in vitro. Only GRID-labeled transplants could be identified by MRI, whereas transplants labeled with PKH26 were not detected in MR images. GRID-labeled cells showed a marked difference in signal compared with the surrounding host brain tissue. T 2-weighted MR images provided the best distinction between host and graft while retaining good neuroanatomical resolution. However, the T 1-weighted scans achieved greater sensitivity as they allowed the detection of cells that dispersed out of the injection tract. In future studies, an assessment of the relaxation characteristics of the density and dispersion of GRID-labeled cells might therefore allow a quantification of graft survival. Proton density-weighted scans did not reveal any additional information. The contrast agent produces such a strong reduction of the spin–spin relaxation time (T 2) that even the shortest TE used in our studies (for the PD-weighted image) was still too long to afford any substantive MR signal in the final image. Therefore, low signal intensity was observed for each imaging method. Both T 1- and T 2-weighted high-resolution MR images may well be a useful combination to assess the graft survival and migration in vivo. A clear advantage of GRID over other contrast agents is the possibility of visualizing the compound by both MRI and fluorescence microscopy. This bifunctionality allows direct visualization of the cells from in vitro labeling by histology without the need for antibodies or colocalization of an antibody with the contrast agent. For instance, Bulte et al. (1999) corroborated their MRI detection of monocrystalline iron oxide nanoparticles (MION)-labeled oligodendrocyte progenitor grafts in the spinal cord by using the Prussian blue reaction against the iron in the cells, and more recently the same group validated their MRI detection of magnetodendrimer-labeled cells with X-gal staining (Bulte et al., 2001), whereas Franklin et al. (1999), who prelabeled transplants with supraparamagnetic iron oxide particles (SPIO), used electron-dense inclusions as visualized by electron microscopy to detect grafted cells histologically. In contrast, the bifunctionality of GRID allowed us to overlay histology and MRI without recourse to another method to visualize the transplanted

cells. Furthermore, the use of fluorescence microscopy allowed us to investigate whether the differentiation of MHP36 cells was altered by the use of a new labeling procedure. These results show that it is possible to detect and monitor grafted stem cells migrating in ischemia-lesioned brains. At present there is little understanding of the mechanisms involved in stem cell-mediated recovery. The use of MRI and related technologies will provide an important step toward understanding how anatomical changes (such as the presence of stem cells in regions of damage) relate to functional modulation and their associated recovery. We believe that the development of in vivo imaging technologies for monitoring stem cell survival will make an important contribution to the assessment of human patients after intracerebral transplantation. ACKNOWLEDGMENTS The authors thank Dr. Paul Kinchesch (Queen Mary and Westfield College) for his assistance with the NMR system and Peter Sowinski (ReNeuron Ltd.) for his help with execution of the global ischemia model. M.M. is supported by predoctoral scholarships from NATO, the Minste`re de l’Education Nationale (MEN) du Luxembourg and ReNeuron Ltd., and is currently supported by MRC Grant G0000966. D.C. is supported by MRC Grant G78/6070. The imaging spectrometer was provided by the University of London Intercollegiate Research Service scheme and is located at QMW. The authors also thank Professor Jeffrey Gray (IoP) for his continued encouragement in the development of novel imaging techniques to assess the viability of neural transplants.

REFERENCES Bulte, J. W. M., Zhang, S.-C., van Gelderen, P., Herynek, V., Jordan, E. K., Duncan, I. D., and Frank, J. A. 1999. Neurotransplantation of magnetically labeled oligodendrocyte progenitors: Magnetic resonance tracking of cell migration and myelination. Proc. Nat. Acad. Sci. USA 96: 15256 –15261. Bulte, J. W. M., Douglas, T., Witwer, B., Zhang, S.-C., Strable, E., Lewis, B. K., Zywicke, H., Miller, B., van Gelderen, P., Moskowitz, B. M., Duncan, I. D., and Frank, J. A. 2001. Magnetodendrimers allow endosomal magnetic labeling and in vivo tracking of stem cells. Nat. Biotechnol. 19: 1141–1147. Franklin, R. J. M., Blaschuk, K. L., Bearchell, M. C., Prestoz, L. L. C., Setzu, A., Brindle, K. M., and ffrench-Constant, C. 1999. Magnetic resonance imaging of transplanted oligodendrocyte precursors in the rat brain. NeuroReport 10: 3961–3965. Gray, J. A., Grigoryan, G., Virley, D., Patel, S., Sinden, J. D., and Hodges, H. 2000. Conditionally immortalized, multipotential and multifunctional neural stem cell lines as an approach to clinical transplantation. Cell Transplant. 9: 153–168. Haas, S. J. P., Bauer, P., Rolfs, A., and Wree, A. 2000. Immunocytochemical characterization of in vitro PKH26-labelled intracerebrally transplanted neonatal cells. Acta Histochem. 102: 273–280. Hodges, H., Nelson, A., Virley, D., Kershaw, T. R., and Sinden, J. D. 1997. Cognitive deficits induced by global cerebral ischaemia: Prospects for transplant therapy. Pharmacol. Biochem. Behav. 56: 763–780. Hu¨ ber, M. M., Staubli, A. B., Kustedjo, K., Gray, M. H. B., Shih, J., Fraser, S. E., Jacobs, R. E., and Meade, T. J. 1998. Fluorescently

MR TRACKING OF TRANSPLANTED STEM CELLS detectable magnetic resonance imaging agents. Bioconj. Chem. 9: 242–249. Jacobs, R. E., and Fraser, S. E. 1994. Magnetic resonance microscopy of embryonic cell lineages and movements. Science 263: 681– 684. Netto, C. A., Hodges, H., Sinden, J. D., Le Peillet, E., Kershaw, T., Sowinski, P., Meldrum, B. S., and Gray, J. A. 1993. Effects of foetal hippocampal field grafts on ischaemia induced deficits in spatial navigation in the water maze. Neuroscience 54: 69 – 92. Norman, A. B., Thomas, S. R., Pratt, R. G., Samaratunga, R. C., and Sanberg, R. P. 1989. A magnetic resonance imaging contrast agent differentiates between the vascular properties of fetal striatal tissue transplants and gliomas in rat brain in vivo. Brain Res. 503: 156 –159. Petito, C. K., Feldmann, E., Pulsinelli, W. A., and Plum, F. 1987. Delayed hippocampal damage in humans following cardiorespiratory arrest. Neurology 37: 1281–1286.

811

Price, J., Uwangho, D., Peters, S., Galloway, D., and Mellodew, K. 2001. Neurotransplantation in neurodegenerative disease: A survey of relevant issues in developmental neurobiology. In Neural Transplantation in Neurodegenerative Disease (J. A. Gray, Ed.), pp. 148 –165. Wiley, Chichester. Pulsinelli, W. A., and Brierly, J. B. 1979. A new model of bilateral hemisphere ischaemia in the unanaesthetised rat. Stroke 10: 267– 272. Sinden, J. D., Rashid-Doubell, F., Kershaw, T. R., Nelson, A., Chadwick, A., Jat, P. S., Noble, M. D., Hodges, H., and Gray, J. A. 1997. Recovery of spatial learning by grafts of a conditionally immortalized hippocampal neuroepithelial cell line into the ischaemia-lesioned hippocampus. Neuroscience 81: 599 – 608. Squire, L. R. 1992. Memory and the hippocampus: A synthesis from findings with rats, monkeys, and humans. Psychol. Rev. 99: 195–231. Williams, B. P., and Price, J. 1995. Evidence for multiple precursor cell types in the embryonic rat cerebral cortex. Neuron 14: 1181–1188.