Introduction

Major human brain and spinal cord injury remain serious problems that currently have no effective treatments. Societal costs, as well as the personal loss sustained by victims and their families, are considerable. The transplantation of stem/progenitor cells may provide an effective treatment for central nervous system (CNS) injury due to the self-renewing and multipotential nature of these cells. Stem/progenitor cells may repair injured nervous tissue through replacement of damaged cells, neuroprotection or the creation of an environment conducive to regeneration by endogenous cells. Stem/progenitor cells can be categorized by their embryonic, fetal or adult origin, and fetal and adult cells can be further identified by their tissue of origin. Several types of precursor cells have been transplanted into the injured CNS,1, 2, 3, 4, 5 including adult bone marrow-derived mesenchymal stromal cells (MSCs). To date, no randomized, controlled human clinical trials have been reported of the transplantation of precursor cells for CNS injury. Clinical trials, however, have been undertaken in spinal cord injury with human olfactory ensheathing glia, Schwann cells, fetal spinal cord and MSCs, as reviewed by Tator.6 This paper reviews the use of MSCs for the repair of CNS injury.

Whole adult bone marrow contains a mixture of hematopoietic cells, including mononuclear cells such as macrophages, and nonhematopoietic marrow stromal cells. Bone marrow-derived MSCs support the growth and differentiation of hematopoietic stem cells (HSCs). MSCs can be isolated by their ability to adhere to the plastic containers used during tissue culture. MSCs are also referred to as mesenchymal stem cells, but this term should be reserved for a subset of these or other cells that demonstrate stem cell activity by clearly stated criteria.7 MSCs contain a population of cells that are self-renewing, and are capable of differentiating into multiple mesodermal tissues, including bone, cartilage, fat and muscle.8 They produce many neurotrophic factors and cytokines,9 can assume other phenotypes, suppress inflammation and trigger production of reparative growth factors as demonstrated after transplantation into other organ systems such as the lung10 and heart.11 Potential advantages of MSCs over other types of transplanted cells are shown in Table 1 which include their ability to be harvested from autologous donors, their relatively rapid expansion ex vivo and the possibility that allogeneic MSCs are nonimmunogenic and, hence, readily available for clinical use,12 although some studies have indicated that immune suppression improves transplant survival.13, 14, 15 Thus, they have been identified as promising candidates in the treatment of CNS injury.

Table 1 Advantages of MSCs versus PBPCs, umbilical cord blood and adult NSPCs
Full size table

Much of the interest surrounding the use of stromal cells in CNS injury began with the discovery that MSCs from both rats and humans give rise to cells with neuronal morphology in vitro, and may also express neuronal or astrocytic markers in vitro.16, 17, 18, 19, 20, 21 However, it has also been shown that continued expression of neuronal markers such as the neuronal nuclei marker NeuN may depend on the continuation of specific, nonphysiological environmental conditions in vitro.22 Furthermore, bone marrow cells expressing green fluorescent protein (GFP) infused into the systemic circulation of irradiated mice did not show neuronal, macroglial or endothelial differentiation, even after CNS injury, and most that were found in the brain were perivascular macrophages.23 This finding is supported by evidence from nonirradiated mice with surgically anastomosed vasculature.24 While adult rat and human MSCs transplanted into embryonic rat brain demonstrate remarkable neuronal differentiation,25 transplantation into the developing CNS is quite different from that of the adult. Cells of the HSC fraction of human bone marrow, when transplanted into lesions of the developing spinal cord in the chicken embryo also differentiate into neurons.26

Therefore, despite these promising initial results, the in vitro and in vivo differentiation of bone marrow-derived MSCs have been recently re-evaluated. Morphological ‘neuron-like’ changes could be created by rapid disruption of the actin cytoskeleton, and an increase in immunolabeling for neural markers may be the result of cellular toxicity, and can also be induced in rat dermal fibroblasts.27, 28 Human mesenchymal cells also transiently acquire the morphological appearance and the biochemical makeup typical of neurons by independently regulated mechanisms.29 While this evidence does not rule out the possibility of transdifferentiation of MSCs, it does indicate the necessity for caution in identifying criteria for the phenotyping of neural cells. If it is true that MSCs do not differentiate into cells of neural lineage after transplantation in adults,30 the role of other mechanisms for promoting functional recovery should be considered. That is, neural differentiation and/or cellular replacement may not be essential requirements for neurological recovery in either brain or spinal cord injury. Other possible beneficial effects and mechanisms of action of MSCs are more likely and should be explored.

This review examines MSC transplantation for both brain and spinal cord repair, and is restricted to MSCs derived from bone marrow. MSCs can also be isolated from other sources such as adipose tissue, umbilical cord blood and mobilized peripheral blood, and have also been used in transplant studies,31, 32, 33, 34, 35 although yields from the latter two are extremely low. Transplantation of MSCs for CNS repair was first reported in 2000 by Chen et al.,36 after administration of bone marrow with brain-derived neurotrophic factor (BDNF) into a model of middle cerebral artery occlusion (MCAo). This group reviewed their data on this topic in 2002,37 but considerable work has been reported since then. The present review will discuss these and subsequent reports of MSC transplantation for adult CNS injury in rodents, primates and humans. Models of brain injury include both traumatic brain injury and ischemia, and these will be discussed first, followed by MSC transplantation for traumatic spinal cord injury.

Brain repair: traumatic brain injury

Nonhuman MSCs

MSCs have been used in experimental repair of the injured brain, and contusion injury models have been utilized to mimic human traumatic brain injury. Chopp and Li37 found functional recovery and attributed the beneficial effects to the enhancement of endogenous restorative and regenerative processes. This research found that MSCs obtained from donor rats or humans could be administered either intra-arterially or intravenously (i.v.), resulting in targeting of injured tissue and improvement of functional recovery, although it was noted that intra-arterial injection provided no added benefit and could cause ischemia. It was felt that therapeutic benefit could be attributed to brain plasticity including mechanisms of angiogenesis, neurogenesis, synaptogenesis, dendritic arborization and reduction of apoptosis in the boundary zone of the injury. Additional publications by this group and others are outlined in Table 2. This group has also emphasized the role of BDNF and nerve growth factor (NGF) in the injured rat brain. Adult rat MSCs previously cultured with either of these neurotrophic factors led to a higher number of engrafted cells after transplantation into the adult rat brain adjacent to the injury site and improved motor function. A small number of cells stained for either astrocytic or neuronal markers,38 but were far too few to provide cellular replacement. This group also reported that i.v. administration of MSCs 1 day after brain injury in the rat brain resulted in an increase in BDNF and NGF but not fibroblast growth factor-2 (FGF2).39 Both intracerebral and i.v. MSC administration promoted endogenous progenitor cell proliferation after traumatic brain injury,40 and functional recovery was dose dependent and persisted for at least 3 months.41 This group has made a significant contribution to this field, and their important discoveries need to be replicated in independent laboratories.

Table 2 Models of traumatic brain injury and transplantation of non-human MSCs
Full size table

Hu et al.42 have shown that intracisternal MSCs migrate into the injured rat brain, improve neurological function and enhance local gene expression of BDNF and NGF.43 The potentially important role of induction of local neurotrophic factors is also supported by Chen et al.,44 who showed that intraventricular injection of MSCs in a murine model resulted in an increase in NGF levels as measured in the cerebrospinal fluid. Other mechanisms have been suggested to explain the action of transplanted MSCs, such as an elevated level of transforming growth factor-β (TGF-β) that helps to reduce the formation of scar tissue, and restoration of cerebral blood flow and the blood–brain barrier in models of traumatic brain injury45 and stroke.47 Most of these studies found a small number of cells expressing markers of neural lineage; however, many have proposed alternative explanations for their findings of functional recovery, either because they found no evidence of neural differentiation after transplantation, or because the cells were so few in number. In contrast, functional improvement was documented with MSCs previously exposed to a ‘neuronal’ induction medium which resulted in increased differentiation into a neuronal phenotype.46

Human MSCs

Transplantation of human bone marrow-derived MSCs into animal models of traumatic brain injury has also been examined by Chopp's group with similar results. When administered i.v. to adult rats with experimental brain injury, improved functional outcome has been documented for at least 3 months. MSCs migrated into the injured brain and preferentially localized around the injury site, with a small number of donor cells expressing neuronal and astrocytic markers.48, 49 These cells enhance connexin43 gap junction intercellular communication in cultured astrocytes, another potential mechanism of action.50

Brain repair: cerebral infarct

Nonhuman MSCs

MSCs have also been studied after transplantation into an ischemia model of MCAo with promising results. MSCs have been shown to migrate preferentially to the ischemic cortex after MCAo occlusion.51 It was first shown in 2000 that intracerebral transplantation of bone marrow with BDNF enhanced differentiation of bone marrow cells and significantly improved motor recovery.36 This was followed by a report that intracerebral transplantation of nonhematopoietic bone marrow cells also improved functional recovery52 (reviewed in Chopp and Li37). Subsequent publications on the transplantation on nonhuman MSCs into models of stroke are summarized in Table 3. As with their studies of traumatic brain injury, Chopp's group found that i.v. MSC injection in a rat stroke model also resulted in functional improvement. MSCs were found to localize preferentially to the ipsilateral ischemic hemisphere, and a reduction in apoptosis, increase in endogenous cell proliferation and increase in FGF2 immunoreactive cells were found in the treated animals.53 MSCs injected 1 month after stroke demonstrated positive effects in a chronic model with a reduction in scar thickness, increased proliferation of endogenous cells and oligodendrocyte precursor cells, and increased expression of the chemokine stromal cell-derived factor-1 (SCF1) along the ischemic boundary zone and its receptor on the MSCs.54 Intracarotid transplantation in this model also markedly increased vessel sprouting, synaptophysin expression and the number of oligodendrocyte precursors in the corpus callosum, suggesting axon–myelin remodeling as a mechanism of action.55 To examine these effects in long term, Chopp found that rats demonstrated significant improvement in neurological outcome at 4 months after transplantation, and showed increased growth-associated protein-43 (GAP43) expression (a marker of axonal growth cones) among reactive astrocytes in the scar boundary and subventricular zone.56 MSCs can also increase astrocytic survival after ischemic injury in vitro and upregulate astrocytic BDNF, vascular endothelial growth factor (VEGF) and FGF2.62

Table 3 Models of stroke and transplantation of non-human MSCs
Full size table

Other groups have shown similar beneficial effects after intracerebral transplantation into a murine model of cerebral infarct. MSCs showed proliferation and migration into the injured cortex, and transplanted cells expressed markers for both neurons and astrocytes.57, 58 Some transplanted MSCs expressed both the neuronal marker MAP2 and the γ-aminobutyric acid A (GABA-A) receptor.59

The potential benefits of FGF2 are supported by studies showing that gene-transfected MSCs expressing this growth factor demonstrated improved functional recovery after stroke compared with MSC transplantation alone.60 Ex vivo gene transfer has also been attempted with hepatocyte growth factor (HGF) which resulted in decreased apoptosis, increased neuron survival in the ischemic boundary zone and functional improvement.61

Human MSCs

Intravenously injected human MSCs transplanted into adult rats after MCAo expressed the immature neuronal marker doublecortin and demonstrated improved functional recovery.63 Several reports have focused on the role of neurotrophic factors such as an increase in insulin growth factor-1 (IGF1) and its receptor which may play a role in neuronal differentiation. Others have suggested that transplanted MSCs facilitate recovery from brain and spinal cord lesions by releasing brain natriuretic peptide (BNP) and other vasoactive factors to reduce edema, decrease intracranial pressure and improve cerebral perfusion.64 Human MSCs cultured in supernatant derived from ischemic brain extracts increased production of BDNF, NGF, VEGF and HGF, lending further credence to the hypothesis that increased expression of trophic factors may be responsible for the beneficial effects of MSCs,65 human MSCs have also been shown to augment collateral remodeling by paracrine signaling through release of several cytokines such as VEGF and FGF2 rather than by cell incorporation into vessels.66 As well, human MSCs can improve neurological functional recovery in mice with experimental autoimmune encephalitis (EAE), possibly via reduction of inflammatory infiltrates and areas of demyelination, stimulation of oligodendrogenesis and by elevating BDNF expression.67, 68 Human MSCs transfected with the BDNF gene also showed improved functional recovery and reduced infarct size through a reduction in apoptosis.69 It has also been shown that combination therapy with MSCs and a nitric oxide donor can enhance angiogenesis and neurogenesis.70

Spinal cord repair

Nonhuman MSCs

Failure of axons to regenerate after spinal cord injury has been attributed to an inhospitable environment containing inflammatory mediators, lack of trophic support and inhibitory molecules. It has been suggested that MSCs can help to overcome some of these barriers to regeneration.

As previously discussed, there is controversy in the literature over the fate of MSCs after transplantation into the injured spinal cord, and also whether they provide any functional recovery. These studies are summarized in Table 4. As in the case of studies with brain, it is unlikely that cellular replacement itself in the injured cord influences outcome, in part because of the small amount of neural differentiation documented and in view of the rapidity of recovery. In vivo fluorescence tracking with GFP suggests that mouse MSCs transplanted into the rat spinal cord migrate towards the injury site within 4 weeks after transplant, and some of these cells express neuronal or astrocytic markers.71 Functional improvement was also observed in a mouse model, either with57 or without72 expression of markers of neurons and astrocytes. Magnetic resonance tracking of implanted adult rat MSCs labeled with iron-oxide nanoparticles into the injured rat brain and spinal cord, either intracerebrally or i.v., demonstrated migration to the lesion site, a very small number of MSCs differentiated into neurons but not astrocytes, and the rats with spinal cord injury demonstrated functional recovery.73 Chopp et al.74 found that intramedullary transplantation of MSCs 1 week after spinal cord injury improved functional outcome over a 5-week period, with a few cells expressing neural markers. Wu et al.75 also showed functional recovery after MSC transplantation at the time of injury and attributed this to enhanced differentiation of endogenous neural stem cells as indicated by in vitro coculture experiments. In contrast, Hofstetter et al.76 observed functional recovery only when MSCs were transplanted in a delayed fashion after 1 week, and suggested that the strands formed by MSCs may provide axon guidance to explain the improvement rather than cellular replacement.

Table 4 Models of SCI and transplantation of non-human MSCs
Full size table

The therapeutic effects of MSCs in chronically injured rats after intramedullary administration have also been examined. When MSCs were transplanted 3 months after injury, functional benefits were seen at 4 weeks after transplantation, which continued to 1 year.77 Grafted MSCs formed cellular bridges across the cavity in the spinal cord, and expressed some astroglial and neuronal markers, along with ependymal proliferation.78

Others failed to find functional recovery with the Basso Beattie Bresnahan (BBB)90 open field locomotor scoring system in rats, but agreed that MSCs were associated with preservation of host tissue including the white matter, and provided directional guidance to regenerating axons.79 These inconsistencies in the literature strongly suggest that more investigation is required to identify the cellular fate of transplanted MSCs to further elucidate the mechanism of action of any functional neurological recovery. Other postulated mechanisms include the production of substances by MSCs such as fibronectin which can counteract the inhibitory effects of the glial scar.91 Remyelination has also been suggested, and transplantation of a murine bone marrow fraction remyelinated the focally demyelinated adult rat spinal cord with a predominantly peripheral myelin, as compared to transplantation of CD34+ hematopoietic cells which also survived in the lesion but did not produce myelin.80 This finding was further supported by the isolation of a mononuclear fraction from adult rat bone marrow, which resulted in remyelination after i.v. transplantation. This myelin was produced, at least in part, by the transplanted cells, was characteristic of both central and peripheral myelin, and resulted in improved conduction velocity of the remyelinated axons.81 Similar results were obtained with direct injection of murine MSCs.82

As with the brain experiments, several groups have explored the possibility of intraventricular, intrathecal or i.v. administration of MSCs for spinal cord repair. Ohta et al.83 examined MSCs infused into rat cerebrospinal fluid (CSF) through the fourth ventricle after T8-9 contusion injury, and found that they promoted functional recovery and reduced cavity formation at the injury site. Lumbar puncture delivery of adult rat MSCs into spinal cord-injured rats resulted in reduced cavitation, which was maximal when cells were transplanted within 14 days of injury.84 Direct transplantation of adult rat MSCs into the spinal cord demonstrated improved survival when compared to i.v. administration.85, 86

In vitro predifferentiation of MSCs has also been attempted before transplantation into the injured spinal cord in an attempt to enhance repair. MSC-derived Schwann cells plus Matrigel, a synthetic scaffold material, were shown to promote axonal regeneration and functional recovery after complete transection of the adult rat spinal cord.87 Cells ‘neurally’ induced in vitro were transplanted into rhesus monkeys resulting in functional recovery.88 In another study, MSCs supported modest growth of sensory and motor axons 1 month after grafting directly into the injured cervical spinal cord of rats, but in contrast, cells ‘neurally’ induced in vitro did not sustain this phenotype in vivo, and did not provide extra benefit.89 Transduction of MSCs to overexpress BDNF resulted in a significant increase in the extent and diversity of host axonal growth but did not show functional recovery.

Human MSCs

Human MSCs have been transplanted into the murine spinal cord either intra-lesionally with a synthetic polymer or i.v., and while both result in functional improvement, no attempt was made to identify the mechanism in this study.92 More recent work shows that oligodendrocytic but not neuronal differentiation was observed after i.v. transplantation into a rat model of spinal cord injury.93 Transplanted cells survived without immunosuppression and resulted in functional recovery at 3 and 4 weeks. Human MSCs, like those of rodents, secrete neurotrophic factors and cytokines, and can enhance functional recovery after transplantation into the injured spinal cord despite very few cells remaining after 11 weeks.94 This suggests that persistence of the transplanted cells may not be required for functional recovery.

In 2005, Park et al.95 reported the first trial of MSCs in humans with spinal cord injury and showed that humans treated with autologous whole bone marrow transplanted into the site of spinal cord injury, along with i.v. granulocyte–macrophage (GM)-CSF, showed improved neurological function, but the full therapeutic value of this protocol is still being investigated. Unfortunately, the fate of the transplants and their mechanism of action are not currently known. There is one other published report about additional current clinical trials of MSCs in humans with spinal cord injury, but no definite results have appeared.93 There are several other unpublished trials underway (personal communication, reviewed in Tator).6

Conclusions

Although several studies show significant functional improvement from transplantation of MSCs after either brain or spinal cord injury, the mechanisms are still unclear. Table 5 shows a list of potential beneficial mechanisms for transplanted MSCs in CNS injury, but it should be noted that these are not mutually exclusive, and it is likely that several factors play a role. While some results suggest the transdifferentiation of MSCs into cells of neural lineage, this is seen at low frequency in vivo, and is, therefore, unlikely to be the predominant beneficial effect of MSC transplantation. The frequency of cell fusion as a possible mechanism is also unclear. It is possible that differing mechanisms may predominate in brain versus spinal cord injury, and in traumatic versus ischemic injury. Indeed, there is significant controversy regarding the therapeutic benefit of MSCs in CNS injury. Discrepancies in the literature may relate to species differences (mouse versus rat versus human), or even intraspecies variation as shown by Neuhuber et al.,96 who reported that human MSCs show significant donor variation after transplantation into the injured rat spinal cord, with variable secretion patterns of growth factors and cytokines. Other differences may be due to culture conditions, the variety of injury models employed, methods of transplantation including number of cells injected, labeling methods and potential transfer to host cells as recently demonstrated with both bromodeoxyuridine (BrdU) and bis-benzamide,15 and strategies for assessing functional recovery.

Table 5 Potential mechanisms of action of transplanted adult bone marrow-derived MSCs in CNS injury
Full size table

There are several limitations to the current data set that require further investigation. For example, investigators use MSCs at different passages. This variable may, in part, be overcome in future studies that define the MSCs used according to internationally proposed characteristics, including immunophenotype.97 For human studies at least, it is also important to document the karyotype of the cultured cells, although to date, cytogenetic abnormalities among passaged human MSCs are admittedly rare. Another issue that remains to be addressed is the importance of cell dose in the transplanted inoculum and whether there is a safe upper limit. Some data suggest that better functional outcomes are dose dependent with one study that showed the benefit of multiple injections. This may help to explain some of the varied results from different studies, but needs to be explored more extensively.

Timing of transplantation is also a key issue, as evidence suggests that cell survival and functional outcome may be improved if cells are transplanted at least 1 week, but less than 14 days, after injury. The extent of cellular division after transplantation is also not known, and while it has generally been suggested to be very low, this finding is not supported by all studies. The best method of administering MSCs is being explored, and while direct administration results in the highest number of cells in the injury site, intra-lesional injection may result in further local damage and is invasive. Of note, cells appear to have improved survival when injected next to the lesion rather than directly into the injured area.

Homing mechanisms to attract MSCs to the area of injury when injecting i.v. or into the cerebrospinal fluid remain to be established but this approach is unlikely to be as effective as direct administration because many cells are distributed widely throughout the body with unknown effects. Our own data suggest that in general, the i.v. route is not optimal for treating local lesions, at least in mouse models, because of prolonged retention in pulmonary capillaries and subsequent loss. Intra-arterial administration of MSCs may cause further ischemia or thrombosis, and has not shown added benefit. Further limitations to the data set include the wide range of animal models used, with a variety of injury types. Rat models of CNS injury have the advantage of providing more highly validated tests of functional improvement than mice and are less costly than primates, enabling large numbers to be investigated. The type of injury is also an important factor. For example, a contusion spinal cord injury is more applicable to the human condition than a cord transection, but is a more difficult model with which to distinguish regeneration from protection of spared fibers and to uncover the underlying mechanisms. The potential benefits of the coadministration of ‘neural factors’ are also not fully appreciated, as these may improve survival of the MSCs, influence their fate or act in synergy to affect the target.

Given these limitations, we therefore recommend further well-designed preclinical studies with appropriate animal models before embarking on new clinical trials in humans. These studies should build on current knowledge regarding optimization of basic transplantation techniques. Such models should be able to distinguish among the contributions to tissue repair mediated by transdifferentiated or fused cells, paracrine effects such as growth factor elaboration, the induction of endogenous progenitor cells, the inhibition of apoptosis, and possibly others. These studies should provide better guidance for the design of clinical protocols for humans. As with all regenerative cell therapy studies, careful documentation of potential toxicities of MSC administration should continue, especially for long-term effects, including tumor formation, ectopic activity and unwanted fibrosis. While promising, much work is needed before MSC transplantation will translate into tangible benefits for humans with CNS injury.