Tracking Stem Cells for Cellular Therapy in Stroke

Abstract

Stem cell transplantation has emerged as a promising treatment strategy for stroke. The development of effective ways to monitor transplanted stem cells is essential to understand how stem cell transplantation enhances stroke recovery and ultimately will be an indispensable tool for advancing stem cell therapy to the clinic. In this review, we describe existing methods of tracking transplanted stem cells in vivo, including optical imaging, magnetic resonance imaging (MRI), and positron emission tomography (PET), with emphasis on the benefits and drawbacks of each imaging approach. Key considerations such as the potential impact of each tracking system on stem cell function, as well as its relative applicability to humans are discussed. Finally, we describe multi-modal imaging strategies as a more comprehensive method to track transplanted stem cells in the stroke-injured brain.

INTRODUCTION

Among the many neurological disorders and types of central nervous system (CNS) trauma, stroke remains one of the most devastating causes of morbidity and mortality. Over 795,000 Americans suffer a stroke each year, making it the third leading cause of death and the leading cause of severe, long-term disability. Not only is stroke injury extremely debilitating, but its protracted course and associated functional deficits make treatment very costly. With the elderly population being amongst the fastest growing demographics in the USA, the prevalence of CNS diseases such as stroke and the costs associated with caring for afflicted patients are likely to increase substantially in coming years.

The discovery of self-renewable neural stem cells (NSCs) and their progenitors in the adult brain brought hope that the CNS could be made to repair damage caused by injuries such as stroke. When cultured ex vivo with appropriate growth factors, NSCs can generate a large number of progeny and differentiate into the various CNS cell lineages, including neurons, glia, and oligodendrocytes. Although it was originally proposed that subsequent delivery of such NSCs to the stroke-injured CNS would allow restoration of the original cytoarchitecture and functional replacement of lost cells, more recent studies indicate that NSCs instead can enhance functional recovery post-stroke via alternative mechanisms, including NSC secretion of neurotrophic factors, immunomodulation, stimulation of endogenous neurogenesis, and neovascularization [1–4].

A critical aspect of understanding and improving NSC-based treatment strategies is the ability to monitor these cells in vivo, ideally via non-invasive imaging systems that allow long-term tracking of transplanted NSCs. By allowing longitudinal assessment without requiring sacrifice of the experimental animal, in vivo imaging techniques can enhance understanding of post-delivery cellular migration and the mechanisms by which stem cells provide their beneficial effects, both of which are still not well understood. Not only do in vivo imaging techniques facilitate progress in developing stem-cell-based therapeutics, but they also are relevant to future clinical therapeutic applications of stem cells. In vivo imaging would help physicians assess the viability and efficacy of transplanted cells. In addition, given the proliferative capacity of stem cells, their in vivo monitoring also would be important for identifying any associated tumorigenicity.

In order to develop successful in vivo imaging modalities for transplanted stem cells, several factors must be considered. An effective system must provide high spatial resolution and high sensitivity, ideally capable of distinguishing individual cells. Such technology also cannot impact cellular viability, motility, differentiation, physiology or functionality, and should not be retained in dead cells. Furthermore, virtually all stem cell imaging technologies require some degree of ex vivo alteration of the stem cells prior to transplantation, and thus efficient alteration of the entire transplant cell population also is a requirement. Finally since these techniques will be used to image cells within the brain, any potential probes must be capable of crossing the blood–brain barrier (BBB).

In this chapter, we review the application of non-invasive imaging methods that are applicable to tracking transplanted stem cells following stroke injury. We discuss the use of optical imaging, magnetic resonance imaging (MRI) and positron emission tomography (PET) as potential imaging strategies and compare the benefits and drawbacks of each approach.

OPTICAL IMAGING

Optical imaging encompasses a variety of cell imaging modalities including fluorescence imaging, quantum dots, and bioluminescence imaging (BLI). All of these methods involve prior ex vivo manipulation of the transplant cell population to introduce a fluorescent label that can allow subsequent detection of photons emitted either by chemical oxidative processes or by external excitation of a fluorophore [5]. In spite of their high sensitivity and superior signal-to-background ratio, all of these optical imaging modalities have poor spatial resolution and poor depth penetration due to high absorption and scatter.

In vivo fluorescence imaging utilizes organic fluorophores in the forms of fluorescent proteins like green fluorescent protein (GFP) that are genetically introduced into the stem cells; fluorescent dyes, such as DiD, DiI and FDA-approved ICG, which label the cell membrane [6]; or fluorochromes bound to various types of molecular probes, including targeting probes, cross-linking probes and enzyme-activatable probes [7,8]. Near-infrared fluorochromes have decreased absorption and enhanced depth penetration, and are thus the most promising [9,10]. However, fluorescent dyes and fluorochromes have a limited half-life and are diluted out of actively proliferating cells, making them sub-optimal for long-term tracking. Although genetic transduction with a fluorescent transgene can allow permanent expression of the corresponding fluorescent protein, this approach runs the risk of introducing unwanted mutations in the stem cell population during transgene incorporation.

Quantum dots are inorganic colloidal fluorescent semiconductor nanocrystals, and have several advantages over their organic fluorophore counterparts. Quantum dots, for example, have a broader absorption spectrum and a narrower emission spectrum, allowing multiple quantum dots with different emission energies to be imaged simultaneously with the same excitation energy [10]. Furthermore, quantum dots can be synthesized to the desired specifications, including size, shape and photon emission energy. They also are more stable, have enhanced quantum yields, decreased photobleaching and less scatter, enhancing their depth penetration [10]. Quantum dots also exhibit increased fluorescence resonant energy transfer, fluorescent ‘blinking’ and heightened fluorescent intensity, resulting in enhanced spatial resolution and sensitivity, with individual dots capable of being imaged in vitro [11–14]. Quantum dots can be solubilized and introduced directly into the transplant cell population or can be conjugated to various molecular probes that can bind specifically to other cellular molecules [15]. However, like fluorescent dyes and fluorochromes, cell labeling with quantum dots is not amenable to long-term tracking. Nonetheless, one study has demonstrated the use of quantum dots to label NSCs and progenitor cells by electroporation and ultrasound-guided biomicroscopy [16]. In vivo fluorescence imaging and quantum dot modalities are, however, in an early stage of use for stem cell tracking. Furthermore, the variable intracellular stability of quantum dots and potential for cellular toxicity necessitate further research assessing safety and efficacy of such techniques.

In contrast, BLI is the best studied of the various optical imaging modalities with regards to NSC imaging. Similar to fluorescence-based tracking, BLI requires the transplant cell population to first be transduced with a transgene encoding the firefly or Renilla luciferase (Luc) enzyme. Following transplantation of a Luc-modified stem cell population, systemic injection of the luciferase substrates, D-luciferin or coalenterazine, results in the emission of photons that can be visualized and quantified wherever Luc-expressing cells are present [17]. The simplicity of this approach, combined with the accuracy and ability to quantify signal, makes BLI the most utilized imaging tool in pre-clinical transplant models [18]. BLI has been used to track survival, migration, immunogenicity and tumorigenicity of transplanted cells in animal models of stroke [19–21], spinal cord injury [22] and intracranial brain tumor [23], which can then be detected and quantified by a charge-coupled device camera system. Drawbacks to this approach include low spatial resolution and signal dampening by overlying tissue [24], which is of particular concern for imaging deep tissue structures such as the brain.

Nonetheless, BLI has been shown to be an effective method to study in vivo stem cell migration and viability in stroke animal models. One study longitudinally followed C17.2 NSC transhemispheric migration along the corpus callosum towards a stroke lesion in rodent stroke models, with the migration pattern subsequently confirmed by histology [19]. Crucially, such migration was not observed in non-stroke controls. Furthermore, histology also confirmed site-specific differentiation with neuronal-type phenotypes and both neuronal (NeuN) and glial cell (GFAP) markers were observed [19].

In a second study, BLI coupled with histology was used to assess NSC viability in vivo, and demonstrated that NSCs had enhanced survival in the stroke microenvironment, and that this may be due to some pro-regenerative effect of inflammation. This study also provided evidence against the notion of the brain being an ‘immunoprivileged’ organ, since transplanted NSCs were found to survive better in immunosuppressed nude mice versus immunocompetent C57 or CD-1 mice [20].

In a study by our group, BLI was used to track the fate of grafted human embryonic stem-cell-derived NSCs (hNSCs) in the stroke-damaged rat brain [21]. The hNSCs were genetically-modified to express a double-fusion (DF) reporter gene encoding GFP and Luc. DF-modified hNSCs were grafted into the ischemic hemisphere of rats subject to the middle cerebral artery occlusion (MCAO) stroke model and monitored in real-time for 2 months. Longitudinal quantitative analysis revealed no significant changes in luciferase activity, demonstrating the nontumorigenic property of the hNSCs (Fig. 1). In addition, the DF transgene did not alter the physiological properties of the cells. Grafted hNSCs migrated to the stroke-damaged area and differentiated into neurons, astrocytes, and oligodendrocytes [21]. Electron microscopy demonstrated that the hNSCs established synaptic contact with the host terminals, and electrophysiological recording of these cells showed that grafted hNSCs behaved functionally like neurons with voltage-gated sodium currents and excitatory post-synaptic currents [21]. In addition, dose–BLI signal relationships in vitro were maintained in vivo and simultaneous hNSC tracking by MRI further corroborated this, as similar cell numbers could be extrapolated from the dose– magnetic resonance signal relationship ((Fig. 1) and [21]).

Fig. 1. in vitro and in vivo BLI of hNSCs.

(a) In vitro imaging analysis of genetically engineered hNSCs showing increasing Luc activity with cell density and a linear correlation (R² = 0.98). (b) Data are representative of 3 independent experiments performed in triplicate. (c–d) Representative BLI of stroke-lesioned rats transplanted with hNSCs and monitored for (c) 4 wks and (d) 8 wks post-transplantation. Quantitative analysis of Luc activity in these animals shows a stable BLI signal, suggestive of graft survival and the nonproliferative property of the hNSCs. Color scale bar is in photons/s/cm²/sr. (Reprinted with permission from Daadi MM, Li Z, Arac A, Grueter BA, Sofilos M, Malenka RC, Wu JC, Steinberg GK. Mol Ther 2009 Jul; 17(7): 1282-91.).

More recently, our group used BLI to track the homing of intravascularly-delivered NSCs in a hypoxia-ischemia mouse stroke model, and used this approach to study the role of the chemokine receptor, CCR2, in NSC injury-homing [25]. In accordance with previous studies, NSCs derived from wild-type mice successfully migrated from the vasculature to the ischemic parenchyma, whereas NSCs from CCR2−/− mice exhibited significantly decreased injury-homing. Similarly, wild-type-derived NSCs injected into mice deficient for CCL2 (CCR2 ligand) also exhibited reduced injury homing. These migration patterns as observed by BLI were further supported by histological detection of NSCs in the ischemic brain, and importantly, mice receiving CCR2+/+ NSCs showed significantly greater functional recovery post-stroke relative to those receiving CCR2−/− NSCs [25].

Together, these studies and others demonstrate that BLI is a reliable method for long-term, noninvasive, real-time monitoring of transplanted stem cells in animal models of stroke. A major advantage of BLI is the specificity and permanent expression of the Luc reporter gene, which is not lost during cell division. In addition, BLI is fast, versatile, and cost effective and offers high sensitivity with the ability to detect 100 to 1,000 cells in superficial anatomical areas [26–29]. As such, variations in the cell density of grafted cells may be quantified, thus offering a simple way to monitor survival, immune reaction against the grafts, and tumorigenicity. However, major disadvantages of BLI are the low spatial resolution and the negative correlation between intensity of light signal and tissue depth, which make current versions of this approach difficult for human application.

MRI

Perhaps the most promising and well studied of the various in vivo imaging modalities is MRI-based tracking of stem cells. Importantly, MRI has been used in clinical practice for the past 30 years to identify pathologic lesions in the brain and is therefore already a standard clinical adjunct for neuropathologies. MRI is based on the distribution of hydrogen atoms in tissues, essentially allowing visualization of water molecules in particular tissues. Each brain region exhibits different water content and has a unique composition of macromolecules, such as proteins and lipids, which bind water in differential amounts. This water composition determines contrast between brain structures, and introduction of MRI contrast agents alter the relaxivity of hydrogen atoms in their vicinity and therefore modify surrounding MR signal.

Early studies used the contrast agent gadolinium rhodamine dextran (GRID) to track NSCs in the stroke-lesioned brain, demonstrating the feasibility of cellular MRI to monitor cell migration in vivo. However, a recent study indicated deleterious effects of GRID-based contrast agents on long-term in vivo functional properties of NSCs grafted in the stroke-injured rodent brain [30]. Instead, supraparamagnetic iron oxide (SPIO) particles have become the contrast agent of choice for NSC labeling and subsequent in vivo tracking. SPIO is degraded safely along physiologic iron metabolism pathways and some SPIOs are already FDA approved for use in humans [31]. SPIO is also thought to be more sensitive and safer than gadolinium, which carries the theoretical risk of toxicity with cellular dechelation [32]. MRI using SPIO has the advantage of superior 3D spatial resolution but unfortunately has relatively poor sensitivity, detecting molecules at the millimolar to micromolar level [24]. Nonetheless, this approach does provide sufficient sensitivity for in vivo quantification of labeled stem cells, providing quantification is performed during initial seeding of the graft and prior to any cell death of the transplant population [33].

SPIO particles cause adjacent protons to experience a large dipolar magnetic field gradient and thereby detection via magnetic resonance as a hypointensity on T₂- and T₂*-weighted magnetic resonance [17]. Unfortunately, hypointensity on T₂ and T₂* can also be due to a number of false positives, including bleeds, air, hemorrhagic transformation of stroke lesions and macrophage phagocytosis of endogenous iron as a part of the inflammatory response to certain pathologies [34,35]. Paramagnetic gadolinium-based agents do not suffer from this problem, as they generate a hyperintense magnetic resonance signal on T₁ and T₂. However, the hyperintense signal on T₂ from gadolinium-based agents can obfuscate pathologic lesions such as strokes, which also generate hyperintense signals on T₂. Fluorinated nanoparticles, which rely on the magnetic resonance phenomenon of ¹⁹F nuclei, are also a promising way to label cells with a significantly improved signal-to-noise ratio, as ¹⁹F is not normally present in the human body [36,37]. However, at present, hydrogen-based MRI and SPIO cell labeling is the most studied modality.

SPIO uptake is affected by many factors, including cell type, species type, type of SPIO used (there are many types with different particle sizes), concentration of the SPIO in the growth medium and length of time the cells are incubated in an SPIO-containing medium [38,39]. The use of positively or negatively charged transfection agents may also enhance SPIO uptake efficiency [38,39]. Examples of transfection agents include dextrans, phosphates, artificial lipids, dendrimers and proteins such as poly-L-lysine and protamine sulfate, which is FDA approved [40]. One study also demonstrated that transfection with hemagglutinating virus of Japan envelopes (HVJ-Es) resulted in significantly more hNSC SPIO uptake, as measured by Prussian blue staining and atomic absorption spectrophotometry, when compared with use of a lipid transfection agent, Lipofectamine™ 2000, alone [38]. Magnetoelectroporation is also an efficient means of cellular iron oxide uptake and does not require the use of any non-FDA approved agents [41,42]. Unfortunately, the concentration of SPIO in NSCs is halved with cell division and thus the signal will degrade with in vivo cellular proliferation [38,39,43]. While this may not be a problem in many pathological states, it may be a problem when the cells are introduced to environments that favor cellular proliferation. In at least one study of C17.2 NSCs labeled with Feridex^® and then introduced into the shiverer dysmyelinated neonatal mouse brain, rapid cellular turnover and asymmetric cellular division resulted in SPIO dilution and poor correlation of magnetic resonance with histology [44]. Further study is needed to determine whether magnetic resonance signal can be detected for extended periods of time after transplantation. Long-term MRI will likely be variable given the type of cell used; the type, means and efficacy of labeling; and the pathologic microenvironment into which the cells are transplanted.

It is also unclear what happens to the released SPIO particles if the labeled cells die in vivo. While SPIO particles can be degraded via normal iron metabolic pathways, one study looking at Feridex-labeled cardiomyoblasts injected directly into rat myocardium found that Feridex could be retained in the native cardiac myocytes even after grafted cardiomyoblast cell death [45]. This led to incorrect and longer graft survival time by MRI versus bioluminescence in the same study. However, it is unclear if this same effect would occur with NSCs grafted into the stroke-injured brain.

Other than SPIO labeling, magnetic resonance reporter genes are another possible means to magnetic-resonance label stem cells. While this technique is still in its infancy, some groups have already demonstrated the ability to transfect cells with genes coding for heavy- and light-chain ferritins, which can be regulated by exogenous chemical agents, such as tetracycline. The subsequent uptake of endogenous iron by the ferritins then allows the cells to be visualized by magnetic resonance. This technique has already been used to image mouse cell lines, including the C6 glioma line [46–48]. Another technique for transfection involves the use of MagA, which is a gene from magnetotactic bacteria that codes for ferromagnetic nanoparticles. Zurkiya et al. labeled a human cell line with doxycycline-regulated MagA and administered doxycycline to induce SPIO synthesis in the transplanted human cells in a mouse model [49]. The cells were subsequently visualized by magnetic resonance. Use of magnetic resonance reporter genes would circumvent the problem of SPIO dilution with cellular proliferation. However, it should be noted that these reporters must be controlled by conditional regulation so as to prevent iron-overload toxicity. Further study into the possibility of magnetic resonance reporter genes is needed to determine if such an approach could be used to track stem cells transplanted into the stroke injured brain.

Our own group has extensively studied the feasibility and biological safety of using MRI and SPIO labeling for in vivo imaging of hNSCs. In one study, hNSCs were grown for 24 hours in a medium with Feridex SPIO (5 μg/ml) and protamine sulfate (2.5 μg/ml) as the transfection agent, for a labeling efficiency of 98% as determined by Prussian blue staining [43]. Labeled hNSCs were then compared with unlabeled hNSCs in vitro. There were no statistically significant differences between the two groups with regard to cell viability, proliferation rate, cell fate under differentiation conditions or electrophysiology of neural progenitor cell-derived neurons [43]. While there are no reports from other groups that SPIO labeling causes changes in cellular differentiation with regard to NSCs, Bulte et al. have contended that human mesenchymal stem cells labeled with Feridex and using poly-L-lysine as a transfection agent have normal osteogenesis, adipogenesis, cellular proliferation and cellular viability, but altered chondrogenesis, although it is unclear whether this was due to the SPIO or the transfection agent [50–52].

Our group also assessed whether SPIO labeling affected cellular viability, migration and differentiation of hNSCs in vivo. It has been shown that in the unique environment of the immature neonatal rodent brain, NSCs from the subventricular zone enter the rostral migratory stream and migrate to the olfactory bulb, where they differentiate and integrate in a site-specific manner [53,54]. Our group was able to visualize this migration with our SPIO-labeled hNSCs in a neonatal NOD–SCID rodent brain. T₂-weighted spin echo and 3D gradient-echo MRI taken at 3, 9, 12 and 18 weeks were able to follow the migration of the SPIO-labeled cells from the injection site in the lateral ventricle to the lateral and fourth ventricles at 3 weeks and along the rostral migratory stream to the olfactory bulb at 9, 12 and 18 weeks. This movement and cell fate was also corroborated histologically by staining with Prussian blue and the human-specific marker SC121. Immunohistochemistry was also employed to show that there were no statistical differences in cell viability, cell migration, cell fate or inflammatory response as measured by the pan-monocytic marker, Iba-1 [43]. We also assessed whether in vivo MRI signal correlated with cell viability. This was shown to be the case with injection of living SPIO-labeled hNSCs into the left striatum and injection of SPIO-labeled hNSCs killed by multiple freeze thaw cycles into the right striatum. While the hypointense MRI signal from the living cells in the left striatum exhibited minimal changes over 35 days, the hypointense signal from the killed cells on the right reduced markedly, by 52%, over the same time frame. Even though we observed a reduction in MRI signal, it did not disappear completely over the 35-day time period. Therefore, the inability to fully differentiate between viable and dead cells represents a limitation of SPIO-based cell imaging techniques and likely reflects transfer of SPIO particles from dying transplanted cells to surrounding host cells [33].

After determining that SPIO labeling did not affect cellular biology in vitro or in vivo, and that it could be used to detect grafted cells and their migration in vivo, we sought to assess whether MRI could be used to follow SPIO-labeled NSCs in vivo in the pathologic stroke environment [43]. 1 × 10⁵ SPIO-labeled hNSCs were transplanted medial to an ischemic cortical lesion 1 week after injury onset in rats subjected to the distal MCAO stroke model. Comparing MRIs from pre- and post-transplantation, the NSC bolus was clearly visible as a hypointensity at 1 week post-transplantation and targeted transparenchymal migration towards the stroke was visible at 5 weeks post-transplantation. At 5 weeks, the rats were sacrificed and the MRI results were confirmed by immunohistochemistry. Furthermore, assessment of cell viability in the pathologic stroked environment showed 51.3% survival of the transplanted SPIO-labeled hNSCs, which was similar to the survival of SPIO-unlabeled hNSCs in prior similar experiments [55].

In a second study, we investigated the fate of grafted SPIO-labeled hNSCs derived from human embryonic stem cells in a stroke environment over time and the fidelity of MRI to detect dose-dependent effects of the grafts on the lesion [21]. Cultures of hNSCs were treated with an SPIO-poly-L-lysine mixture for 3 days in vitro, harvested, and transplanted medially to the stroke at an increasing density of 50,000, 200,000, and 400,000 cell doses. MRI analysis clearly detected the grafts as hypointense areas in the striatum and in the stroke zone as a hyperintense region in the striatum and cortex on T2-weighted images (Fig. 2). The graft size for each analysis in the serial MRI scans demonstrated a linear correlation between the injected cell dose and the MRI size of the transplant. Three dimensional reconstructions of the grafts and stroke by surface rendering from the MR scans allowed for an accurate representation of both the graft and the stroke sizes and visualized the relationship between small and large grafts to stroke region. Immunohistochemical analysis of SPIO-labeled grafts performed with confocal microscopy and Prussian blue staining confirmed the MRI data and demonstrated the survival of the grafts and migration toward the stroke-damaged areas [21].

Fig. 2. MRI imaging analysis of hNSC grafts in an experimental stroke model.

(a–c) MRI horizontal and (d–f) frontal scans show dose-dependent size of the SPIO-labeled hNSC grafts as hypointense areas in the striatum (arrow) and medially in the penumbral zone of the stroke region distinguished as strongly hyperintense areas on T2-weighted images. The cells doses are 50,000 cells (a,d), 200,000 cells (b,e), and 400,000 cells (c,f). (g) Quantitative analysis of graft size, in consecutive coronal MRI scans, 600 μm spaced, in the 3 animal groups (N = 15) over the post-transplant survival time confirm the BLI data and show a stable graft size demonstrating survival of the graft. Three-dimensional surface rendering reconstruction of grafted rat brain from high resolution T2-MRI illustrate the grafts (green) and stroke (pink, red) in a representative animal from the (h–j) low dose and (k-m) intermediate dose group. (n) The MRI measured graft size show a strong correlation (R² = 0.99) with the cell dose transplanted. (o–q) Histological analysis using Prussian blue staining for SPIO particles demonstrate cytosolic deposition of blue crystals in the grafted hNSCs and migration of hNSCs toward the stroke area (asterisks in o, p). Interrupted line in (o) shows the boundary of the stroke zone. Scale bars = (p) 50 μm; (q) 20 μm. (Reprinted with permission from Daadi MM, Li Z, Arac A, Grueter BA, Sofilos M, Malenka RC, Wu JC, Steinberg GK. Mol Ther 2009 Jul; 17(7): 1282-91.).

Several studies have used MRI in novel ways to follow NSCs in vivo. As mentioned earlier, magnetic resonance has already been shown to be extremely useful for monitoring NSCs in animal models of stroke [43,56–60]. However, use of magnetic resonance has also enhanced our understanding of how stem cells can be utilized in the stroke environment. For example, GRID-labeled stem cells implanted into the contralateral hemisphere in stroked rats were followed by magnetic resonance and found to migrate transhemispherically along the corpus callosum towards stroke lesions [58]. Bimodal contrast agents, such as GRID, containing an MRI marker and a fluorescent marker can facilitate confirmation of cell transfection and subsequently histological detection of the transplanted cells using fluorescence microscopy. Another study found that SPIO-labeled NSCs that are placed in the uninjured mouse brain are activated by a stroke and subsequently migrate towards the stroke lesion as detected by magnetic resonance [56]. Song et al. showed that SPIO-labeled hNSCs injected intravenously into the tail vein of mice that had received middle cerebral artery occlusion strokes migrated to the peri-infarcted areas in the brain as soon as 3 days after injection, and the cells could be imaged in vivo by MRI [60]. This was later confirmed with Prussian blue and anti-BrdU staining. Monitoring of the initial cellular distribution after intravascular cell injection is critical and SPIO-based imaging has been demonstrated to be a valuable tool for this purpose [61]. By allowing a constant longitudinal assessment of cellular migration, we are gaining new perspectives on the mechanisms and means by which NSCs can be used in the ischemic brain.

More recently, Wang et al. described a novel fluorescent-magnetic nanocluster (FMNC) labeling system, which they developed to allow increased sensitivity and long-term MRI tracking of transplanted cells [62]. In a mouse MCAO model, FMNC-labeled mesenchymal stem cells (MSCs) were infused into the contralateral hemisphere and MRI was used to track their migration along the corpus callosum and into the stroke-injured hemisphere. This novel labeling system appears to allow increased MRI sensitivity, and use of a polystyrene scaffold to embed magnetite nanoparticles and the fluorescent probe promoted stable, long-term incorporation into the transplanted MSCs [62].

MRI has even been used in human clinical cases of stem cell transplantation. In one case, a patient with traumatic brain injury had exposed neural tissue collected and cultured to obtain a sample of NSCs that were subsequently labeled using SPIO, Feridex and a lipofection transfection agent, Effectene^®[63]. The cells were then implanted stereotactically near the region of brain damage and followed with T₂-weighted MRI every week for 10 weeks with a 3 T magnet. The hypointense signal generated by the cells demonstrated movement of the cells from the implantation site to the periphery of the lesion as early as the first week. However, the hypointense signal disappeared by the seventh week, which the group attributed to NSC proliferation [63]. In a second clinical study, spinal cord injury patients were transplanted with SPIO-labeled autologous bone marrow CD34+ cells and compared to patients receiving SPIO particles alone. Subsequent MR imaging at 20 and 35 days showed a persistent hypointense signal around the lesion site in 5 of 10 graft-receiving patients, which was absent in control patients [64]. In a third clinical trial, multiple sclerosis patients were injected intrathecally and intravenously with autologous SPIO-labeled MSCs and monitored by MRI 24-48 hours and 1-3 months post-transplantation [65]. Hypointense signal attributed to the transplanted cells was detected in the meninges of the spinal cord, nerve roots and the spinal cord parenchyma, however no accounts of the signal’s persistence over time or its relative consistency across the nine patients receiving SPIO-MSCs was provided [65]. Finally, a Swiss study assessed the safety of transplanting human cadaver islet cells into four patients with type 1 diabetes [66]. Transplanted cells were SPIO-labeled prior to intraportal delivery and then monitored by MRI at 5 days, 5 weeks, or 6 months post-transplant. Notably, all patients achieved insulin independence following cell transplantation. Although persistent hypointense spots could be detected in 3 out of 4 patients at 5 weeks post-transplant, this signal did not correlate with initial transplant dose, and changes in spontaneous liver hypointensity following iron treatment was suggested as a confounding factor [66].

Overall, MRI has several advantages as a strategy to monitor transplanted stem cells, but also has some drawbacks. Advantages include clinical applicability, high spatial resolution and, with the use of SPIO agents, some of which are already FDA approved, safety without perturbing cellular vitality or functionality. Drawbacks to this approach include dilution with in vivo proliferation, false positives from underlying pathology, and relatively poor sensitivity. Some groups have proposed use of magnetic resonance-sensitive reporter genes that generate SPIOs or ferritin and thereby sequester endogenous iron [67]. Such an approach may overcome some of the drawbacks associated with MRI, however, future studies will have to address whether expression of magnetic resonance-sensitive reporter genes interferes with stem cell function and/or reduces stem cell viability due to potential cytotoxicity associated with increased intracellular iron.

PET

PET is a third imaging modality that may be used to track transplanted stem cells in the stroke-injured brain. Given the accuracy and noninvasiveness of PET, it is routinely used in the clinic for both diagnosis and post-therapy monitoring of patients. The most commonly used radiolabeled probe to image brain activity is ¹⁸F-fluorodeoxyglucose ([18F]-FDG). Imaging of [18F]-FDG uptake by PET provides an estimate of regional glucose utilization, which can be normalized to a reference region of the brain. FDG-PET is approved by the FDA and is currently the most commonly performed imaging modality in the clinic.

Indeed, FDG-PET already has been used to image brain metabolism after neuronal progenitor cell transplantation into stroke patients [68]. Following neuronal transplantation, patients were injected with 7 mCi of [18]-FDG and PET imaging was performed 40 minutes later. PET images were normalized and the [18F]-FDG uptake in the infarct and per-infarct zones was expressed as percentage of baseline. PET scans performed 6 months post-transplantation revealed a 15% increase in [18F]-FDG uptake at the transplant site and the ipsilateral adjacent parenchyma in 6 of 11 patients [68]. Furthermore, improved motor outcome correlated with increased metabolic activity in the stroke (p= 0.02) and surrounding regions (p= 0.006) [69]. However, it is not known if the increased FDG uptake is related to metabolic activity in the transplanted cells, host neurite outgrowth, neovascularization, improved endogenous metabolism at the transplant site or a host inflammatory response.

While the aforementioned study indicates that PET can be used to detect downstream functional consequences of cell transplantation, additional studies have focused on the use of PET to directly image transplanted cells. This approach requires prior genetic modification of the transplant cell population to incorporate a transgene encoding an appropriate enzyme capable of uniquely binding a radiolabeled substrate. The herpes simplex virus type 1-derived enzyme, thymidine kinase (HSV1-tk), provides an example of this, with the key features being that this enzyme is not normally found in host tissue and is able to uniquely phosphorylate substrates composed of acycloguanosines. As a consequence of this phosphorylation step, the acycloguanosines accumulate in the cell, and thus prior radiolabeling of an acycloguanosine substrate allows specific detection of cells expressing the HSV1-tk transgene.

Although the use of HSV1-tk in conjunction with PET was originally developed as a method for specific detection and killing of cancer cells [70], it has since been used to successfully track human embryonic stem cells transplanted in the heart [71] and murine neural progenitor cells (NPCs) transplanted in an intracranial glioma animal model [72]. In addition, one human study has demonstrated the clinical feasibility of this approach, showing that HSV1-tk-modified cytolytic CD8+ T cells injected into a patient with glioblastoma multiforme were detectable in the tumor site by PET following systemic injection of the radiolabeled substrate, 9-[4-[¹⁸F]fluoro-3-hydroxymethyl)butyl]guanine ([¹⁸F]FHBG) [73]. However, it is important to note that while [¹⁸F]FHBG can be used to image the CNS in situations where the BBB is compromised, this substrate cannot cross an intact BBB and thus cannot detect labeled cells present in healthy brain tissue [72]. Nonetheless, this method is a promising approach for tracking stem cells in the context of brain injuries such as stroke, given that resulting damage to the CNS typically involves localized opening of the BBB.

Indeed, our group recently applied PET to track HSV-tk-modified hNSCs transplanted into the stroke-injured rodent brain. Prior to transplantation, hNSCs were genetically modified to express a triple fusion (TF) reporter gene encoding monomeric red fluorescent protein (mRFP), firefly luciferase (Luc), and a truncated version of HSV1-tk (HSV-ttk). hNSCs also were labeled with SPIO particles prior to transplantation, thus allowing us to simultaneously track cells by BLI, PET, and MRI and also to measure the functional efficacy of hNSCs on stroke recovery [74]. Using this multimodal imaging strategy, we found that grafted hNSCs reduced the infarct size of moderate strokes in a dose-dependent manner, while larger strokes could not be spared in this manner. In addition, PET imaging revealed increased metabolic activity in grafted animals and allowed visualization of functional grafted hNSCs in vivo [74]. Subsequent immunohistochemical analysis at 3 months post-transplantation confirmed survival and dispersal of hNSCs within the stroke-lesioned parenchyma, and showed that grafted cells differentiated into neurons, astrocytes, and oligodendrocytes [74]. Thus, this study demonstrates that the PET/HSV1-tk system can be used to monitor stem cells transplanted in the stroke-injured brain, and moreover, a multimodal imaging strategy as utilized in this study can provide a more comprehensive method of tracking grafted cells and can help overcome limitations inherent to each approach.

Given the benefits and drawbacks of each imaging modality, continued exploration of their combined use will be an important focus for future studies. While BLI can allow long-term monitoring and high sensitivity, complementary imaging by MRI can provide enhanced spatial resolution and detection in deep tissue structures, whereas use of the PET-HSV1-tk system allows for greater sensitivity and a sustained signal even during cell division. Given that the use of PET/HSV1-tk to track stem cells is a relatively new approach, more studies are needed to determine the safety of this technology and if its use affects stem cell viability or function post-transplantation. In addition, grafted cells that express HSV1-tk are vulnerable to anti-herpes drugs, such as acyclovir and gancyclovir [75,76], thus it will be important to identify or develop alternative enzyme-substrate systems that are compatible with PET. One example of an alternative system is the xanthine phosphoribosyl transferase reporter enzyme, which has the added advantage that xanthine reporter probes can cross the BBB [77]. Further characterization of this and other probe systems will help increase the utility of PET for monitoring transplanted cells. It is clear that each of these discussed imaging modalities will continue to be important to advance our understanding of stem cell-based therapies. In the future, when stem cell transplantation becomes an accepted treatment for stroke and other neurological disorders, development of safe and effective clinical imaging modalities will be indispensable.