Introduction

OPC transplantation therapies for regenerative neurology

Oligodendrocyte precursor cells (OPCs) are proliferative, stem-like cells of the central nervous system (CNS) that have emerged as a key transplant population to promote repair of myelin (the protective, fatty insulating sheath around nerve fibers)[1]. Myelin damage is a key contributor to the pathology of Multiple Sclerosis and spinal cord injury (SCI)[2–5]. The regenerative properties of OPCs are largely due to their capacity to generate oligodendrocytes, the cells that form myelin around nerve fibers[1] (Figure 1), but some evidence suggests that these cells may also dampen destructive processes in pathology sites[6].

Figure 1
figure 1

Schematic diagram illustrating the developmental stages of the oligodendroglial lineage. Oligodendrocyte precursor cells (OPCs) are proliferative cells which generate myelinating oligodendrocytes, as shown. The insets show typical OPCs (A2B5+) and oligodendrocytes (MBP+) derived from primary rat cultures (cell culture and immunostaining protocols are detailed in Additional file1).

Full size image

Transplantation of OPCs derived from a range of cell sources enhances myelin repair in animal models, including extensive myelin genesis and rescue from lethal conditions in dysmyelinating/ hypomyelinating mutant rodents[6–9]. Introduction of human OPCs into newborn shiverer mice resulted in extensive myelination, neurological improvement and enhanced survival in ~26% of mice[10]. Givogri et al. transplanted primary OPCs into a neonatal mouse model of metachromatic leukodystrophy, a genetic disorder leading to demyelination and extensive loss of oligodendrocytes[11]; transplant populations generated myelinating oligodendrocytes, identifiable one year post-transplantation, with motor function significantly improved compared with controls. Human embryonic stem cell (ESC)-derived OPCs, transplanted into adult rodent models of SCI, demonstrated remyelination and associated improvement in motor function[12]. From a clinical perspective, OPC transplant populations can be derived from numerous sources[13–16], expanded in vitro[14–16], have a good preclinical safety record[17] and have been approved for clinical trial (Geron Corporation, California; GRNOPC1 cells; phase I clinical trial for transplantation of human ESC-derived OPCs into acute SCI[17–19]). This trial recruited 5 of the 10 patients originally intended[20], but has now stopped enrolling, a decision taken by Geron on financial grounds[18, 21–23]. No adverse effects have been reported within one year of transplantation, and a US clinical trials database now lists this study as ‘complete’ (http://www.clinicaltrials.gov, trial identifier NCT01217008, accessed 01 July 2014)[18, 22]. Patients will be followed-up at both 5 and 15 years post-transplantation[20]. Through a deal with BioTime, Asterias Biotherapeutics have acquired the GRNOPC1 stocks (renamed AST-OPC1) and ‘plan to seek FDA clearance to reinitiate human clinical trials’ (asteriasbiotherapeutics.com/our-clinical-focus/opc1/, accessed 01 July 2014)[24]. In a review of 24 preclinical OPC transplant studies for SCI models, no instances of teratomas, systemic toxicity, allodynia, increased mortality or allogeneic immune responses were recorded[17].

While such progress in OPC transplantation is highly promising, neural cell therapies still face a number of technical issues/hurdles with respect to testing their efficacy for clinical translation. Here, we review the evidence that magnetic nanoparticles (MNPs) have high utility as multifunctional tools in addressing key challenges in OPC transplantation therapies, most notably in cell tracking. The term MNP encompasses physicochemically diverse synthetic particles, the common element being a magnetic component (Figure 2). This review will focus on particles containing the most widely used magnetic material, iron oxides - these have a good safety profile, with some formulations receiving approval for clinical applications: e.g. ferucarbotran (Resovist), Feridex (Endorem), Ferumoxsil (Lumirem/Gastromark) and Ferumoxtran-10 (Combidex/Sinerem) as MRI contrast agents[25]; NanoTherm for hyperthermic tumor therapy[26]; ferumoxytol (Feraheme) for iron-deficiency anemia[27].

Figure 2
figure 2

Schematic diagram illustrating possible MNP features. Iron oxides (typically magnetite, Fe3O4, or maghemite, γ-Fe2O3) provide contrast for MRI and confer ‘superparamagnetism’ to the final particle. A protective biocompatible coating may be functionalized to carry drugs, cell targeting molecules, fluorophores for histological detection and/or nucleic acids for gene delivery.

Full size image

Review

The MNP platform can address key challenges confronting OPC transplantation therapies

For neural cell therapies, non-invasive tracking of transplanted cells is essential to correlate functional neurological recovery with transplant cell biodistribution[28]. Further, post-mortem histological analyses are required to assess transplant cell survival, rejection, differentiation profiles and integration, including the extent of myelin genesis. MNPs have been shown to be broadly suitable for both non-invasive and histological imaging, serving as contrast agents for MRI and being readily detectable in post-mortem tissue[29–34]. MRI offers critical advantages for non-invasive imaging including (a) detailed anatomical imaging of inflammation, demyelination/remyelination and assessment of lesion size[35], in parallel with transplant cell detection[26]; (b) lack of potentially harmful radiation (in contrast to CT and PET scanning[28]); and (c) existence of significant infrastructure and expertise in place at clinics worldwide. MNPs provide MRI contrast for imaging through high magnetic moments, which disturb local magnetic field homogeneities[36], resulting in short relaxivity times in water protons in the immediate vicinity of the particles and loss of signal in T2*-weighted MRI images[35, 37, 38]. The contrast generated is proportional to the magnetization of the metal, and inversely proportional to the distance between metal and water protons, so particles designed with high iron content and/or iron near their surface are likely to provide enhanced contrast[26, 37]. Clinical MRI scanners have a resolution of ~500 μm, but high magnetic field (up to 9 T) research scanners have demonstrated a resolution of ~10 μm[26] (although these are unlikely to be safe for human clinical use), with recent refinements allowing for the identification of individual transplant cells[39]. It should be pointed out here that MRI cannot distinguish between intracellular and extracellular MNPs, and dead/dying MNP-labeled cells can therefore provide false-positives[40]. In order to address this confounding issue, studies have correlated MRI contrast with the presence of transplanted cells by post-mortem analyses such as immunostaining. Other methods, such as spatial correlation of MNPs with transplant cell-associated transgene expression or myelin production have also been used to unambiguously identify MNP-labeled OPCs within host tissue.

OPCs can be labeled for imaging applications with physicochemically diverse MNPs, but there is significant inconsistency within the literature in respect of experimental methodologies, particle design/characterization, and outcome measures – leaving doubt regarding the particle properties required to achieve optimal cell labeling. For example, Bulte et al. (1999) reported that CG4 cells (an oligodendroglial cell line) did not exhibit MNP-labeling when incubated with dextran-coated MNPs, although the authors report that “significant” MNP-labeling of these cells was achieved when the same particles were conjugated with anti-transferrin-receptor antibodies (no numerical data were reported in this study regarding the extent of cellular labeling; Table 1)[29]. When transplanted into spinal cord of myelin deficient (md) rats and normal littermates, these cells migrated up to 8.4 mm from the injection site (over 14 d), with ex vivo MRI signal correlating well with iron staining and new myelin. In contrast, Franklin et al. (1999) successfully labeled >60% of CG4 cells using a dextran-coated MNP without specific cell targeting strategies[30]. These cells were detected ex vivo seven days post-transplantation into adult rat ventricles. Frank et al. (2003) investigated MNP uptake in CG4 cells using the clinically-approved formulation Feridex (dextran-coated iron oxide particles[41–43]) with and without a complexed transfection agent (Lipofectamine Plus or poly-L-lysine, PLL)[33]. Labeling with unfunctionalized MNPs was reportedly “low” (not detectable using Perls’ Prussian blue iron stain), consistent with Bulte et al.’s (1999) study, but the cells were successfully labeled using both transfection agents. PLL-functionalized Feridex MNPs were also used by Lepore et al. (2006) who reported that “large numbers of Feridex particles were taken up” by transgenic OPCs co-cultured with “neuronal-restricted precursor cells”; however these co-cultures were uncharacterized and OPC-specific labeling was not quantified[32]. Five weeks post-transplantation into adult rat spinal cord, these cells were detected using ex vivo MRI demonstrating migration (up to 5 mm), with good correlation between MRI contrast, iron staining and transgene expression[32]. From these studies, there is insufficient data to reach conclusions regarding the potential physicochemical basis for the different labeling results obtained with dextran coated MNPs in OPCs, as properties such as size and zeta potential differ substantially between the studies, or are entirely unreported.

Table 1 Comparative data from MNP studies involving OPCs or oligodendroglial cell lines
Full size table

The Bulte group reported comparable uptake levels in CG4 cells, OPCs and other cell types, concluding that MNP-uptake is non-specific and independent of cell type[31, 45]. However, our group has reported substantial variability in MNP-uptake dynamics between neural cell types[52]. Concentration- and time-dependent uptake of carboxylated polystyrene MNPs was shown for four neural cell types (microglia, astrocytes, OPCs and oligodendrocytes) derived from primary cultures. Up to 60% of OPCs were labeled, with heterogeneity in the extent of MNP-loading. Notably, microglia exhibited very avid and extensive MNP uptake compared with the other cell types, with oligodendrocytes demonstrating the lowest levels of uptake[52].

Hohnholt et al. (2010, 2011) used MNPs with the goal of studying iron metabolism and toxicity, rather than labeling, in OLN-93 cells (an oligodendroglial cell line) reporting concentration-dependent uptake of both citrate- and dimercaptosuccinic acid (DMSA) coated MNPs (up to 300-fold increases in average intracellular iron)[47–49]. In a subsequent study, Petters et al. (2014) functionalized these DMSA-coated MNPs with a fluorophore and demonstrated uptake comparable to particles lacking conjugated fluorophores (69 nmol Fe/mg cellular protein control; ~1700 nmol/mg without fluorophore; ~1800 nmol/mg with fluorophore; to aid comparisons with other studies, we have re-calculated these values, as described in Additional file1; respectively, these values are ~1 pg Fe/cell, ~23 pg Fe/cell and ~24 pg Fe/cell)[55]. Importantly, the authors characterized these particles before and after functionalization, an oft-omitted step (Table 1;[29, 33]): size increased by 17%, zeta potential changed from -20 to -28 mV. This study noted nine-fold greater levels of uptake in the absence of serum, compared to serum-supplemented medium, illustrating the influence of the biochemical composition of media on particle-cell interactions.

Many MNPs are readily detected due to their metal content, for example by simple histochemical iron staining, which in turn correlates well with MRI observations of MNP-labeled OPCs post-transplantation[29, 32]. For particles not amenable to metal-based detection (e.g. due to low iron content), fluorophores can be incorporated, either internally or attached to the particle surface, facilitating post-mortem detection by fluorescence imaging. For example, Kircher et al. demonstrated detection of a cyanine dye (Cy5.5)-tagged dextran-coated MNP through fluorescence microscopy of post-mortem tissue, although this particle was used to delineate a brain tumor, rather than to track a transplant population[56].

Long-term tracking of transplanted cells is highly dependent upon label retention, but dilution of MNP-labeling has been observed in vitro and in vivo, being attributed at least in part to cell proliferation[29, 40]. This represents a particular challenge for imaging the biodistribution/migration of proliferative populations such as OPCs. Although MNP retention by OPCs has been reported for 7 d in vitro and 6 weeks post-transplantation (upper limits not determined)[31, 45], no studies have systematically quantified proliferative dilution of MNPs, or distinguished between particle loss due to cell proliferation versus cellular excretion by exocytosis. A further concern is whether particles are retained during differentiation into mature oligodendrocytes, as the primary goal of OPC therapy is to replace lost/damaged oligodendrocytes[1, 5]. Therefore, the ability to image these differentiated cells long-term in areas of regeneration is key for myelinating therapies. Oligodendrocytes are post-mitotic cells, therefore particle loss due to proliferative dilution is eliminated. Indeed in our experiments, when pulse-labeled OPCs were subsequently differentiated and maintained for 30 days, a significant proportion (>50%) of oligodendrocytes displayed MNP-labeling, suggesting that the differentiated progeny can ‘inherit’ MNPs and retain the label for long-term imaging[53].

MNPs have promising safety profiles in OPCs

In order to develop MNPs for clinical cell therapies, it is of paramount importance to assess their potential cytotoxic effects in neural transplant populations. Oligodendroglial cells contain more iron than any other CNS cell type, but are also the most vulnerable to excess iron, which typically leads to oxidative stress due to reactive oxygen species (ROS)[49]. It is of note that oxidative stress has been linked with oligodendrocyte damage in diseases such as Multiple Sclerosis[57, 58], indicating that MNP-induced genesis of ROS could be similarly deleterious to labeled transplanted oligodendroglial cells. In other neural cells, MNPs have been shown to impair cellular function through mechanisms including disruption to the cytoskeleton/cell membrane[59, 60] or intracellular trafficking processes[61, 62], and direct damage to intracellular organelles[61] including by iron release during particle degradation[63]. Through these or other mechanisms, MNP uptake could also perturb cellular behavior, including capacity for migration or proliferation[64].

The Dringen group have used the OLN-93 oligodendroglial cell line to conduct the most detailed MNP-OPC toxicity studies to date, including demonstration of uptake of citrate-coated MNPs without affecting viability, morphology or proliferation, and without evidence of iron leaching[47]. Ferritin was greatly upregulated in response to increased Fe levels, storing Fe in a redox inactive form and protecting against iron-related toxicity[47]. A battery of assays found no evidence of acute cytotoxicity (72 h) for DMSA-coated MNPs[49]. For the same MNPs and cells, another study reported morphological changes, decreased glutathione (an antioxidative molecule) and increased ROS, but these changes were reversible and did not affect viability[48, 65]. Consistent with these data, OPCs labeled with other MNPs are generally reported as having viability and behavior comparable to unlabeled OPCs (Table 1;[29, 31, 32, 45, 52, 53, 55]).

Combinatorial therapies and OPC transplantation: using multimodal MNPs to achieve multiple therapeutic goals

While cell therapy alone is demonstrably efficacious, a widely-held view in the regenerative neurology community is that ‘combinatorial’ therapies (e.g. cell transplantation plus drug/gene delivery) achieve more impactful clinical regenerative outcomes than single therapeutic strategies[66–70]. For example, transplanting OPCs genetically engineered to secrete neurotrophic factors showed significantly greater improvement in SCI injury models than transplanting unmodified OPCs, or fibroblasts secreting the same neurotrophins[68, 71]. A major translational challenge currently is to achieve safe and effective genetic engineering of transplant populations. We have shown that MNPs can deliver both reporter and therapeutic genes to OPCs, a process significantly enhanced by the use of state-of-the-art ‘magnetofection’ strategies (applied static or oscillating magnetic fields to enhance particle-cell contact; up to 21% transfection efficiency in OPCs derived from primary sources)[50]. In contrast to the precursor cells, differentiated oligodendrocytes showed far lower transfection levels (up to 6%), suggesting that the proliferative or endocytotic properties of the OPCs may make these cells relatively amenable to MNP-mediated transfection compared with their progeny[72]. As far as we are aware, these are the only reports of MNP-mediated gene delivery to cells of the oligodendrocyte lineage available.

A further translational challenge is achieving targeted delivery of transplant cells to lesions while limiting secondary pathology. Spatial manipulation of MNP-labeled cells has been demonstrated using external magnetic fields – a technique that could retain/localize transplant cells at target sites by magnetic cell ‘capture’ following intravenous/intrathecal delivery, of high relevance in situations where a limited cell source exists. For example, an implanted magnet localized (limited dispersion of) MNP-labeled cells at a rat spinal cord lesion site following intrathecal delivery of mesenchymal stem cells[73] and bone marrow stromal cells[74]. Magnetic fields have also been used to localize MNP-labeled cells at a specific region of the retina following intravitreal injection (reportedly ~360000 transplanted cells, compared to ~10000 cells without applying a magnet), and following intravenous delivery of cells (~42000 cells, compared to ~4000 cells without applying a magnet)[75]. Magnetic cell targeting has not yet been demonstrated for OPCs, but may be feasible for SCI as above. Furthermore, the superparamagnetic properties of particles used for cell labeling (i.e. where particle magnetic properties are exhibited only in the presence of a magnetic field) can help overcome issues of cell aggregation and blockade of capillaries following systemic delivery[76]. In conjunction with the imaging potential of MNP-labeled transplant populations, the above findings highlight the high potential of MNPs to serve as a ‘multifunctional platform’ to address key challenges in neural cell therapy[54], summarized in Table 2.

Table 2 Challenges for cell transplantation therapies and the relevant utility of magnetic nanoparticles
Full size table

However, it should be noted that the overwhelming majority of MNPs described so far for neural applications have been unimodal. Rapid advances in materials chemistry in recent years have led to the development of a spectrum of complex, multimodal MNPs which simultaneously mediate multiple functions[54]. For cell therapies, purpose designed multimodal MNPs could mediate cell imaging, genetic modification and magnetic cell targeting. One such multimodal MNP was recently described, with high iron content for MRI contrast (and possibly cell targeting), a fluorophore for histological detection, and potential for gene delivery which was demonstrated in astrocytes – a major neural transplant population[54]. The particles were synthesized using a chemical grafting procedure to link polyethyleneimine (PEI) covalently to the surface of Fe3O4 MNPs. This methodology allowed for permanent linking of the PEI to the MNP surface, of benefit for use in biological fluids with high electrolyte concentrations; this also enabled overall particle size to be restricted to <50 nm and resulted in high iron content for the particles (ca 65% by weight), of potential benefit for imaging and magnetic targeting applications[54]. A red dye (rhodamine B isothiocyanate, RITC) was then bound to the PEI skeleton, with the final particles being denoted Fe3O4-PEI-RITC. The particles could be imaged using standard fluorescence/confocal microscopy and MRI, and were compatible for use with a range of histological methods[54]. Further, the chemical design of the particles also allowed for high versatility with respect of the use of other functional polymers and binding chemistries for nanoparticle functionalization, therefore particles of greater functional complexity can be evolved from this basic prototype. Using a simple one-step procedure, we have found that ~60% of OPCs could be safely labeled with these novel MNPs (Figure 3, previously unpublished data; particle properties can be found in Table 1, ref[53]), highlighting the therapeutic potential of multimodal particles in OPC transplantation[53]. Whilst our previous work showed that the gene delivery capacity of the particles was low overall (<1%), we consider that with further work directed towards enhancement of their transfection potential, such particles can prove a valuable ‘theragnostic’ tool for the developmental testing and clinical translation of neural cell therapies for regenerative neurology[54].

Figure 3
figure 3

OPCs can be labeled with a multimodal MNP. ~60% of OPCs derived from a primary source exhibit uptake of a multifunctional MNP. (a) Phase contrast micrograph showing Perls’ iron staining of MNPs (arrows). (b) Z stack fluorescence micrograph confirming intracellular presence of MNPs, both perinuclear (crosshairs and dashed arrow) and cytoplasmic (arrow). Arrowhead indicates extracellular particle accumulation. OPCs were derived from primary rat cerebral cortex cultures and plated 24 h before MNP incubation: Fe3O4-PEI-RITC MNPs, 20 μg/ml, 24 h[54]. Cultures were then fixed (4% paraformaldehyde) and immunostained[54]. Further details can be found in Additional file1. Particle characteristics are detailed in Table 1. A2B5 is an OPC marker. RITC = rhodamine B isothiocyanate.

Full size image

Conclusions

Biological perspectives: the need for standardization of reporting

The MNP platform offers high promise for neural transplantation applications, but the field is still in its relative infancy. In-depth and cross-disciplinary studies between materials chemists and transplantation neurobiologists are required to fully evaluate MNPs as an adjunct tool for OPC transplantation. For example, despite the key advantages offered by multimodal MNPs for OPC transplantation, there is a critical lack of neurocompatible and multimodal MNPs, representing a major scientific and commercial gap. The potential for magnetic cell targeting of OPCs to injury foci has never been assessed, and the processes of proliferative dilution and particle ‘inheritance’ by daughter oligodendrocytes are poorly understood. Further, much of the research investigating MNP uptake and handling by OPCs has relied on cell lines, whose behavior can differ markedly from primary cells – consequently, biological data derived from cell lines may have limited predictive value. For example, Pinkernelle et al. report six-fold greater MNP-labeling in the ‘neuron-like’ cell line PC12 than in primary neurons[51]; similar comparative analyses are required for OPCs.

The standardization of data reporting from MNP-labeling studies is essential to guide advances in nanoparticle synthesis and design. As with many biomaterials studies, MNPs used for OPC labeling are typically not fully characterized, yet these details are essential to identify parameters relevant for improving biomaterial design. There has been little systematic attempt to correlate MNP physicochemical properties with extent of OPC labeling (Table 1), of high relevance from a cell therapy perspective. Findings regarding the ability of oligodendroglial cells to take up MNPs without conjugated targeting molecules/transfection agents are contradictory (e.g. Bulte[29] and Frank[33] versus Franklin[30], all using dextran-coated MNPs); the reasons underpinning these differences are difficult to address in the absence of detailed particle characterization. Typically, reports should include size, shape and surface charge/functionalities of the final particle, measured within physiologically relevant media. The evaluation of OPC interactions with MNPs possessing a wider range of physicochemical properties can inform the tailored development of MNPs for specific transplantation applications. Such investigations should ideally include ultrastructural analyses of particle-cell interactions, along with evaluations of intracellular handling and particle fate to establish cellular processing mechanisms for different particles. This information can guide the development of MNPs with potential for endosomal escape, or suggest specific uptake mechanisms to which MNPs should be preferentially targeted for optimal labeling.

Other substantial knowledge gaps are apparent from the literature. Few studies report the proportions of OPCs exhibiting MNP-labeling, or conduct assessments of the extent of MNP-loading and its correlation with imaging capacity. More often, researchers provide an average iron content per cell measurement, which will mask any heterogeneity of particle accumulation within a cell population. This is particularly relevant to primary populations (the most likely cell source for transplantation therapies) which show considerable heterogeneity in behavior including particle uptake[52], unlike cell lines which behave in a relatively clonal manner[77]. Most studies report limited MNP-associated cytotoxicity in OPCs, but generally without numerical viability/safety data, a significant shortcoming as this information is vital to developing biocompatible particles and safe labeling protocols. Microarray/proteomic analyses are essential for detailed molecular analyses of MNP toxicity, particularly the long term safety of transplant populations. This should progress in parallel with functional assays of the regenerative capacity of transplanted MNP-labeled OPCs (e.g. cell migration and myelin genesis). It can be predicted that such work can facilitate the development and application of this platform technology to neural cell therapies, in order to promote repair mechanisms following neurological pathology – currently a key goal for regenerative medicine globally.