Turning Brain into Blood: A Hematopoietic Fate Adopted by Adult Neural Stem Cells in Vivo

See allHide authors and affiliations

Science  22 Jan 1999:
Vol. 283, Issue 5401, pp. 534-537
DOI: 10.1126/science.283.5401.534


Stem cells are found in various organs where they participate in tissue homeostasis by replacing differentiated cells lost to physiological turnover or injury. An investigation was performed to determine whether stem cells are restricted to produce specific cell types, namely, those from the tissue in which they reside. After transplantation into irradiated hosts, genetically labeled neural stem cells were found to produce a variety of blood cell types including myeloid and lymphoid cells as well as early hematopoietic cells. Thus, neural stem cells appear to have a wider differentiation potential than previously thought.

Stem cells have been identified in adult tissues that undergo extensive cell replacement due to physiological turnover or injury such as the hematopoietic, intestinal, and epidermal systems (1). These cells have been found in the central nervous system (CNS) (2), a tissue thought to be capable of extremely limited self-repair. CNS stem cells can generate the three major cell types found in the adult brain: namely, astrocytes, oligodendrocytes, and neurons (3). This is consistent with the view that the developmental potential of stem cells is restricted to the differentiated elements of the tissue in which they reside. However, some developmental peculiarities suggest certain cells may be able to differentiate into cell types that are not of the same dermal origin (4). Hence, we sought to determine whether neural stem cells (NSCs) could produce hematopoietic progeny.

A bone marrow (BM) repopulation assay (5) was used to test this hypothesis. Hematopoietic stem cells from unfractionated adult BM (107cells per animal) or NSCs cultured from either the embryonic or adult forebrain (106 cells per animal) (6) of ROSA26 mice were systemically injected into sublethally irradiated Balb/c recipient animals (7). ROSA26 animals were selected as the source of donor tissue because they are of a different immunological background than Balb/c and are transgenic forlacZ (8), which encodes for the Escherichia coli enzyme β-galactosidase. Importantly, to eliminate possible contamination of NSCs with cells of mesodermal origin (9), we also injected clonally derived adult ROSA26 NSCs (Fig. 1).

Figure 1

Cloning of adult ROSA26 CNS stem cells. (A) Single adult ROSA26 NSC plated in isolation in a single well (arrow). After (B) 1 and (C) 8 days, a cluster of cells formed, which was serially subcultured every 4 days to establish a continuous culture. (D) All cells expressed the NSC antigen, nestin. Differentiation was induced by plating a fraction of these cells in 1% fetal bovine serum, in the absence of growth factors. (E) The simultaneous detection of neurons [red; 25.3 ± 0.9% of total cell number (TCN);n = 6, ± SEM], astroglia (blue; 71.6 ± 6.3% TCN; n = 6), and oligodendroglia (green; 0.9 ± 0.01% TCN; n = 6) among the progeny of this cell indicate its tripotentiality (19). Secondary clones of the cell displayed in (A) produced an average of 48 ± 3.2 (n = 6) cells capable of producing tertiary, tripotential clones, thereby demonstrating self-maintenance (2). Data are from clone 2H1. Bars: (A) to (C), 90 μm; (D), 60 μm; (E), 40 μm.

Polymerase chain reaction (PCR) (10) was used to assay for the presence of lacZ in splenic DNA isolated from animals transplanted 5 to 12 months earlier. The lacZ gene was not detected in samples from either unirradiated or irradiated Balb/c mice that received vehicle (EBSS) (Fig. 2), whereas a strong signal was observed from both untreated ROSA26 animals and irradiated Balb/c mice that received ROSA26 BM (Fig. 2). A strong lacZ signal was also detected in animals injected with embryonic, adult, or clonally derived adult NSCs (Fig. 2). Because the behavior of NSCs derived from three clones (2H1, 4E8, and 3C6) was indistinguishable from that of bulk cultures, only results from a representative clone (2H1) and embryonic NSCs will be considered for the remainder of this report.

Figure 2

Detection of lacZ in splenic DNA isolated from animals injected with ROSA26 NSCs by PCR. ThelacZ gene was not detected in control (lane 1) or irradiated Balb/c animals injected with vehicle (lane 3). ROSA26 animals (lane 2) and Balb/c mice that received either ROSA26-derived BM (lane 4), embryonic NSCs (lane 5), adult NSCs (lane 6), or clonal adult NSC 2H1 (lane 7) produced strong amplification signals for lacZ(upper band; 374 base pairs). GAPDH was also amplified in the same reaction tube as an internal control (10) (lower band; 309 base pairs).

To test whether engrafted NSCs adopted a hematopoietic identity, we used in vitro clonogenic assays, immunocytochemistry, and flow cytometric analysis. For the clonogenic assays, cells from the BM of transplanted animals were plated in methylcellulose in the presence of defined cytokines (11). Ten to 14 days after plating, colonies founded by single hematopoietic progenitor cells were subjected to X-Gal histochemistry to detect β-galactosidase activity (12), thereby identifying hematopoietic precursors of a NSC origin. None of the colonies derived from the BM of irradiated Balb/c animals injected with EBSS stained positively for β-galactosidase (Fig. 3A). Conversely, BM isolated from recipients of either embryonic or adult NSCs formed colonies that reacted strongly to X-Gal (Fig. 3, B and C). A few colonies (<5%) did not stain positively for β-galactosidase (Fig. 3C), showing that some endogenous hematopoietic progenitors had survived the sublethal irradiation. Different types of colonies generated from BM isolated from adult NSC recipients included pure granulocyte (13%), granulocyte-macrophage (Fig. 3, D, E, and F) (30%), and pure macrophage (Fig. 3, G, H, and I) (22%), as well as mixed colonies (19%). Megakaryocytic and B cell colonies were also present, although at lower frequencies (<1% and <10%, respectively). Erythroid cells were not taken into account in this analysis because their evaluation cannot be reliably carried out by X-Gal staining. None of the NSC cultures proliferated or formed colonies when used in the same clonogenic assay before injection. Thus, ROSA26-derived NSCs can give rise to hematopoietic precursors after engraftment into irradiated Balb/c hosts.

Figure 3

NSCs produce early hematopoietic cells after transplantation into irradiated Balb/c recipients. An in vitro clonogenic assay (11) was used to analyze hematopoietic precursors in Balb/c mice injected with either embryonic or clonal adult NSCs, both of which yielded identical results. X-Gal histochemistry was used to identify hematopoietic clones derived from NSCs after 10 to 14 days in culture (12). Whereas none of the colonies from Balb/c BM stained positively for β-galactosidase (A), colonies from the BM of animals that received adult NSCs displayed β-galactosidase activity (blue) (B). In the same cultures a few unlabeled clones were found in Balb/c mice injected with clonal adult NSCs (C) (arrow), showing the persistence of a small number of endogenous (Balb/c) hematopoietic precursors in recipient animals. NSC-derived (β-galactosidase–reactive) colonies were further characterized by morphology. Colonies with distinctive granulocyte-macrophage (D and E) and pure macrophage (G andH) characteristics before (D and G) and after (E and H) reaction with X-Gal are shown. The X-Gal reaction was stopped prematurely to allow for morphological identification of individual cells within these colonies after staining. High-power microphotographs show the identifying morphology of the cells derived from the two types of colonies as stained by May-Grünwald-Giemsa [(F), granulocyte-macrophage; (I), pure macrophage]. Bars: (A), 40 μm; (B) and (C), 90 μm; (D), (E), (G), and (H), 40 μm; (F) and (I), 15 μm.

To further demonstrate the engraftment of NSCs into the hematopoietic system, we exploited the fact that distinct cell surface antigens are expressed by ROSA26 (H-2Kb) and Balb/c (H-2Kd) mice. We assayed cells isolated from the spleen, BM, and peripheral blood of transplanted and control animals by flow cytometry (13) using antibodies specific to H-2Kb and H-2Kd (14). Whereas no H-2Kb–positive (H-2Kb+) cells were found in animals injected with EBSS alone, numerous H-2Kb+ cells were detected in animals that received either ROSA26 BM or, more importantly, NSCs (Fig. 4A). By this approach, early experiments showed effective engraftment with between 105 and 107 NSCs per animal. With 106NSCs per animal and 107 BM cells per animal, 100% of BM, 100% of embryonic NSCs, 70% of adult NSCs, and 63% of clonal-adult NSC recipients showed positive engraftment by flow cytometric analysis (n = 20, 20, 40, and 30 animals per group, respectively). In addition, donor-derived (H-2Kb+) cells were first detected in the peripheral blood of NSC recipients 20 to 22 weeks after injection compared with 16 to 18 weeks for ROSA26 BM recipients. After initial detection in peripheral blood, spleen cells were harvested and processed for dual-label flow cytometry (13) with antibodies to H-2Kb in combination with either antibodies to CD3e (anti-CD3e) (T lymphocytes), anti-CD19 (B lymphocytes), or anti-CD11b (myeloid cells) (15). A significant number of ROSA26-derived NSCs gave rise to B and T lymphocytes or myeloid cells after transplantation into Balb/c hosts (Fig. 4, A and B). None of the hematopoietic antigens tested was expressed by any NSCs before transplantation.

Figure 4

Identification of differentiated hematopoietic cell types derived from ROSA26 NSCs. Hematopoietic cells from transplanted mice were processed for flow cytometry (13). (A) A significant number of H-2Kb+ cells were found in the BM, spleen, and peripheral blood of Balb/c animals injected with ROSA26 BM, and embryonic or adult clonal NSCs, but not in unirradiated Balb/c or EBSS recipients. Similarly, none of the CD3e, CD19, or CD11 immunoreactive cells identified in the spleens of control animals (Balb/c, EBSS) expressed H-2Kb. Conversely, a significant proportion of CD3e, CD19, or CD11b immunoreactive cells isolated from the spleens of ROSA26 animals as well as ROSA26 BM, embryonic, or adult clonal NSC recipients were H-2Kb+. All percentages were calculated relative to the total events gated (n = 6, ±SEM; * indicates P < 0.05 compared with Balb/c by analysis of variance). (B) Representative dual-label FACS plots identifies CD3e (panels 1, 4, and 7), CD19 (panels 2, 5, and 8), and CD11b (panels 3, 6, 9) immunoreactive cells (y axes) that express H-2Kb (x axes) in Balb/c animals that received EBSS (panels 1 to 3), and in ROSA26 BM (panels 4 to 6) or adult clonal NSC (panels 7 to 9) recipient animals.

Thus, NSCs isolated from the embryonic and adult murine forebrain, which generate neurons and glia, engraft into the hematopoietic system of irradiated hosts to produce a range of blood cell types. This demonstrates that the actual differentiation potential of adult NSCs, which are currently viewed as tripotent neural precursors, is much broader than expected.

We chose sublethal irradiation because it eliminates a significant fraction of the endogenous hematopoietic precursors without killing the mouse. We felt this would be necessary because NSCs would likely need more time to acquire a hematopoietic fate than a lethal dose would allow. That the repopulation of the immune system took on average 3 weeks longer, combined with a slightly weaker engraftment for NSC compared with BM recipient animals, seems to supports this idea. This extra time required suggests that NSCs undergo additional steps of fate determination, differentiation, and maturation with respect to BM cells to produce hematopoietic progeny.

It has been suggested by studies in various model systems that most somatic cell specialization may not involve irreversible genetic changes (16). The seminal demonstration of conserved genomic totipotentiality in adult somatic cells was provided in a study that describes the cloning of an adult ewe (17). Our work is complementary to these findings and suggests that the reactivation of dormant genetic programs may not require nuclear transfer or experimental modification of the genome. NSCs appear naturally endowed with the appropriate machinery required to express an otherwise silent genomic potentiality in response to an appropriate pattern of stimulation.

Given the fact that human NSCs can be continuously expanded for extended periods of time (18), the finding presented here may have implications for the treatment of a number of human disorders. If they behave similarly to their murine counterparts, human NSCs may provide a renewable, characterized source of cells that could be used in approaches aimed at hematopoietic reconstitution in various blood diseases and disorders.

  • * These authors contributed equally to this work.

  • Present address: University of Washington, Department of Biochemistry, Seattle, WA 98195–7350, USA.

  • To whom correspondence should be addressed at University of Washington, Department of Biochemistry, Seattle, WA 98195–7350, USA, e-mail: adanac{at} (for C.R.R.B.) and at Istituto Nazionale Neurologico C. Besta, Via Celoria 11, Milan, Italy I-20133, e-mail: vescovi{at} (for A.L.V.).

  • § Present address: Walter Eliza Hall Institute, Medical Research, Parkville, Victoria, Australia 3050.


View Abstract

Stay Connected to Science

Navigate This Article