"From Marrow to Brain: Expression of Neuronal Phenotypes in Adult Mice".

Timothy R. Brazelton, Fabio M. V. Rossi, Gilmor I. Keshet, and Helen M. Blau*

Department of Molecular Pharmacology, CCSR 4215, 269 Campus Drive, Stanford University, Stanford, CA 94305-5175, USA.
*   To whom correspondence should be addressed.
E-mail: hblau@stanford.edu

After intravascular delivery of genetically marked adult mouse bone marrow into lethally irradiated normal adult hosts, donor-derived cells expressing neuronal proteins (neuronal phenotypes) developed in the central nervous system. Flow cytometry revealed a population of donor-derived cells in the brain with characteristics distinct from bone marrow. Confocal microscopy of individual cells showed that hundreds of marrow-derived cells in brain sections expressed gene products typical of neurons (NeuN, 200-kilodalton neurofilament, and class III -tubulin) and were able to activate the transcription factor cAMP response element-binding protein (CREB). The generation of neuronal phenotypes in the adult brain 1 to 6 months after an adult bone marrow transplant demonstrates a remarkable plasticity of adult tissues with potential clinical applications.

Until recently, the fate of adult cells has been thought to be restricted to their tissues of origin. Cells are well known to be capable of replenishing damage in tissues in which they reside, such as blood, muscle, liver, and skin. However, the finding that adult cells could be reprogrammed to express genes typical of differentiated cell types representing all three lineages (mesoderm, endoderm, and ectoderm) when fused to cells in heterokaryons was quite unexpected (1-3). This degree of plasticity demonstrated that the differentiated state is reversible and requires continuous regulation to maintain the balance of factors present in a cell at any given time (4). The cloning of frogs (5) and later sheep (6) further showed that previously silent genes could be activated in adult nuclei. Although remarkable, these examples of plasticity all involved extensive experimental manipulations. More recently, findings have been made that suggest that stem cells can assume diverse fates under physiologic conditions. Both transformed and primary neural cells can give rise to a range of phenotypes typical of their site of implantation within the central nervous system (CNS) (7, 8). Bone marrow cells can yield not only all cells of the blood but also cells with a liver phenotype (9). Perhaps the greatest plasticity yet demonstrated is the "homing" of bone marrow-derived cells to damaged muscle in irradiated dystrophic mdx mice (10, 11). Muscle-derived and CNS-derived stem cell-like populations have also been reported to reconstitute the blood and rescue lethally irradiated mice (11, 12). Here we report that after lethal irradiation, bone marrow-derived cells administered by intravascular injection yielded cells that expressed genes specific to neurons (neuronal phenotypes) in the CNS. Moreover, both the sources and the recipients of these cells were adults.

To examine whether bone marrow-derived cells could give rise to cells in the brain, adult marrow was harvested from transgenic mice (8 to 10 weeks of age) that ubiquitously expressed enhanced green fluorescent protein (GFP) (13). GFP-expressing (GFP+) bone marrow was administered by tail vein injection (6 × 10⁶ nucleated cells per recipient) into lethally irradiated, isogenic adult (8- to 10-week-old) recipients (14). Brains harvested several months after the transplant (post-transplant) and examined by light microscopy revealed the presence of GFP+ cells throughout the CNS, including the olfactory bulb, hippocampus, cortical areas, and cerebellum. No cells expressed detectable GFP in five age-matched bone marrow recipients (controls) transplanted with marrow that was not genetically engineered to express GFP. Thus, bone marrow-derived GFP+ cells were clearly present in the brain.

A comparison by flow cytometry of GFP+ cells derived from the bone marrow and the brains of transplanted animals revealed substantial differences (15). Isolated adult brains from transplanted mice were minced and treated with proteases as previously described (16) in order to obtain a single-cell suspension. Dissociated brain and bone marrow cells from the same mouse were stained with antibodies to the surface marker CD45 (15), which identifies all nucleated mature blood lineages (17), and to CD11b, which is expressed by all myelo-monocytic cells (18), including microglial cells in the CNS (19) (Fig. 1). Flow cytometry of whole bone marrow preparations obtained from recipients demonstrated that essentially all of the GFP+ cells that engrafted also expressed CD45 (Fig. 1, A and B). By contrast, analysis of entire dissociated brains indicated that a significant subset (up to 20%) of the GFP+ cells present in the CNS lacked both CD45 and CD11b (Fig. 1, A and C). These findings suggested that exposure to the CNS environment may have led a subpopulation of bone marrow-derived cells to acquire novel phenotypes. In order to identify such phenotypes unambiguously, we performed a single-cell analysis of brain sections.

Using laser scanning confocal microscopy, we determined definitively that individual cells coexpressed GFP and neuron-specific proteins (20). By using parameters that resulted in a calculated optical section thickness of 0.3 to 0.4 µm, this rigorous analytic method ensured that the colocalization of markers observed represented true coexpression of different proteins within the same cell. The olfactory bulb (OB) was selected for in-depth quantification because of its high frequency of ongoing neural regeneration relative to other parts of the brain (21-23). The olfactory bulbs of five recipients of GFP+ bone marrow were analyzed in-depth.

Thousands of GFP+ individual cells in the brain were subjected to confocal analysis in order to obtain quantitative data on morphology and gene expression (20, 24). An average of 220 (SD ± 96) GFP+ cells were observed per coronal OB section. The majority of GFP+ cells in all sections expressed F4/80, which is indicative of a macrophage/microglial phenotype. However, because many GFP+ cells in the population lacking F4/80 had morphologies consistent with those of neuronal cells, additional sections were stained with antibodies specific to neuronal proteins (Tables 1 and 2), including NeuN and the 200-kD isoform of neurofilament (NF-H) (25) as well as the astrocyte-specific marker, glial fibrillary acidic protein (GFAP) (24). GFP+ cells were categorized according to their morphology and their expression of neuronal and non-neuronal proteins (Table 1 and Figs. 2 and 3). Neuronal gene expression was detected in round-, oval-, and rod-shaped cells, but was most abundant in triangular cells, many of which had one or more extensions of up to 70 µm.

A compilation of the number of individual cells analyzed by confocal microscopy that expressed the neuronal proteins, NeuN or NF-H, is presented in Table 2. A total of 165 NeuN-expressing and 128 NF-H-expressing cells were scored in eight and four OB sections, respectively. Given the total number of neurons estimated per section, about 0.2 to 0.3% of the total number of neurons were derived from the bone marrow by 8 to 12 weeks after the bone marrow transplant. Although this percentage is small, the total number of cells observed is substantial, providing strong evidence that cells derived from the bone marrow can express neuron-specific genes and assume a neuronal phenotype within the brains of adult animals.

The distribution within the OB of GFP+ cells expressing neuron-specific genes was analyzed further. When coronal cross sections of the OB were analyzed, the majority of GFP+ cells expressing neuronal markers were found in the superficial axon layer (SAL) and in the glomerular layer just adjacent to the SAL (Table 2). This finding is in good agreement with studies by others that indicate that the outer regions of the OB have a high degree of regenerative activity (21-23).

Although astrocytes expressing GFAP were readily detected in four OB sections (24), no GFP+ cells were found in these sections that expressed GFAP (Table 2). This finding is in contrast to previous reports, which suggested that cells of the astrocytic lineage could be obtained from nonstromal bone marrow in vivo (26). This may be due either to a difference in experimental protocol or to the higher resolution made possible by using GFP-labeled cells in conjunction with confocal analysis.

A substantial proportion of the marrow-derived cells assayed coexpressed multiple neuron-specific gene products. Sections from five different mice were analyzed by confocal microscopy for the coexpression of two neuronal proteins, 200-kD neurofilament (NF-H) and class III -tubulin (TuJ1) (Fig. 2). Twenty-four of a total of 80 GFP+ cells selected for analysis based on morphology were found to coexpress these two cytoskeletal proteins in addition to GFP. Two sections with groups of cells that coexpressed both NF-H and class III -tubulin are shown in Fig. 3, A through E and F through I, respectively.

Bone marrow-derived neural gene expressing cells were well integrated in the microenvironment of the OB in a manner consistent with their neighbors (Fig. 4). Without the GFP marker, the cells are not easily distinguishable from their neighbors. This is true not only with respect to neuronal gene expression, for example NeuN, but also with respect to their ability to phosphorylate the transcription factor CREB. Indeed, nine cells in the field in Fig. 4 exhibited all three markers: GFP, NeuN, and pCREB (27). CREB is a transcription factor that is activated by phosphorylation (pCREB) in a number of diverse cell types. In neurons of the brain, CREB is activated in response to extracellular stimuli such as glutamate, neurotrophins, trauma, or stress (28, 29). The GFP+ cells from the marrow exhibited phosphorylation of CREB in a manner similar to the cells surrounding them. We observed several such regions in the outer layers (glomerular and superficial axonal layers) of the OBs of all three transplanted mice that were stained for NeuN and pCREB. These cells were in layers in which the majority of neuronal regeneration has been documented to occur in adult mice (21-23). In the OB, the cells that phosphorylate CREB most strongly are neurons (30); however, the existence of NeuN+ cells that are lacking pCREB is not surprising, as the OB contains diverse types of neurons (31). Taken together, the CREB phosphorylation shown here indicates that a major signal transduction pathway present in neurons is also intact in bone marrow-derived cells that express neuron-specific genes in the brain. These results suggest that the cells are able to respond to cues in their environment in a manner consistent with the surrounding cells.

Our results clearly show that adult cells from the marrow can gain access to the adult brain and assume
characteristics of CNS neurons. Hundreds of GFP+ cells in the OB expressed neuron-specific genes. Of these, only a few had long axon-like extensions (Figs. 2 and 3), whereas short nonbranching extensions were prevalent. However, neuronal cells with this morphological characteristic are not unexpected in the OB, as they appear to be typical for class III -tubulin expressing "neuroblasts" migrating into the OB from the subventricular zone (SV2) (22, 23). To determine definitively the origin of cells in the OB will require marking studies and time courses as described by others (22, 23). Many factors remain to be explored that could increase the frequency at which cells derived from marrow produce neuron-like cells in the brain: for example, the influence of donor and recipient age, the relative role of transient or chronic injury, the time period after transplantation, and the availability of specific trophic factors. Nonetheless, the results reported here reveal an unexpected degree of plasticity of adult bone marrow-derived cells. Recently, marrow-derived cells were reported to express neuronal markers in tissue culture (32); however, their ability to yield neuronal phenotypes in response to physiological signals in vivo has not
previously been shown. Thus, our findings are not only of fundamental interest but also, once more robust, could have application as a cell-mediated therapy. Not only could neurons be contributed to the adult brain, but, if genetically engineered, they could be a potentially useful tool for treating disorders characterized by defective neuronal function or a loss of neurons such as Parkinson's disease, lysosomal storage disorders, psychiatric disorders, trauma, and other types of CNS injury.

 References and Notes:

    1. B. T. Spear and S. M. Tilghman, Mol. Cell. Biol. 10, 5047 (1990).
    2. W. E. Wright, Exp. Cell Res. 151, 55 (1984).
    3. H. M. Blau, C. P. Chiu, C. Webster, Cell 32, 1171 (1983).
    4. H. M. Blau and D. Baltimore, J. Cell Biol. 112, 781 (1991).
    5. J. B. Gurdon, J. Embryol. Exp. Morphol. 10, 622 (1962).
    6. I. Wilmut, A. E. Schnieke, J. McWhir, A. J. Kind, K. H. Campbell, Nature 385, 810 (1997).
    7. A. L. Vescovi and E. Y. Snyder, Brain Pathol. 9, 569 (1999).
    8. F. H. Gage, G. Kempermann, T. D. Palmer, D. A. Peterson, J. Ray, J. Neurobiol. 36, 249 (1998).
    9. B. E. Petersen, et al., Science 284, 1168 (1999).
   10. G. Ferrari, et al., Science 279, 1528 (1998).
   11. E. Gussoni, et al., Nature 401, 390 (1999).
   12. C. R. Bjornson, R. L. Rietze, B. A. Reynolds, M. C. Magli, A. L. Vescovi, Science 283, 534 (1999).
   13. M. Okabe, M. Ikawa, K. Kominami, T. Nakanishi, Y. Nishimune, FEBS Lett. 407, 313 (1997).
   14. Marrow was isolated in a sterile environment from 8- to 10-week-old, male transgenic mice that
ubiquitously expressed enhanced green fluorescent protein (GFP) and non-GFP control mice. After lethal
irradiation, 8- to 10-week-old C57BL/6 mice (Stanford) received 6 × 10⁶ cells by tail vein injection.
   15. Isolated brains were minced with a razor blade, dissociated with proteases, washed, stained with TriChrome (TC)-conjugated rat antibody to mouse CD11b and allophycocyanin (APC)-conjugated rat antibodies to mouse CD45, and analyzed by flow cytometry.
   16. F. H. Gage, et al., Proc. Natl. Acad. Sci. U.S.A. 92, 11879 (1995).
   17. J. A. Ledbetter and L. A. Herzenberg, Immunol. Rev. 47, 63 (1979).
   18. T. Springer, G. Galfre, D. S. Secher, C. Milstein, Eur. J. Immunol. 8, 539 (1978).
   19. H. Akiyama and P. L. McGeer, J. Neuroimmunol. 30, 81 (1990).
   20. Each GFP+ cell was analyzed for antibody staining by three-dimensional confocal laser scanning microscopy. Data was collected with sequential laser excitation to eliminate bleedthrough and with confocal parameters (e.g., pinhole sizes) selected to minimize the thickness of the calculated optical section.
   21. J. W. Hinds, J. Comp. Neurol. 134, 287 (1968).
   22. M. B. Luskin, Neuron 11, 173 (1993).
   23. C. Lois and A. Alvarez-Buylla, Science 264, 1145 (1994).
   24. Eight or 12 weeks after bone marrow transplant, experimental mice and age-matched control mice were perfused and fixed with 1.5% paraformaladehyde (PF)/0.1% glutaraldehyde, snap frozen in TISSUE-TEK O.C.T. compound, cryosectioned, and stained as floating sections with antibodies against NeuN, 200-kD neurofilament, class III -tubulin, GFAP, F4/80, and CD45. All sections were blocked with 25% normal goat serum, 0.25% Triton-X 100, and antibody to CD16/32. Goat antibodies to mouse and to rabbit conjugated to Texas Red or Cy5 were used as secondary antibodies.
   25. R. J. Mullen, C. R. Buck, A. M. Smith, Development 116, 201 (1992).
   26. M. A. Eglitis and E. Mezey, Proc. Natl. Acad. Sci. U.S.A. 94, 4080 (1997).
   27. Mice were anesthetized with Methoxyflurane and surgically decapitated, and the OBs were rapidly isolated and incubated in Tyrode solution. OBs were fixed in 1.5% paraformaldehyde/4 mM EGTA and were sliced to yield coronal sections. The samples were blocked and permeabilized in 0.3% Triton-X 100, 3% bovine serum albumin (BSA), monoclonal antibody to mouse CD16/CD32, and 100 mM glycine. Staining was performed with polyclonal antibody to pCREB and monoclonal antibody to NeuN, washed, and stained with secondary antibodies (Texas Red, goat antibody to rabbit; Cy5, goat antibody to mouse).
   28. K. Deisseroth, H. Bito, R. W. Tsien, Neuron 16, 89 (1996).
   29. A. J. Shaywitz and M. E. Greenberg, Annu. Rev. Biochem. 68, 821 (1999).
   30. N. Liu, et al., J. Biol. Chem. 274, 3042 (1999).
   31. G. M. Shepherd, Physiol. Rev. 52, 864 (1972).
   32. D. Woodbury, E. J. Schwarz, D. J. Prockop, I. B. Black, J. Neurosci. Res. 61, 364 (2000).
   33. P. S. Eriksson, et al., Nature Med. 4, 1313 (1998).
   34. Web video and supplemental text are available at Science Online at:
      www.sciencemag.org/cgi/content/full/290/5497/1775/DC1.
   35. We wish to thank neurobiologists S. McConnell, K. Deisseroth, and J. Weimann for their expertise and ongoing guidance; U. Wang and S. Heck for technical expertise; B. Blakely for insightful comments;
M. Okabe for transgenic GFP mice; and M. Greenberg for antibody to pCREB. This research was
supported by the Life and Health Insurance Medical Research Fund and a NIH predoctoral training grant
(T.R.B.), a fellowship from Human Frontiers in Science Program (F.M.V.R.), a postdoctoral fellowship
(G.I.K.), and NIH research grants CA59717, AG09521, and HD18179 (H.M.B).

1. Frenster JH, and Herstein PR, "Gene De-Repression", New Eng. J. Med. 288: 1224 (June 7, 1973).

2. Lagasse E, Connors H, Al-Dhalimy M, Reitsma M, Dohse M, Osborne L, Wang X, Finegold M, Weissman IL, and Grompe M, "Purified Hematopoietic Stem Cells Can Differentiate Into Hepatocytes In Vivo", Nature Medicine 6: 1229 (November, 2000).

3. Mezey E, Chandross KJ, Harta G, Maki RA, and McKercher SR, "Turning Blood into Brain: Cells Bearing Neuronal Antigens Generated In Vivo from Bone Marrow", Science 290: 1779 (December 1, 2000).