Remyelination of the Rat Spinal Cord by Transplantation of Identified Bone Marrow Stromal Cells

GFP expression in bone marrow cells from GFP-expressing mice

Bone marrow was examined from a transgenic mouse line with an enhanced GFP (EGFP) cDNA under the control of chicken β-actin promoter and cytomegalovirus enhancer (C57BL/6-TgN(ACTbEGFP)1Osb; The Jackson Laboratory). This model is suggested to have GFP expression in all tissues, except for erythrocytes and hair (Okabe et al., 1997). We found that in acutely isolated raw bone marrow from this strain that 19.6 ± 4.9% of cells were GFP positive and that, in the mononuclear layer separated on a density gradient, that 40.9 ± 9.8% of cells were GFP positive. In the adherent cell population of the bone marrow, i.e., stromal cells, 62.5 ± 5.0% were GFP positive at 7 d in culture, and, at 14 and 21 d in culture, 62.4 ± 12.2 and 66.78 ± 7.9%, respectively, were positive for GFP. This indicates that a significant proportion of cells in the stromal cell fraction express GFP in this model system. At 2 weeks in culture, the adherent cells increased in number from 0.42 × 10⁶ to 2.1 × 10⁶ cells/ml, indicating a near fivefold increase in the number of cultured stromal cells.

Figure 1A is a field of living adherent cells in culture obtained at 2 weeks showing a number of GFP-expressing cells. The overlay of the fluorescent image of Figure1A with a differential interference contrast image (Fig. 1B) shows that both GFP-positive and -negative cells are present in this field. A variety of cell morphologies was present at this time culture. These included flattened process-bearing cells, fusiform cells, and rounded cells. There was no relationship between GFP expression and cell morphology, i.e., GFP-positive and -negative cells were present across the cell population. Figure1C is a higher-power field of fixed cells showing a cluster of flattened, process-bearing GFP-positive stromal cells at 2 weeks in culture. Stromal cells isolated in this manner have been reported to express collagen type I, fibronectin, and CD44 (Zohar et al., 1997;Azizi et al., 1998; Conget and Minguell, 1999; Pittenger et al., 1999). Many of the GFP-expressing cells coexpressed GFP with collagen I (Fig.2A), fibronectin (Fig.2B), or CD44 (Fig. 2C). The isolation of plastic adherent cells plated from the mononuclear layer of bone marrow and the expression of the above three surface antigens places these cells as conventionally defined marrow stromal cells. Although we did not do triple-labeling experiments for these antigens, the large percentage of cells that coexpressed GFP with each of the antigens studied suggest that collagen I, fibronectin, and CD44 are coexpressed on the stromal cells.

Remyelination after transplantation of bone marrow stromal cells

In preparation for injection into the demyelinated spinal cord, the stromal cells were grown in culture for 2 weeks, trypsinized, washed, and suspended in NPBM at a cell concentration of ∼5.0 × 10³ cells/μl. Control injections were with NPBM alone to control for possible effects of the trophic factors and mitogens in the medium. Cells or control medium solution were injected into the center of the lesion (see Materials and Methods), and the immunosuppressed rats were prepared for histology at 3 weeks or forin vitro electrophysiological recording. Figure3A1 shows a low-power micrograph of a plastic-embedded section from the normal dorsal columns. A higher-power micrograph (Fig. 3A2) from this section shows the abundance of myelinated axons in the normal dorsal funiculus. In the EB-X-lesioned dorsal funiculus (Fig. 3A3), there is virtually no myelin; the field is composed of demyelinated axons, cellular debris, and phagocytic cells. It is important to note that, in this model system, virtually no endogenous repair occurs for up to 6–8 weeks after lesion induction (Blakemore and Crang, 1985,1989; Honmou et al., 1996). Three weeks after direct microinjection of bone marrow stromal cells into the demyelinated spinal cord, relatively extensive remyelination is observed (Fig. 3B). The low-power micrograph in Figure 3B1 shows the lesion site. Thedarker stained area in the central region of the dorsal funiculus is replete with remyelination profiles. The boxed areas (2 and 3) in Figure 3B1 are shown at higher power in Figure 3, B2 and B3, respectively. Notice the near complete remyelination in the center of the lesion (B2) and the partial remyelination at the lateral margin of the lesion (Fig. 3B3). The lighter stained areas on the lateral margins of the lesion zone showed poor myelination (Fig. 3B1), and the remyelination was most competent in the dorsoventral axis of the dorsal funiculus.

The electron micrographs in Figure4 show the structure of the remyelinated axons in greater detail. A lower-power field shows both demyelinated and remyelinated axons obtained at the edge of the repair zone (Fig.4A). Some of the myelinated profiles are associated with large nuclei characteristic of peripheral myelin (arrows). One can also see cells with large nuclei that have cytoplasmic regions surrounding axons but have not as yet formed myelin (asterisks). A higher magnification of a myelinated axon associated with a large nucleus (Fig. 4B) indicates that these myelinated axons are surrounded by a basement membrane (arrowheads). These morphologies suggest that, whereas some axons show stable myelin formation, others appear to be in the process of myelinating. Moreover, a significant number of remyelinated axons have morphological features characteristic of peripheral myelin, which include a large cytoplasmic and nuclear surround.

The low-power field of the dorsal funiculus in Figure5A from a transplanted rat shows an abundance of GFP fluorescence. A higher-power micrograph (Fig.5B) shows cellular elements with GFP fluorescence that may be myelinated axons. We parceled tissue into adjacent blocks for frozen and plastic-embedded sections. Figure 5, C and D, is from adjacent blocks of the same transplanted animal from a frozen and plastic section, respectively. In virtually all transplanted animals in which plastic-embedded sections revealed remyelination, adjacent frozen sections showed intense GFP fluorescence. In animals in which no or little remyelination was observed, GFP fluorescence was weak.

Figure 5E is a high-power light micrograph from a frozen section stained with hematoxylin–eosin (H–E) taken within the transplant zone. The arrows point to collapsed axon cylinders surrounded by expanded myelin, which is typical with this histological technique. Note the eosin-stained nuclei (asterisks) associated with some of the myelinated axons. Figure 5F is an unstained frozen section obtained from the same block as in E, showing a cluster of cells that are GFP positive. A field of remyelinated axons from a plastic-embedded section is shown in Figure 5G for comparison.

Immunostaining was also performed for MBP and P0 on sections from spinal cords that were transplanted with GFP-expressing stromal cells (Fig. 6). Plastic sections from adjacent blocks of the same spinal cords indicated an abundance of remyelination. Colocalization of both MBP (Fig.6A–C) and P0 (Fig. 6D–F) with GFP-expressing cells was observed in the transplant zone. Whereas MBP staining was very specific for myelin (both central and peripheral myelin on attached spinal roots), there was considerable MBP staining in the lesion site that was not colocalized with GFP cells. P0 staining was very specific for peripheral roots and cells within the transplant zone. The enhanced MBP staining in the lesion zone may reflect MBP epitope from central myelin debris within the lesion. These results, although not quantitative, indicate the colocalization of transplanted GFP-expressing stromal cells with myelin proteins.

Conduction velocity improves after transplantation of stromal cells

Electrophysiological studies were performed in vitro to examine the conduction properties of the axons (Fig.7). Field potential recordings of compound action potentials were recorded with a glass microelectrode from the dorsal columns of control, demyelinated, and bone marrow stromal cell-injected rats, respectively, at sequential longitudinal distances (Fig. 7A). The early negativity was indicative of the fastest conducting fiber group and is shown at three distances along the conduction trajectory for normal, demyelinated, and stromal cell-transplanted dorsal funiculi (Fig. 7B1–3). The demyelinated axons displayed considerable conduction slowing (0.89 ± 0.11 m/sec; n = 12) compared with control (10.76 ± 1.24 m/sec; n = 12). The transplant group had a conduction velocity of 6.03 ± 1.62 m/sec (n = 9), which was significantly faster than the demyelinated axons (Fig. 7C). These data indicate that at least a subpopulation of remyelinated axons in the bone marrow stromal cell-injected rats showed increased conduction velocity.

Donor cells as the source of myelin-forming cellsin vivo

The lesion model that we used combines focal X-irradiation of the lesion zone to delay endogenous repair in combination with microinjection of ethidium bromide, which kills glia within the lesion zone (Blakemore and Crang, 1985, 1989; Honmou et al., 1996). Although our lesion model is unique in that it represents an aglial zone and most naturally occurring demyelinating disorders are associated with gliosis, this system does provide an important and simple model system to study the interaction of transplanted precursor cells and demyelinated axons in vivo. Virtually no myelination is observed in the lesion zone for 6–8 weeks, at which time, some endogenous myelin repair begins. Our studies were performed at 3 weeks after lesion induction, which is well within the time at which persistent demyelination is observed in this model. However, one cannot completely rule out the possibility that the injected cells recruited or facilitated an endogenous repair mechanism. To address this issue, we used GFP-expressing donor mouse MSCs. The majority of the MSCs cultured from the GFP-expressing mouse strain were GFP positive. It is also important that, in our control experiments, we did not observe remyelination. There are a number of trophic and mitogenic agents in the control media that was injected without cells. This strongly indicates that the medium alone did not enhance or accelerate an endogenous repair mechanism.

MSCs in culture were immunopositive for CD44, collagen I, and fibronectin, which have been shown to be expressed on MSCs (Zohar et al., 1997; Azizi et al., 1998; Conget and Minguell, 1999; Pittenger et al., 1999). Although we did not triple label the cells, the large proportion of cells immunopositive for these three antibodies suggest colocalization. Our results indicate that, in the spinal cord of rats showing remyelination after injection of donor GFP-expressing MSCs, that GFP-expressing cellular profiles were observed in the remyelination zone and were associated with the myelinated axons. In experiments in which no remyelination was observed, some GFP-like fluorescence was observed, but the intensity was low and no cellular profiles associated with axons were observed. Although we cannot say conclusively that every myelinated axon was the result of the injected donor cells, which is a caveat for all such studies, the intense GFP fluorescence and the cellular morphology of the GFP profiles in areas of remyelination strongly suggest that the donor cells contributed to the remyelination.

Morphology of the remyelinated axons

The remyelinated axons were densely packed and tended to cluster in the area of cell injection. The lateral margins of the dorsal funiculus remained unmyelinated, suggesting that either cell injection number was relatively low or the cells had limited ability to migrate. However, remyelination was observed along the entire dorsoventral extent of the dorsal funiculus. Interestingly, many of the remyelinated profiles were associated with a large cytoplasmic and nuclear surround, which in turn was enshrouded by a basement membrane. These are hallmark characteristics of peripheral myelin (Berthold, 1978). Neural precursor cells derived from the subventricular zone of postnatal rat (Keirstead et al., 1999) and adult human (Akiyama et al., 2001) brain also can give rise to both central (oligodendrocyte-like) and peripheral (Schwann cell-like) myelin.

There was no obvious neuronal differentiation within the transplantation zone, although marrow cells do have the ability to differentiate into neurons and astrocytes when transplanted into other regions of brain (Azizi et al., 1998; Woodbury et al., 2000). It is not clear whether there are different populations of cells within these progenitor groups or whether regions in the host (transplant recipient) provide signals to direct these cells along different glial or neuronal lineages. It is also unclear as to why these cells develop an almost exclusive myelinating phenotype when transplanted into the demyelinated lesion. One possibility is that the demyelinated white matter in our lesion model is replete with bare axon membrane and phagocytic cells. The ethidium bromide treatment kills virtually all endogenous cells, including astrocytes and oligodendrocytes within the lesion zone, and no endogenous neurons are present in the dorsal funiculus. Axon membrane is known to provide important signals to promote myelin-forming cells to form myelin (Wood and Bunge, 1975; Salzer et al., 1980; Maurel and Salzer, 2000). It is possible that these local axonal signals in the demyelinated lesion direct the lineage of the transplanted MSCs to a myelinating phenotype.

Conduction velocity of the remyelinated axons

In vitro electrophysiological studies were performed to assess conduction velocity of the remyelinated axons. The results indicate that conduction velocity was significantly increased in the transplantation group compared with the demyelinated group. The velocity was not as fast as controls but was several times greater than the demyelinated axons. However, the increase in conduction velocity as studied over several millimeters of conduction suggest that functional and sequential internodal segments are deposited on at least a subset of the fibers. We limited this study to 3 weeks after transplantation. It will be important in the future to determine whether continued maturation of the myelinated axons is accompanied by greater improvement in conduction velocity. These results do indicate that at least a level of functional recovery of conduction is elicited by the MSCs transplantation-induced remyelination.

Oligodendrocytes are the cells that normally myelinate CNS axons, but peripheral myelin-forming cells, such as Schwann cells (Blakemore, 1977) and olfactory ensheathing cells (Franklin et al., 1996; Imaizumi et al., 1998), can myelinate CNS axons in vivo and restore near normal conduction properties (Honmou et al., 1996; Imaizumi et al., 1998). Peripheral myelin-forming cells may have the advantage, if used as a cell therapy in multiple sclerosis (MS) patients, of not having the antigenic properties of oligodendrocytes, which are targets of an immune response in MS patients. Harvesting sufficient numbers of Schwann cells from peripheral nerve biopsy and cell expansion is problematic. However, the development of human clonal neural precursor cells derived from either embryonic or adult CNS (Keirstead et al., 1999; Kukekov et al., 1999; Vescovi et al., 1999; Akiyama et al., 2001) may allow for an abundant source of myelin-forming cells to be used in transplantation studies. Brustle et al. (1999) have demonstrated that human embryonic stem cell-derived glial precursors can be used as a source of myelinating cells in the CNS. Advances in the cell biology of progenitor cells derived from embryonic, fetal, or adult CNS open the prospect of developing cell lines as a potential source of a cell therapy for demyelinating diseases.

Bone marrow stromal cells may offer advantages if developed for cell therapies because the cells are relatively easy to isolate from small aspirates of bone marrow that can be obtained under local anesthesia, and they could provide a vast source of autologous cells for reparative therapies. Thus, MSCs have the potential to provide an efficient and renewable source of cells for autologous transplantation studies for demyelinating diseases. The demonstration that bone marrow stromal cells can form functional myelin when transplanted into demyelinated CNS regions suggests a potentially important interaction between bone marrow cells and CNS white matter repair mechanisms.