Increasing clinical interest exists in the use of transplantable stem cells as a means of repairing neurodegenerative disease or for gene delivery. The success of such approaches will be dependent on not only the source of donor stem cells and their expansion in vitro
, but also their mode of delivery, where localized surgical delivery to the brain carries inherent risks of morbidity. Studies have begun to evaluate the clinical efficacy of using intravenously administered bone marrow stem cells in diseases such as spinal cord injury [18
]. This study demonstrates that MRI is a useful technique that allows longitudinal noninvasive monitoring of stem cell migration in the brains of individual animals, giving important information on cell dynamics and interactions with the host CNS.
Implanted eNCSCs or BMSCs were visible as a hypointense region in T2 MRI images, allowing them to be tracked as they migrated within the brain. Implanted cells remained visible at the lesion site by MRI for the duration of the experiment (up to 13 weeks). Stem cells implanted directly into the brain were found to migrate preferentially along the major white matter tracts in both healthy and lesioned animals. In animals with a focal inflammatory lesion, the transplanted cells displayed unidirectional pathotropic migration to the lesion, where they ceased migrating and remained visible on MRI for several months. Moreover, and of potential benefit for clinical applications, both cell types similarly displayed migration to a focal inflammatory brain lesion from the peripheral vasculature.
Several studies have demonstrated the migration of stem cells after intravenous administration in models of neurodegeneration [19
]. However, those models involved cerebral ischemia and compression injury, which result in a more severe lesion. Our study has shown that relatively minor, focal lesions provide sufficient cues to attract stem cells from the peripheral vasculature. Our LPS-lesion model induces a focal demyelination, resulting in an influx of microglia; such lesions have been shown to stimulate the release of cytokines such as IL-1β, TNF, and IL-6 [23
], which may have provided the chemoattractant signals for eNCSCs and BMSCs. The migration of transplanted stem cells through the brain parenchyma may also be aided by their secretion of matrix metalloproteases (MMPs), because BMSCs require MMP1 to migrate toward human gliomas [24
]. Because no donor cells could be detected in other regions of the brain, this suggests that the cells were not simply trapped within the cerebral blood vessels, but were responding specifically to chemotactic factors released from the lesion. It has been reported that eNCSCs are not migratory when transplanted into the lesioned spinal cord [25
], whereas our work shows that eNCSCs have homing properties similar to those of BMSCs and invade brain lesion from the peripheral vasculature. This suggests that the local CNS environment may have an important effect in regulating NCSC attraction and invasion, possibly by altering the expression of NCSC chemokine receptors or proteases essential for cell migration.
Although the intravenous administration route has obvious clinical advantages, we found that one challenge of using this route will be to get sufficient cells to the lesion for therapeutic benefit. Thus, we estimate that only about 8% of the intravenously injected stem cells migrated to the LPS lesion. By contrast, more than 10 times this number of stem cells reached the lesion site after intracranial administration some 3 mm from the lesion site. It will be of interest to determine whether other vascular routes, such as the carotid artery, might increase the proportion of injected stem cells that enter the brain-lesion site. It will also be of interest to examine whether stem-cell engraftment can be maximized by changing the postlesion timing of cell delivery. Any delay in delivery, sufficient to permit the in vitro expansion of autologous stem cells, would obviously be desirable from a practical and clinical perspective.
Concerning the fate of the administered stem cells, we found that BMSCs and eNCSCs were able to adopt glial fates in the brain, although we did not detect stem cell-derived neurons. A previous study demonstrated that BMSCs have the ability to differentiate into astrocyte-like and oligodendrocyte-like cells after implantation into an ischemic lesion [17
]. A separate study, by using eNCSCs in a in rat spinal cord crush lesion model, detected limited S-100 differentiation of implanted eNCSCs, indicative of mature Schwann cells [25
]. We did not detect S-100 differentiation of the implanted eNCSCs in our focal LPS-lesion model, although several implanted cells expressed GFAP, an early marker of glial differentiation for both astrocytes and immature Schwann cells [26
]. It remains unclear whether, had the experiment been allowed to proceed for a longer period, the implanted eNCSCs in our model would eventually mature into myelinating Schwann cells.
Previous studies within the group have shown that dead cells, prelabeled with IO-TAT-FITC, are cleared away from the local area and do not contribute to a signal void (see Supplementary data 1 in Additional file 1
). In addition, and consistent with this study, previous studies using live stem cells have shown that whereas a few of the cells are engulfed by IB4+
microglia, the majority remain nonphagocytosed after BMSC implantations into a rodent model of Parkinson disease [14