Progenitor cell therapies for traumatic brain injury: barriers and opportunities in translation

Approximately 1.5 million people suffer yearly from traumatic brain injury (TBI) in the USA. The annual mortality figure is around 50,000 patients, with the remainder of victims affected by varying levels of long-term sequelae (Thurman et al., 1999). Overall, 6.5 million patients are burdened by the physical, cognitive and psychosocial deficits associated with TBI (Consensus, 1999) and the total, annual economic impact is almost US $60 billion (Faul et al., 2007).

The deleterious effects of TBI occur during distinct primary and secondary periods. Primary injury, owing to mechanical force, results in shearing and compression of both neural and vascular tissue. An area of direct or primary injury and cellular necrosis is surrounded by a more diffuse area of neuronal injury (penumbra) (Moppett, 2007). Studies have shown that any further insult occurring during the secondary period is derived mainly from ischemia (Bouma et al., 1991) and inflammation (Israelsson et al., 2008). The acute care of TBI is still based on support, and focuses on controlling intracranial pressure while maintaining adequate cerebral perfusion. Long-term therapy consists of rehabilitation to improve cognitive and motor skills (Consensus, 1999). Despite aggressive clinical management, neurons have little ability for repair and there is no therapeutic modality available currently to reverse the injury on either a cellular or subcellular level. Based on preclinical research, adult stem cells might offer a potential therapeutic avenue to treat TBI.

Although preliminary evidence has identified a potential role for cellular therapy in the treatment of TBI, careful examination of the preclinical data also reveals barriers to the interpretation and translation of potential treatment options, as discussed in detail in this review, including: (1) therapeutic progenitor cell heterogeneity; (2) the variable efficacy/durability of cell labeling; (3) routes of delivery of cell therapeutics and implications for the mechanism of action; and (4) potential adverse events related to cell therapeutics.

By definition, all stem cells are capable of self-renewal and are multipotent as they can develop into a variety of cell lineages (Weiner, 2008). Whereas a number of investigators have focused on embryonic-stem-cell-based therapies, the Cox laboratory has not pursued this line of discovery for TBI owing to regulatory issues placed upon embryonic stem cells and their potential to create tumors in the post-TBI brain (Riess et al., 2007). Alternatively, the harvesting of adult stem cells is not strictly regulated and a growing body of literature suggests their efficacy in treatment protocols. Therefore, this review will focus on non-embryonic or adult stem/progenitor cells in the treatment of TBI.

Throughout the human body, adult stem cells are maintained in niches where careful regulation protects progenitor cell populations from depletion or overproliferation. Stem cell niches are anatomic microenvironments that regulate the role of the progenitor cells in tissue regeneration, maintenance and repair (Scadden, 2006). Adult stem cell population niches that are currently under investigation for potential therapeutic use in TBI include: the bone marrow mononuclear fraction, the brain subventricular zone, the umbilical cord blood mononuclear fraction and adipose tissue. The availability, multipotency and capability for self-renewal make adult stem cells theoretical prime candidates for the treatment of a wide array of human diseases. This paper will review some of the progress and challenges involved with the use of adult stem cell types for the treatment of TBI.

Mesenchymal stem cells (MSCs) were first isolated from bone marrow upon the observation that plastic adherent monolayer colonies formed after culture of marrow aspirate (Friedenstein et al., 1970). Further research has clearly shown the multipotency of MSCs and their capability for self-renewal (Caplan, 1991; Pittenger et al., 1999; Prockop, 1997). Classically, MSCs were defined by their ability to differentiate into chondrocytes, osteocytes and adipocytes. A number of terms are used to refer to MSCs including marrow stromal or stem cell, and mesenchymal stem cell. The ISCT (International Society for Cellular Therapy) has standardized MSCs as a multipotent, mesenchymal stromal cell, referring to plastic-adherent fibroblastoid cells from the bone marrow that are capable of multilineage connective tissue differentiation.

The capability to differentiate down multiple cell lineages and the ease of in vitro expansion make MSCs attractive therapeutic agents. MSCs have demonstrated potential therapeutic benefit in a wide array of diseases, including neurological injuries such as TBI and stroke; however, there have been numerous conflicting reports regarding engraftment and therapeutic efficiency. This might be related to the use of variable cell types, or to the effects of media, cell passage number/techniques or isolation methods. As a result, there has been a growing standardization of immunophenotyping methods, as well as differentiation capacity.

The ISCT has set the following minimal criteria for defining human MSCs: (1) MSCs must be plastic-adherent and spindle shaped when maintained in standard culture conditions. Fig. 1 shows cells from a typical heterogeneous bone marrow population that are characterized by both their plastic adherence and their spindle shape by the third passage. (2) All MSCs must express the surface molecules CD105, CD73 and CD90, and lack expression of CD45, CD34, CD14 or CD11b, CD79α and CD19. An example of this expression profile is shown in the accompanying figures (Figs 2 and 3), which emphasize how the cellular immunophenotype changes with cell passage and media conditions. In the panels shown in Fig. 2, cell surface markers change with each passage, up to the third passage, to become more homogeneous. Furthermore, media selection might have a substantial effect on the expression of certain markers (Fig. 3). (3) MSCs must be able to differentiate into osteoblasts, adipocytes and chondroblasts in vitro. Fig. 4 demonstrates the typical differentiation pathways needed to complete the cellular characterization of MSCs (Fig. 4) (Dominici et al., 2006).

A large amount of preclinical research using MSCs in models of TBI has shown the migration of stem cells to injury sites and the subsequent survival of MSCs and their differentiation into neurons and astrocytes, leading to improved motor function (Mahmood et al., 2001a; Mahmood et al., 2001b; Mahmood et al., 2003; Mahmood et al., 2005). Recent in vivo studies have shown that, 3 months after cortical impact injury, MSCs that are engrafted into the brain differentiate into neural cells, resulting in functional neurologic improvement (Mahmood et al., 2006; Lu et al., 2001). Despite the promising results of these initial studies, doubts remain regarding the ability of MSCs to transdifferentiate into neural cells, as well as the frequency of this transdifferentiation and the clinical significance of cell engraftment (Castro et al., 2002; Roybon et al., 2006; Wagers et al., 2002; English et al., 2006). Recent research has indicated that the potential therapeutic benefit of MSC therapy could be the result of the following mechanisms: the secretion of growth factors (Chen et al., 2002); the exchange of genes and proteins through cell-to-cell fusion or contact (Sprees et al., 2003); the induction of angiogenesis (Chen et al., 2003); and the effects on immune modulation (Aggarwal and Pittenger, 2005) (these are discussed later in the text).

Adult tissue stem cells are defined as undifferentiated cells, derived from the bone marrow, that are capable of self-renewal, differentiation and proliferation (Loeffler and Roeder, 2002). Compared with embryonic stem cells, adult tissue stem cells have a decreased ability to self-renew and normally differentiate through only one germ cell lineage (Verfaillie, 2005). However, specific adult stem cell cultures have been found to differentiate into all three germ cell lineages: visceral mesoderm, neuroectoderm and endoderm cell lines (Jiang et al., 2002). The observed multipotency is the result of a more primitive cell line termed multipotent adult progenitor cells (MAPCs). MAPCs have extensive proliferative ability and are likely to be precursors for adult tissue progenitor cells. MAPCs have been found to engraft in mice and can generate lung, liver, gut, vascular, endothelial and hematopoietic cells (Jiang et al., 2002; Serafini et al., 2007; Aranguren et al., 2008). MAPCs have also been shown to differentiate in vivo and in vitro. Markers for human MAPCs include CD10, CD13, CD49b, CD49d, CDw90 and Flk1 (Ortiz-Gonzalez et al., 2004).

The differentiation of transplanted MAPCs into neural cells demonstrates a possible therapeutic avenue for TBI treatment because such differentiation has been noted in vitro (Jiang et al., 2003). In addition, MAPCs that were derived from murine bone marrow and injected into mouse blastocysts were found to differentiate into neural and glial phenotypes, which then showed normal development (Keene et al., 2003). The small amount of preclinical research carried out using MAPC therapy has shown a potential therapeutic benefit; however, the translation of MAPC therapeutics into clinical trials will require a larger body of both in vitro and in vivo work.

Adipose-derived stem cells (ASCs) represent a promising source for large quantities of stem cells. There are several terms that have been used to describe these cells, including adipose-derived stem/stromal cells, adipose-derived adult stem cells, adipose-derived stromal cells, adipose stromal cells, processed lipoaspirate cells and adipose mesenchymal stem cells. The currently accepted term to describe these cells, as per the International Fat Applied Technology Society, is adipose-derived stem cells (Bunnell et al., 2008). ASCs can be harvested from liposuction waste tissue by collagenase digestion and differential centrifugation. The resultant stromal vascular fraction contains the adipocyte progenitor cells (Bunnell et al., 2008; Gimble and Guilak, 2003). ASCs have been shown to differentiate into adipocytes, osteoblasts, myoblasts, chondroblasts and neural cells, with maintenance of their characteristics through several cell passages. The age of the donor might affect the differentiation potential of these cells (Bunnell et al., 2008; Rodriguez et al., 2005).

Intravenous administration of ASCs after induced hemorrhagic stroke in a murine model demonstrated a decrease in inflammation and chronic brain degeneration and promoted long-term functional recovery (Kim et al., 2007). Additionally, Kang et al. demonstrated an improvement in motor function in a rat model of spinal cord injury after intravenous administration of ASCs, some of which differentiated into neurons and oligodendrocytes (Kang et al., 2006). Similarly, in a model of cerebral ischemia, transplantation of ASCs resulted in functional improvement without resultant inflammation or an immune response (Kang et al., 2003). Secondary to their relative ease of availability, ASCs represent an intriguing candidate for future research into cell therapeutics.

Classically, embryonic stem cells have been the most common source for neural stem cells (NSCs) (Weiner, 2008); however, recent research has shown that there are areas of active neurogenesis in the hippocampus and dentate gyrus of both the adult rat brain and the human brain (Altman and Das, 1965; Eriksson et al., 1998). NSCs have been isolated from adult murine striatum and have been shown to differentiate into both neurons and glial cell lines (Reynolds and Weiss, 1992). NSCs that have been isolated from the subventricular zone of the lateral ventricle of adult rats can be expanded in culture as neurospheres (Chen et al., 2007). NSCs can be further characterized by the immunocytochemical markers nestin, an intermediate filament (Hockfield and McKay, 1985), musashi 1, an RNA-binding protein (Sakakibara et al., 1996), and the transcription factors SOX1 and SOX2 (Pevny et al., 1998; Ellis et al., 2004) Nestin and musashi 1 are specific to neural progenitor cells and are not expressed in fully differentiated neurons. Although SOX proteins are not specific to NSCs, they can be used to differentiate further between neural progenitor cells and adult neurons (Wegner and Stolt, 2005).

Transplantation of NSCs into the cortex of injured rats showed that these cells can differentiate into neurons and produce an array of trophic factors, and might also improve cognitive functioning (Gao et al., 2006). Direct transplantation of NSCs after cortical impact injury has shown that between 1–3% of cells become engrafted within 2 weeks of treatment (Fig. 5). NSC engraftment was associated with an improvement in motor function, as indicated by the rats’ increased ability to run on a rotarod; however, no recovery of cognitive function was identified (Harting et al., 2008b). Work completed in the Roh laboratories investigated the possible anti-inflammatory effect of systemic NSC infusion in a rat intracerebral hemorrhage model. NSCs were transfused at 2 and 24 hours after injury and were compared with NSCs that were directly transplanted at the same times. The NSCs that were infused systemically during the hyperacute phase (2 hours) showed reduced initial neurologic deterioration, decreased brain edema formation, decreased inflammatory infiltration and lower levels of apoptosis when compared with the other therapeutic modalities. In addition, early infusion of NSCs also attenuated the activation of tumor necrosis factor alpha (TNF-α), interleukin 6 (IL-6), and the transcription factor NF-κB (Lee et al., 2008). Recent investigation of NSC seeding onto polymer scaffolds and their subsequent implantation into infarction cavities of mouse brains has shown an increase in neuronal differentiation and an elaboration of neural processes. It is postulated that the mesh provides a framework for neurite outgrowth, allowing improved interaction between the donor and host cells, which ultimately leads to connectivity and regeneration of cortical tissue (Park et al., 2002).

Umbilical cord blood is an abundant source of hematopoietic stem cells (HSCs); however, further investigation has revealed a subgroup of cells in cord blood that do not express the hematopoietic cell marker CD45 and that do not differentiate into hematopoietic cells in vitro. This mononuclear cell fraction can be expanded in culture and, when stimulated with basic fibroblast growth factor (bFGF) and human epidermal growth factor (hEGF), has been shown to differentiate into cells that are positive for the neural markers beta-tubulin III and glial fibrillary acidic protein (GFAP) (Bicknese et al., 2002). Attempts at isolating MSCs from umbilical cord blood have been inefficient, with approximately a 10% yield, however, multiple studies have shown that up to 100% efficiency can be achieved when harvesting MSCs from cord stroma (Secco et al., 2008; Lu et al., 2006). Therefore, umbilical cord stroma or placental tissue, rather than cord blood, offers a potentially non-invasive and readily available source for large amounts of human MSCs. When used as a mononuclear cell fraction, these cells have demonstrated marked anti-inflammatory effects making them a potentially attractive therapeutic agent, even without the isolation of MSCs (Vendrame et al., 2006).

Transplantation of the human umbilical cord blood (HUCB) mononuclear fraction into the subventricular zone of neonatal rats resulted in up to 20% engraftment of GFAP- and class III beta-tubulin (TUJ1)-positive human cells by 1 month after transplantation (Zigova et al., 2002). In addition, the Chopp laboratory infused HUCB into the tail veins of rats 24 hours after cortical impact injury. The specimens were followed with serial rotarod testing and neurologic severity scoring to assess neurological functioning. The results showed reduced motor and neurological deficits, as well as engraftment of cells that were positive for GFAP and other neuronal markers (Lu et al., 2002). Despite the evidence of neuronal differentiation, more recent studies have failed to show engraftment of HUCB after intravenous infusion in a rat middle cerebral artery occlusion model (Borlongan et al., 2004).

To evaluate alternative therapeutic mechanisms, CD34-positive cells that had been isolated from HUCB were infused into immunocompromised rats using the middle cerebral artery occlusion (MCAO) model, 48 hours after injury. Endogenous neurogenesis and neovascularization were found to be enhanced at the border of the ischemic zone. When antiangiogenic agents were infused to suppress neovascularization, neurogenesis was impaired suggesting that HUBC has an angiogenic effect that supports endogenous neurogenesis (Taguchi et al., 2004).

Although previous in vivo research has shown a therapeutic benefit from stem cell transplantation, the exact mechanism of action has yet to be delineated. Tissue-specific stem cells such as MSCs and NSCs have the ability to proliferate into loco-regional progenitor cells that promote tissue-specific regeneration (Caplan, 1991; Pittenger et al., 1999; Prockop, 1997; Kaiser et al., 2007; Dominici et al., 2006). Additional studies investigating MSC cell therapy have shown the differentiation of MSCs into multiple cell lineages including cardiomyocytes (Kawada et al., 2004), hepatocytes (Ling et al., 2008), keritinocytes (Sasaki et al., 2008) and neurons (Mahmood et al., 2006; Lu et al., 2001). The concept of stem cell plasticity or transdifferentiation would explain the capacity of adult tissue stem cells to proliferate into different cell lineages (i.e. engraftment of neurons after MSC therapy). However, doubts still remain about the frequency and clinical relevance of transdifferentiation (Castro et al., 2002; Roybon et al., 2006; Wagers et al., 2002; English et al., 2006).

Stem cell therapy may also stabilize damaged or diseased cells via gene/protein transfer through cell-to-cell contact or fusion. Early investigation into the treatment of heat-shocked small airway epithelial cells with MSCs demonstrated a cellular infusion incidence of up to 1% (Sprees et al., 2003). Current research is being conducted into loading MSCs with the Duchenne’s muscular dystrophy (DMD) gene, which encodes dystrophin. The aim is to develop a model for DMD treatment through directed stem cell fusion and complementation with the deficient gene (Goncalves et al., 2008). Despite evidence supporting cellular fusion, additional research needs to be conducted to evaluate its clinical significance as a possible mechanism of action.

Additional investigation has been completed evaluating the secretion of growth factors from stem cells. MSCs cultured with the supernatant of homogenized ischemic brain tissue from a middle cerebral artery occlusion rat model showed an increase in brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF), vascular endothelial growth factor (VEGF), epidermal growth factor (EGF), fibroblast growth factor 2 (FGF2) and insulin-like growth factor 1 (IGF1) when compared with a sham model (Qu et al., 2007). MSCs that have been cultured with supernatant from the brains of rats that have experienced closed cortical impact injury also showed an increase in BDNF, NGF and VEGF, and an increase in hepatocyte growth factor (HGF) (Chen et al., 2002). The increase in growth factor production by MSCs could stimulate the resident tissue cells to repair themselves and might account for the functional benefit seen with stem cell therapy.

Stem cells are known to migrate towards sites of inflammation and mediate inflammatory markers, thereby reducing the amount of overall inflammation and edema. Co-culture of MSCs with purified immune cells such as natural killer cells, dendritic cells, and both naïve and effector T cells caused an increase in production of the anti-inflammatory interleukins IL-4 and IL-10, while decreasing the amount of TNF-α and IFN-γ. As seen in Fig. 6, the increase of IL-4 in accordance with the decrease in INF-γ promotes a shift in the helper T cell subsets, from cytotoxic Th1 cells to Th2 cells. In addition, a decrease in TNF-α together with an increase in IL-10 could lessen the maturation of dendritic cells, while increasing the number of regulatory T cells that promote an anti-inflammatory or tolerant response (Aggarwal and Pittenger, 2005). Additional studies have shown that treatment with MSCs can suppress the proliferation of T cells (Ramasamy et al., 2008). A decrease in both overall inflammation and the immune response could correlate with a possible decrease in the cell damage associated with stem cell therapy.

Recently, an investigation using a rat stroke model showed that, after ischemic accident, there was a decrease in splenic mass with a concomitant reduction in the number of CD8-positive T cells. Transfusion of systemic HUCB into rats that had experienced a stroke showed that MSCs homed towards the rat spleen, with stabilization of both splenic mass and CD8-positive T cell count. In addition, T cell proliferation was decreased in accordance with both an increase in IL-10 and a decrease in IFN-γ. The reduction in the peripheral inflammatory response was associated with up to an 85% decrease in cerebral infarct volume (Vendrame et al., 2006).

It has also been theorized that stem cell therapy induces angiogenesis, which accounts for tissue repair (Chen et al., 2003). In a rat model, intravenous injection of MSCs after induced ischemic stroke resulted in increased levels of endogenous VEGF and VEGF receptor 2, as well as increased angiogenesis in the transition zone (Caplan and Dennis, 2006). Further investigation has shown that combination therapy with MSCs and a nitric oxide donor can significantly increase blood vessel diameter and endothelial cell proliferation when compared with MSC therapy alone (Chen et al., 2004).

Promising preclinical data has led to the initiation of several clinical trials to study cell therapy in neurologic diseases. A search of the NIH clinical trials database showed that there are currently multiple trials investigating the use of peripheral stem cell therapy, either after or in conjunction with chemotherapy, for the treatment of central nervous system (CNS) neoplasms. (http://clinicaltrials.gov/ct2/results?term=stem+cells+and%2For+brain) Under current FDA guidelines, the research of all new cell therapeutics must be completed in accordance with an investigational new drug protocol (www.fda.gov). Following these guidelines, we have initiated a Phase I clinical trial using autologous, intravenous bone-marrow-derived mononuclear cells as an adjunctive treatment for children (ages 5–14) with isolated, severe acute TBI (defined as having a Glasgow coma score of 5–8). The study is currently ongoing and represents the sole clinical trial examining cellular therapeutics for TBI in the USA.

One limiting variable in the development of stem cell therapeutics is the identification of a delivery method that is both effective and minimally invasive. Increasing amounts of research are being conducted to evaluate multiple routes for delivery of stem cells, such as intravenous, intra-arterial and direct routes.

Intravenous administration offers easy access to the circulation, with the possibility of distribution throughout multiple tissues including the liver, lung, kidneys, spleen, bone marrow and mesenchymal tissues (Gao et al., 2001; Allers et al., 2004). Administration of green fluorescent protein (GFP)-labeled MSCs to irradiated baboons resulted in up to 2.7% engraftment of MSCs throughout many organ systems (Devine et al., 2003); this phenomenon has been demonstrated by several investigators using SPECT imaging (Chin et al., 2003; Lappalainen et al., 2008).

An initial drawback of intravenous application is the large proportion of first-pass pulmonary sequestration (Barbash et al., 2003; Tolar et al., 2006). Recent work has been completed in the Pelletier lab to estimate the diameter of pulmonary capillaries and MSCs. In these studies, MSCs were initially isolated, purified and then transfected with luciferase. Subsequently, 4, 10 and 15 μm fluorescence-labeled microspheres were first compared to MSCs to estimate the diameter of these cells, and then infused with or without sodium nitroprusside^† treatment. Analysis of lung tissue, which was harvested 30 seconds after infusion, showed no evidence of the 4 μm microspheres in either treatment group. However, a significant decrease in the number of 10 and 15 μm microspheres was noted in the group treated with sodium nitroprusside. The study estimated that MSCs have a diameter of 15–19 μm, whereas the pulmonary capillary diameter ranges from approximately 5–9 μm, and showed a decrease in the first-pass pulmonary effect observed with sodium nitroprusside pretreatment (Schrepfer et al., 2007). Fig. 7 depicts the primary effects and mechanisms of intravenous stem cell therapy for the treatment of traumatic or ischemic brain injury.

Two of the major issues in human MSC therapeutics are cell engraftment and the factors that affect MSC homing. The specific mechanisms are still under investigation but it is believed that, in the bone marrow, CXCR-4 (chemokine receptor 4) and SDF-1 (stromal-cell-derived factor 1 ligand) play a key role in homing and colonization of the bone marrow. The importance of these two cell adhesion molecules has been demonstrated in specific knockout models, where knockout of either CXCR-4 or SDF-1 results in bone marrow failure. Following multiple (more than two) passages of stem cells, there is a marked loss in CXCR-4 receptor sites (Potapova et al., 2008). Additionally, CXCR-4 upregulation in cytokine-stimulated rat MSCs resulted in a notable increase in stem cell engraftment (Shi et al., 2007). Upregulation of SDF-1α is another strategy that has been used to increase stem cell engraftment. After transplanting muscle cells, some of which had been transfected with a plasmid encoding SDF-1α, into infracted heart muscle, an increased number of stem cells were observed in the group with the SDF-1α transfection. Additionally, this study identified an improvement in angiogenesis and left ventricular function (Elmadbouh et al., 2007). Research into the role of chemokine receptors and ligands in stem cell homing is in its infancy. The work completed on CXCR-4 and SDF-1 suggests that evaluating the effectiveness of alternate receptors/ligands could generate exciting new results that could increase the efficacy of stem cell delivery.

Intra-arterial administration offers a method to further localize the placement of stem cells, whilst bypassing the high pulmonary first-pass effect. Treatment of ischemic stroke in a murine model by intracarotid transfusion of MSCs resulted in engraftment of 21% of these cells in the middle cerebral artery distribution, with functional neurologic improvement (Li et al., 2001). However, recent investigation into cerebral blood flow, following intracarotid infusion of MSCs immediately after induced ischemic stroke, showed a great deal of variability in cell delivery. In 37% of study specimens, there was a significant (80–90%) decrease in cerebral blood flow, which was associated with rapid death (67% of cases) (Walczak et al., 2008). Therefore, the threat of cerebral artery emboli/occlusion might limit the role of intra-arterial MSC administration in cell therapeutics. Fig. 8 displays possible mechanisms of therapeutic effect from intracarotid stem cell transfusion.

Direct or intracerebral implantation of stem cells would maximize the stem cell load at the site of disease/injury. Investigation into the intracerebral placement of stem cells using a murine TBI model showed engraftment with stem cell migration to the sight of injury (Mahmood et al., 2001b). Immunohistochemical analysis indicated a significant increase in progenitor cell proliferation in the boundary zone and subventricular zone of the lateral ventricle when compared with intravenous administration (Mahmood et al., 2004). Despite the increased localization of progenitor cell placement, investigators must consider the invasiveness of the intracerebral approach and the possibility of further tissue damage during cell transplantation.

The ideal stem cell delivery vehicle would combine the accessibility of intravenous administration with the ability to focus progenitor cell proliferation, as provided by intracerebral/direct transplantation. Further investigation is needed with regards to the following: (1) methods to decrease the first-pass pulmonary effect; (2) safety of intra-arterial infusion; (3) the difference in neurologic function, secondary to cell proliferation, associated with intracerebral transplantation, as compared with intravenous infusion; and (4) the benefit from modulation of the inflammatory response and the need for direct cell placement at the site of injury.

Stem cell therapy is being evaluated as a possible therapeutic agent in many animal diseases. It is also being used in a variety of clinical studies including those investigating cardiomyopathy, diabetic disease, systemic lupus erythematous, pancreatic disease, liver disease, stroke and TBI. In studies involving stem cell transplants in different diseases, identifying the location, migration and biodistribution of transplanted cells, and distinguishing them from endogenous cells, is the key to elucidating their underlying mechanism of action. Moreover, different tracing techniques have been applied in different studies to assess the sensitivity, dynamic range, convenience and reliability of MSC assays. Therefore, we will review the different tracing techniques and their applications in stem cell transplantation, including both experimental studies and preclinical trials (Yan et al., 2007; Harting et al., 2008c). We also discuss some of the pitfalls and limitations of the specific labeling techniques (Table 1).

The method of labeling CNS-destined cells with thymidine analogs (bromodeoxyuridine [BrdU], chlorodeoxyuridine [CldU], iododeoxyuridine [IdU], titrated thymidine) has been used for many years. During the S phase of cell division, these analogs are incorporated into the DNA of the target cell (Vega and Peterson, 2005; Burns et al., 2006).

Recent studies suggest that caution must be exercised when using thymidine analogs as a marker for stem cell transplantation. Burns et al. (Burns et al., 2006) demonstrated that, after transplantation of stem cells, the surrounding, dividing, non-transplanted neural cells appeared to take up thymidine. This was suspected because large numbers of labeled cells could be seen after transplantation with labeled dead cells. The mechanisms for this uptake by surrounding, non-transplanted cells might involve apoptosis of transplanted cells and/or the uptake of any thymidine analogs that have been released by transplanted cells (Burns et al., 2006). Additionally, continued proliferation of cells can dilute the nuclear signal of the thymidine analog, this is most commonly observed with BrdU (Cooper-Kuhn and Kuhn, 2002).

Thiol- and amine-reactive tracers [commercially available as CellTracker and CellTrace (Invitrogen)] represent another labeling technique for identifying stem cells. These tracers diffuse into the target cells, where cleavage by certain esterases results in a fluorescent product, allowing detection of the target cells. Cells that contain CellTracker usually retain the ability to be detected for several divisions, whereas cells with CellTrace may be detected for much longer (Invitrogen, 2008). Advantages of this method include its ease of use, rapid labeling and bright fluorescence.

Carbocyanine compounds [Dil, DiA (Di-Asp), DiO, CMDil, PKH26], which are lipophilic and bind to cell membranes, can be used to track cells in vivo. The carbocyanine membrane (CM) dye Dil is associated with a 45–70% toxicity level, therefore only 3–4% of proliferating cells can be visualized after 35 days. PKH26 has lower levels of toxicity and cannot be transferred from labeled to non-labeled cells, and PKH26-stained cells have been detected, in vivo, up to 60 days after transplantation (Yan et al., 2007). Although the ease of labeling and in vivo detection make membrane dyes an attractive labeling technique, possible drawbacks to this approach include decreased detection with increasing rounds of mitotic division and the need for rapid analysis of cell labeling (Ferrari et al., 2001).

Nanocrystals provide another method of cell labeling. Many types of nanocrystals are available for cell labeling; the nanocrystals that are most commonly used for stem cell labeling are of the semiconductor type (quantum dots) [commercially available as Qtracker (Invitrogen)], in which tracking is provided by emission of certain wavelengths after excitation by light. There are many potential mechanisms for cellular uptake of nanocrystals, including endocytosis, microinjection and peptide-based reagents. Nanocrystals do not seem to affect the properties of stem cells, including their capacity to differentiate; however, detection does decrease with each round of mitotic cell division. Other potential mechanisms of decreased detection include instability of the particles and their uptake by surrounding cells. This might affect the usefulness of nanocrystals for long-term cell tracking. However, some studies have demonstrated that cells labeled with quantum dots may be detected at up to 22 days.

A long in vivo detection time is advantageous, but a disadvantage arises in discriminating individual cells from the native tissue background, which can be difficult. Additionally, a major disadvantage of nanocrystals is the potential for cytotoxicity, which may be mediated by several mechanisms such as free radical formation, the release of free cadmium/gadolinium (secondary to degradation) or through certain intracellular interactions/reactions (Hardman, 2006; Lin et al., 2007; Shah et al., 2006; Shah et al., 2007).

Other nanocrystal-tracking methods are available; most of them exploit the use of metal ions (such as gadolinium, iron, manganese or superparamagnetic iron oxide), which can then be detected by cellular magnetic resonance imaging. Other potential mechanisms for cell labeling include the use of microbeads, where material is attached to an antibody that is specific for certain cell surface proteins, or surface coating with substances, i.e. dextran, polymers that render these particles susceptible to endocytosis by the target cells (Sykova and Jendelova, 2007).

Another approach to cell labeling involves the transfection of cells with specific vectors that have specific properties enabling the detection of cells. Most commonly, foreign material destined for target cells is transfected by specific viruses or plasmids. Issues that must be taken into account include the immunogenicity of the vector, the affect on cell properties, methods for detection and the possibility of toxicity. GFP and LacZ are the two most common vectors used in animal studies. One of the challenges with this method involves ensuring that target cells express the transfected genetic material. Transduction with viruses may also risk tumor induction. Nucleofection, the transfer of genetic material directly into the nucleus by plasmids and without the use of a viral intermediary, seems to be the most promising approach, with a highly efficient rate of transfer and minimal effects on the properties of target cells (Lu et al., 2003; Wiehe et al., 2005; Lakshmipathy et al., 2004; Aluigi et al., 2006).

Stem cells isolated from transgenic animals are another potential way to track transplanted cells. Several studies have used GFP-tagged transgenic animals as a method for cell labeling and tracking (Okabe et al., 1997; Brazelton and Blau, 2005). However, work in our laboratory has demonstrated that this approach can be fraught with quandaries. The expansion of MSCs results in decreased expression of GFP. Detection of these cells by microscopy can be difficult and the use of a GFP antibody may be needed to improve fluorescence and, additionally, increase the number of GFP-positive cells that can be detected (Harting et al., 2008c).

Fig. 9 shows that expansion of MSCs through multiple passaging of cells results in the loss of the GFP label, with approximately 50% of cells remaining GFP-positive after in vitro cell culture. The two major epigenetic modifications that control gene expression in this system are histone acetylation/deacetylation and DNA methylation. To explore the relative contributions of each factor, MSCs were sorted into GFP-positive cells, GFP-negative cells, and the normal GFP wild type (consisting of a 50:50 mixture of GFP-positive and GFP-negative cells). In culture, the cells were exposed to either 5-azacytidine, a DNA demethylator, or trichostatin A, a histone deacetylase inhibitor. Following exposure to trichostatin A, histone acetylation increased, but GFP expression was not altered. In contrast, GFP expression increased following incubation with 5-azacytidine (demethylation), as shown in Fig. 10.

The conclusions drawn from this data illustrate how the technical issues associated with cell labeling can impact the interpretation of cell delivery and engraftment data. Specifically, unless GFP expression is monitored in MSCs that are used for transplantation, there is a risk of significantly underestimating the number of cells present, i.e. if GFP fluorescence is used as a guide. Further, the regional cell microenvironment (proinflammatory cytokines, reactive oxygen species, etc.) may affect labeled gene expression through epigenetic modifications.

The ideal labeling technique is one that: (1) is non-invasive (after the initial labeling); (2) allows cells to be tracked for a long period of time; (3) is stable and not transferred to the surrounding cells; and (4) does not genotypically or phenotypically affect the labeled cell.

Cell labeling for biodistribution in humans is important because it enables transplanted stem cells to be tracked; certain properties are necessary, including the ability to track cells non-invasively. The best currently available method of cell labeling is MRI, which is similar to the use of nanocrystals (discussed above). MRI is a non-invasive method that can be used to track labeled transplanted cells. The cells are labeled with paramagnetic and supermagnetic iron oxide particles that are then detected by MRI. Iron-based particles are preferred because they are more likely to disturb the surrounding magnetic field; consequently, it is easier to detect a smaller number of particles.

Preclinical studies using MRI have demonstrated promise. After transplantation of cells from a Maudsley hippocampal neural stem cell line, labeled with a gadolinium-based contrast agent, the labeled cells could be detected by MRI up to one month later (Modo et al., 2002a; Modo et al., 2002b; Modo et al., 2008). Studies have also demonstrated the effectiveness of detecting a variety of labeled cells, including NSCs, MSCs and ASCs (Rice et al., 2007; Kim et al., 2008).

Further studies are still required to assess the durability, length of action, long-term toxicity and the effect on function and degradation of these particles by the human body (Bulte et al., 2002). In a 1-year study of NSCs in a rat stroke model, transplantation of NSCs labeled with a gadolinium-based contrast agent resulted in no behavioral improvement when compared with control animals, and an increase in lesion size. In contrast, rats that received NSCs labeled with the red fluorescent dye PKH26 showed behavioral improvement with a 35% reduction in lesion size. The increase in lesion size might have arisen from the decreased survival of the labeled cells, possibly because of degradation/disintegration of the label leading to release of gadolinium (Modo et al., 2008). It is also important to determine whether the label remains in the progenitor cell and that it is not transferred to a macrophage or a surrounding cell. Although this can be determined histologically, this is not a useful technique for non-invasive imaging. Other issues include distinguishing the particles on the MRI from changes in the surrounding tissue, such as those caused by swelling or inflammatory responses. The studies evaluating human stem cell labeling with MRI-compatible particles are mostly preclinical and have shown great promise. There are several human studies that have also demonstrated the efficacy of this technique of cell labeling (Zhu et al., 2006; Kraitchman et al., 2008).

Using a Y chromosome marker is an efficient method of tracking cells and is also a relatively simple technique. Bone marrow or stem cells from male donors are transplanted into female recipients and are subsequently detected by fluorescent in situ hybridization (FISH) (Yan et al., 2007). Previous analysis of neural tissue from human female bone marrow recipients of male donors has revealed neurons and astrocytes labeled with the Y chromosome in the neocortex, hippocampus, striatum and cerebellum (Crain et al., 2005). Despite the high stability and specificity of Y chromosome markers, their use is limited in autologous stem cell transplantation because of the need for sex-matched studies (Yan et al., 2007).

Another potential imaging strategy is the use of SPECT imaging. In a middle cerebral artery occlusion rat model of stroke, detection of transplanted ¹¹¹In-oxine-labeled embryonic stem cells by SPECT imaging was possible both immediately after and at 24 hours after transplantation. When administered by intravenous infusion, ¹¹¹In-oxine-labeled neural progenitor cells were found in the internal organs but not the brain (Lappalainen et al., 2008). In addition, this method has shown efficacy following transplantation of labeled human umbilical cord blood stem cells and MSCs. The majority of studies have demonstrated the potential role of SPECT imaging in stem cell transplantation, particularly in the realm of in vivo cell distribution (Chin et al., 2003; Makinen et al., 2006; Lappalainen et al., 2008).

An important, although controversial, issue surrounding stem cell transplantation is the level of immunogenicity, or lack of, shown after transplantation. In one model, neural progenitor cells were implanted below the kidney cortex, which is not a classical site for immune privileged cells, to study the resultant immune response. After a period of approximately 4 weeks, they were found to develop along their expected pathway without evidence of immune reaction. It is interesting to note that when neonatal cerebellar tissue was implanted, a significant immune response was noted at 7 days. However, when the animal was sensitized with spleen cells before transplantation, the progenitor cells were not able to become established and grow, suggesting that antigens regulating immune response are present on these cells (Hori et al., 2003).

It has been demonstrated that MSCs possess major histocompatibility complex (MHC) class I and class II antigens, which are upregulated upon exposure to interferon γ (IFN-γ). The co-stimulatory molecules CD80, CD86, CD40 and the CD40 ligand do not appear to be present and are not expressed, even after stimulation with IFN-γ. T cells do not seem to attack MSCs when both are cultured together; further, no immune response was seen after osteogenic differentiation of the MSCs. It appears that MSCs have an immunomodulatory property, which might involve altering levels of IL-10, IFN-γ and TNF-α, because after co-culture with MSCs, the T cells still react against peripheral blood mononuclear cells (Klyushnenkova et al., 2005). It is interesting to note that there appears to be upregulation of certain immunological markers (MHC class 1 and class 2 antigens, CD45 and CD11b) in areas around ischemic lesions. Transplantation of MHP36 cells (an immortalized murine neuroepithelial stem cell line) does not seem to increase the immune response (by histological evaluation) in the immediate 2-week post-transplantion period; further, administration of cyclosporine does not improve the outcome with regard to cell survival (Modo et al., 2002a; Modo et al., 2002b). In addition, certain cytokines, such as TNF-α and IL-6, have been shown to variably regulate the expression of MHC antigens on human, rat and monkey NSCs (Johansson et al., 2008).

Other studies have also demonstrated a lack of an immune response against MSCs, even after differentiation. However, an immune response may occur after transplantation of differentiated cells (Liu et al., 2006). Additionally, an MHC-incompatible stem cell transplant may provoke immune rejection, which indicates that MSCs might not be immune privileged (Uccelli et al., 2006; Eliopoulos et al., 2005).

Although the promise of stem cell therapy entices many, the side-effect profile of this therapy in humans remains largely unknown owing to a paucity of clinical trials. However, preclinical studies offer a valuable insight. Additionally, a significant amount of literature on this topic is generated from HSC transplantation for hematologic malignancies.

Rubio et al. first demonstrated the spontaneous transformation of human stem cells after prolonged (i.e. 4-5 months) culture in vitro (Rubio et al., 2005). This has been reproduced in other stem cell populations, including MSCs from a murine model, which have been shown to developing into cells that are capable of becoming fibrosarcomas (Miura et al., 2006). Age might be a factor that contributes to the tumor potential of stem cells (Shi et al., 2007).

In addition, Lyle et al. demonstrated that, in two out of ten animals studied, a benign chondroma developed in relation to the needle tract that was used for the injection of embryonic stem cells (in the TBI model) (Li et al., 2007). Development of osteosarcoma like lesions in the lung, as well as sarcoma in the extremities, has also been noted after transplantation of MSCs in mice (Aguilar et al., 2007; Tolar et al., 2007).

Large retrospective analyses have demonstrated that there is an increased risk for the development of post-transplant malignancies. The most common malignancies include basal cell carcinoma, myelodysplastic syndrome, acute myeloid leukemia and post-transplant lymphoproliferative disorder. However, many other malignancies have developed, including melanoma, sarcoma and neuroblastoma (Darabi et al., 2005).

Emboli have been reported during stem cell transplantation. A case report was published discussing paradoxical embolus to the brain following human stem cell transplantation (the patient had a patent foramen ovale) (Peters et al., 2005). In addition, pulmonary cytolytic thrombi (pathologically described as thrombi with leukocytes) have been reported following allogenic HSC transplantation (Morales et al., 2003; Kounami et al., 2003), and pulmonary thromboembolism has also been reported (Baker et al., 2003). Further, increased stem cell engraftment has been associated with a decrease in cerebral blood flow in a model analyzing intracarotid therapy for the treatment of ischemic brain injury (Walczak et al., 2008).

Previous work has shown the benefit of cell therapeutics for the treatment of TBI in animal models. Multiple stem cell types have been used, however, investigation to quantify the relative cellular efficacy has yet to be completed. In addition, further research is needed to evaluate both the therapeutic significance of trophic factor secretion from locally transplanted cells and the role of systemic inflammatory modulation. Intravenous infusion is the most attractive delivery vehicle, but methods to decrease the first-pass pulmonary effect and thus increase both peripheral and injury-site-specific cell engraftment need to be developed. In addition, the development of experimental or clinical protocols is dependent upon the ability to track cells both in vitro and in vivo, therefore, the investigation of cell-labeling techniques that can be both accurately and non-invasively monitored, without changing cellular properties, will become a cornerstone of future research.

Much progress has been made with adult stem cell therapy and TBI in the preclinical arena. Our understanding of the origin, potential mechanisms of action, and immune-related issues surrounding these cells have all improved, in addition to improvements in our methods of tracking transplanted cells. However, a large amount of work remains, especially translation of the information into clinical trials so that the safety and efficacy of these therapies can be evaluated.

We would like to thank Lillian S. Kao for her efforts in reviewing the manuscript. Supported by grants: NIH T32 GM 08 79201, NIH P018/N01 HB 37163, M01 RR 02558 and by the Texas Higher Education Coordinating Board.