Multipotent stem cells from adult olfactory mucosa



Multipotent stem cells are thought to be responsible for the generation of new neurons in the adult brain. Neurogenesis also occurs in an accessible part of the nervous system, the olfactory mucosa. We show here that cells from human olfactory mucosa generate neurospheres that are multipotent in vitro and when transplanted into the chicken embryo. Cloned neurosphere cells show this multipotency. Multipotency was evident without prior culture in vitro: cells dissociated from adult rat olfactory mucosa generate leukocytes when transplanted into bone marrow–irradiated hosts, and cells dissociated from adult mouse olfactory epithelium generated numerous cell types when transplanted into the chicken embryo. It is unlikely that these results can be attributed to hematopoietic precursor contamination or cell fusion. These results demonstrate the existence of a multipotent stem-like cell in the olfactory mucosa useful for autologous transplantation therapies and for cellular studies of disease. Developmental Dynamics 233:496–515, 2005. © 2005 Wiley-Liss, Inc.


In recent years, neural stem cells have been shown to be capable of differentiating into non-neural cell types, suggesting a broad developmental capacity that can be directed by the surrounding environment in vitro or in vivo (Reynolds and Weiss, 1992; Bjornson et al., 1999; Clarke et al., 2000; Rietze et al., 2001). A potential source of such neural stem cells is the olfactory mucosa, the sense organ of smell. Vigorous neurogenesis continues in this tissue throughout adult life, replacing the olfactory sensory neurons (Graziadei and Monti Graziadei, 1979; Calof and Chikaraishi, 1989; Mackay-Sim and Kittel, 1991; Pixley, 1992; Mahanthappa and Schwarting, 1993; MacDonald et al., 1996; Newman et al., 2000). Olfactory neurogenesis is regulated by the same growth factors that control neurogenesis in the central nervous system (Mackay-Sim and Chuah, 2000) and occurs in adult human (Wolozin et al., 1992; Murrell et al., 1996; Féron et al., 1998). We reasoned that the olfactory mucosa may contain a neural stem cell that would have a developmental potency similar to other neural stem cells.

Neural stem cells from the forebrain typically grow in vitro as “neurospheres” (Reynolds and Weiss, 1992), and these neurospheres have become a surrogate assay for neural stem cells. Neurospheres derived from olfactory mucosa can give rise to neurons and glia (Roisen et al., 2001). The aim of the present study was to test whether the olfactory mucosa contains a neural stem cell with the broad developmental potency ascribed to other adult stem cells and able to give rise to cells outside the neural lineage. This potential is predicted from recent studies in which it is reported that basal cells of the olfactory epithelium gave rise to the non-neural cells of the olfactory mucosa (Carter et al., 2004; Chen et al., 2004). The specific aim here was to investigate whether this multipotency extended to cells of tissues outside the olfactory mucosa. Our hypothesis is that the presence of a sufficiently primitive “stem cell” will exhibit potency in response to the environment into which the cells are placed (Clarke et al., 2000).

“Stem cells” from adult tissues are generally named after the tissue to which they normally give rise, such as “neural stem cell” and “hematopoietic stem cell.” A stem cell should exhibit properties of self-renewal and the ability to give rise to multiple cell lineages. Definitions of stem cells vary from pluripotent cells that give rise to all cells of the body (as presumed for embryonic stem cells), to multipotent cells that give rise to all cells within a tissue (as presumed for hematopoietic stem cells and neural stem cells). The question addressed here is whether a cell derived from the olfactory mucosa has properties of a stem cell, that is, self-renewal and multipotency. A second question is whether the developmental potency extends to tissues beyond the olfactory mucosa.

Three complementary sets of experiments are presented here that investigate the multipotency of a putative olfactory stem cell. (1) Neurospheres derived from human olfactory mucosa were examined for their multipotency in a variety of environments in vitro and after transplantation into the chick embryo. (2) Cells dissociated from mouse olfactory epithelium were transplanted into chick embryo. (3) Cells dissociated from rat olfactory mucosa were transplanted into bone marrow–irradiated hosts. The results of these experiments are all consistent with the presence of a stem-like cell whose developmental potency is directed by signals in its “environmental niche.” Experiments are presented that suggest that this multipotency is unlikely to be the result of fusion with other cells in vitro or in vivo.


Human Olfactory Mucosa Biopsies Produced Neurospheres

Approximately 7 to 10 days after plating cells from olfactory biopsies, floating, phase-bright spheres of cells were found in all cultures from both lamina propria and epithelium. At this time, the spheres comprised approximately 1,000 cells. Typically, spheres arose over every 1 to 4 mm2 of the culture dish from groups of attached cells, beginning as clusters of approximately 50 cells and growing to 40,000 cells after longer periods eventually floating free in the medium (Fig. 1A). This method produced a yield of 500 to 2,000 spheres per biopsy after 7–10 days in vitro. After gentle trituration, these primary spheres were harvested from the supernatant after which additional primary spheres were produced by the remaining cells. Some primary cultures of olfactory cells continued to produce spheres for many months. Primary spheres were frozen, thawed, and successfully re-plated with a survival rate of 83%. Neurospheres were generated from biopsies from all participants whose ages ranged from 20 to 78 years.

Figure 1.

Human olfactory neurosphere proliferation and differentiation. A: Phase-bright spheres of cells (“neurospheres”) after 10 days in vitro. B: Reverse transcriptase-polymerase chain reaction (RT-PCR) revealed nestin mRNA in human olfactory mucosa (lane 2) and olfactory neurospheres (lane 3). Lane 1, DNA size markers (100 bp and 200 bp); lane 4, water control. The predicted band for nestin is 197 bp. C: Optical section through olfactory neurosphere showing cells immunostained for Nestin (green) and glial fibrillary acidic protein (GFAP, red). Some cells were double labeled (yellow). DI: When neurospheres were dissociated and grown on plastic some were immunopositive for GFAP (D), a marker for astrocytes; O4 (E) and GAL C (F) markers for oligodendrocytes; and MAP5 (G), neurofilament (I), and β-tubulin III (H), markers for neurons. Scale bars = 40 μm in A, 10 μm in C–J.

Human Olfactory Neurospheres Gave Rise to Neurons and Glia

Primary neurospheres expressed nestin, an intermediate filament protein expressed in neural stem cells in the brain (Lendahl et al., 1990). It is present in the olfactory epithelium (Osada et al., 1995; Doyle et al., 2001) and within the olfactory ensheathing cells of the olfactory nerve (Sonigra et al., 1999). Nestin mRNA was present in primary neurospheres, identified using reverse transcriptase-polymerase chain reaction (RT-PCR, Fig. 1B) and confirmed by sequencing the PCR product. Nestin immunoreactivity was present in many cells within the neurosphere (Fig. 1C). Some neurosphere cells expressed glial fibrillary protein (GFAP; Fig. 1C); others expressed β-tubulin III (data not shown). When primary neurospheres were dissociated and replated at low density, differentiated cells were observed to express GFAP, a marker for astrocytes (Fig. 1D); O4 and GalC, markers for oligodendrocytes (Fig. 1E,F); or neurofilament, β-tubulin III, and MAP5, markers for neurons (Fig. 1G–I,).

When neurospheres were dissociated into single cell suspensions and replated, their growth varied with culture conditions (Fig. 2). After 5 days in culture on uncoated plastic, cell density was lowest in serum-free medium compared with medium with 10% serum, fibroblast growth factor-2 (FGF2) or transforming growth factor-alpha (TGFα; Fig. 2A). Cell density also varied with culture surface, with lowest densities when cells were grown in a medium with serum for 5 days on polylysine and polyornithine (Fig. 2B). Matrix molecules such as laminin, fibronectin, and collagen had no effect on cell density, compared with uncoated plastic (Fig. 2B). Accordingly, neurospheres were dissociated and replated onto uncoated wells and grown for 5 days in different media (Fig. 2C). In defined, serum-free medium (Dulbecco's modified Eagle medium [DMEM] containing only insulin, transferrin, and selenium) the majority of cells (50.3% ± 8.09) expressed GFAP, whereas only 4.5% ± 0.29 expressed β-tubulin III, and 1.33% ± 0.33 expressed O4. Fetal calf serum decreased the proportion of GFAP expressing cells (44.67% ± 1.76) and increased the proportions of cells expressing β-tubulin III (18.33% ± 0.88) and O4 (3.67% ± 0.88). The highest proportion of cells expressing GFAP (69% ± 3.21) was obtained with ciliary neurotrophic factor (CNTF) in the serum-free medium. The highest proportion of neurons (25.33% ± 1.45) was obtained with nerve growth factor (NGF). The highest proportion of cells expressing O4 (50.67% ± 2.96) was obtained with retinoic acid.

Figure 2.

Growth factor regulation of human neurosphere proliferation and differentiation. Neurospheres were dissociated and grown in various media and substrates. A: Fetal calf serum (FCS), fibroblast growth factor-2 (FGF2), and transforming growth factor-alpha (TGFα) increased cell proliferation compared with control (CON) cultures in serum-free medium when grown on plastic culture wells (n = 5 for all conditions). Analysis of variance indicated a significant difference among the growth conditions (F3,11 = 17.54; P = 0.0007). B: Cells were grown in fetal calf serum on various substrates. Culture wells coated with poly-l-lysine (Ply) and poly-ornithine (Plo) reduced cell proliferation compared with uncoated plastic culture wells (Pla), and wells coated in collagen (Col), fibronectin (Fib), and laminin (Lam), n = 3 for all conditions. Analysis of variance indicated a significant difference among the growth conditions (F5,17 = 9.02; P = 0.0009). C: Cells were grown on plastic culture wells in various media. Differentiation of cells into neuronal, astrocytic, and oligodendroglial phenotypes (observed by β-tubulin III, glial fibrillary acidic protein, and O4 staining, respectively) was influenced by different growth factors (n = 5 for each growth condition): ciliary neurotrophic factor (CNTF) induced astrocyte differentiation, nerve growth factor (NGF).induced neuronal differentiation, and retinoic acid (RA) induced oligodendrocyte differentiation. Two-way analysis of variance indicated significant effects of growth factor (F4,30 = 5015; P = 0.0028), cell type (F22,30 = 248.2; P < 0.0001), and their interaction (F8,30 = 64.4; P < 0.0001).

Human Olfactory Neurospheres Were Self-Replicating

Self-replication is an important defining characteristic of stem cells (Reynolds and Weiss, 1992). To demonstrate this, a two-stage process was developed in which the cells from “primary” neurospheres were dissociated and proliferated on plastic followed by plating on poly-L-lysine to form “secondary” neurospheres. During the first stage, the cells attached to the culture well and underwent rapid proliferation reaching confluency within a week but did not create secondary neurospheres. After the second stage, secondary spheres were derived with the same characteristics as primary spheres. Like primary neurospheres, the secondary neurospheres could be collected by gentle trituration leaving the remaining cells to produce further neurospheres. The secondary neurospheres gave rise to neurons and glial cell types when grown in similar conditions to primary neurospheres (data not shown). Secondary neurospheres were able to produce tertiary neurospheres, which could also differentiate into neurons and glial cell types.

Human Olfactory Neurospheres Gave Rise to Cells of Non-neural Lineage In Vitro

A transwell induction assay was used to investigate whether neurosphere-derived cells could differentiate along non-neural lineage pathways. In this assay, the neurosphere-derived cells grow separated from but in medium conditioned with soluble factors derived from living neonate tissue slices. In each case, cells migrated from the neurospheres and differentiated into cells with the morphology of, and expressing proteins normally found in each of the cocultured tissues (Fig. 3A–F). Cells growing beneath liver were ovoid or blunted in morphology and appeared laden with ferritin and albumin (Fig. 3A,B). Cells beneath cardiac muscle had a multipolar fibrous appearance with numerous autophagosomes. Early cardiac muscle cultures exhibit autophagosomes both intra- and extracellularly as they reorganize their cytoskeletons to become myofibrillar (Nag et al., 1990). These olfactory neurosphere-derived cells were immunopositive for the cardiac muscle antigens, sarcomeric alpha actin, and cardiac troponin I (Fig. 3C,D). Cells beneath skeletal muscle were elongate and tended to concatenate, reminiscent of early myocyte cultures. Cells were immunopositive for the muscle antigens, skeletal myosin, and striated tropomyosin (Fig. 3E,F). The percentages of non-neural phenotypes were different for each inducing tissue and were different in medium containing serum and in serum-free medium (Table 1). Experimental control cultures, where neurospheres were plated in medium containing 5% fetal calf serum (FCS) and that were subjected to the same immunochemical analyses did not exhibit the same morphological or specific staining phenotype as test cultures (Fig. 3H). These results indicate that olfactory neurosphere-derived cells can differentiate into non-neural lineages in vitro when given the appropriate tissue-derived, soluble signals.

Figure 3.

Human neurospheres differentiated into non-neural lineages. In a transwell induction assay, human olfactory neurospheres were plated under live slices of neonate rat liver, heart, and skeletal muscle. After 4 days, cells were fixed and immunostained for proteins expressed by the inducer tissues. A,B: Under liver, cells were positive for ferritin (A) and albumin (B). C,D: Under heart, cells were positive for sarcomeric alpha actin (C) and cardiac troponin I (D). E,F: Under skeletal muscle, cells were positive for skeletal myosin (E) and striated tropomyosin (F). G: Immunochemistry control: human olfactory neurosphere cells grown under skeletal muscle but with no primary antibody and a secondary antibody only. H: Experimental control: human olfactory neurosphere cells grown in fetal calf serum only and subsequently stained for striated tropomyosin. When human olfactory neurosphere cells were transplanted into the chick embryo, they gave rise to cells in multiple chick tissues, including heart. K: Human cells in wall of a 3-day-old chick heart identified with an antibody to human specific cyclin E (bright green foci) using tyramide fluorescence amplification. L: An untransplanted control chick heart does not show fluorescence above background. I,J: Corresponding Nomarski images are shown. M: Confocal optical section through chick myocardium illustrating human-derived cells in red (cyclin E and 1,1′, di-octadecyl-3,3,3′,3′,-tetramethylindo-carbocyanine perchlorate [DiI] labeling in the same fluorescence channel) and cardiac troponin I, which shows its structural banding, in green. A fluorescence image of the nuclei (blue, stained with Hoechst) has been overlaid. In the head region of the same embryo a large cluster of cyclin E–stained cells contained subsets of neurofilament-positive cells (not shown). Scale bars = 10 μm in A–H, in J (applies to I–L), 5 μm in M.

Table 1. Transwell Induction Conversion Rates (%)a
Inducing tissuePhenotypic marker5% FCS in DMEMITS in DMEM
  • a

    Percentages are for cells growing out from spheres; cells remaining in spheres are not included in this analysis. Two alternative media were used for each experimental situation: 5% fetal calf serum in DMEM and a serum-free medium (ITS: insulin, transferrin, and selenium in DMEM). The data represent numbers of cells derived from single spheres repeated three times for each condition. FCS, fetal calf serum; DMEM, Dulbecco's modified Eagle medium.

LiverFerritin73.57 ± 4.5665.00 ± 11.82
 Albumin21.50 ± 8.6025.03 ± 13.29
Skeletal muscleSkeletal myosin71.56 ± 9.5181.23 ± 2.07
 Striated tropomyosin91.13 ± 7.7180.90 ± 5.10
HeartCardiac troponin-I15.50 ± 0.3666.13 ± 10.74
 Sarcomeric α-actin96.03 ± 3.3599.17 ± 1.18

Cloned Human Olfactory Neurospheres Were Multipotent In Vitro

Clonal neurospheres were derived from primary neurospheres and tested for their developmental potency. This method was achieved in a three-stage process in which cells were dissociated from primary neurospheres, plated at clonal dilution on plastic and allowed to grow to over 200 cells, and subsequently plated onto poly-L-lysine to form neurospheres. Of the 1,920 wells originally seeded at clonal dilution, 245 wells contained a single cell, confirmed visually. Of these, 74 (30%) proliferated and gave rise to clonal populations that were plated onto poly-L-lysine. After 10 days, 29 of 245 (12%) had formed neurospheres. The developmental potency of these clonal neurospheres was assessed using the same procedures used for primary neurospheres. When plated onto fibronectin, these clonal neurospheres produced cells with astrocyte, neuronal, and oligodendrocyte phenotype (data not shown). Some clonal cultures were genetically labeled with green fluorescent protein (GFP) and used in the same transwell induction assay described above. They produced cells of non-neural phenotype, including cardiac muscle (Fig. 4), skeletal muscle, and liver. The presence of the GFP gene confirmed that these cells were of human origin.

Figure 4.

Clonal human olfactory neurosphere cells differentiated into multiple lineages. Primary human olfactory neurosphere cells were green fluorescent protein (GFP) -labeled using a retroviral vector and cloned by limiting dilution, and secondary neurospheres were generated. These were dissociated and grown under neonate tissues in transwell induction assays. Clonal human olfactory neurosphere cells grown under cardiac tissue: A: GFP fluorescence. B: Nomarski image. C: Immunostaining for sarcomeric α-actin. The asterisk indicates the position of the neurosphere that washed away during immunoprocessing. Scale bar = 10 μm in B (applies to A–C).

Human Olfactory Neurospheres Gave Rise to Cells of Non-neural Lineage In Vivo

A chick embryo transplantation assay was used to investigate whether human neurosphere-derived cells could differentiate along non-neural lineage pathways. Primary neurospheres were dissociated, labeled with 1,1′, di-octadecyl-3,3,3′,3′,-tetramethylindo-carbocyanine perchlorate (DiI; Molecular Probes), and transplanted into the chick gastrula. Red DiI fluorescence was noted in many tissues, including cardiac atrium, cardiac ventricle, blood vessels, brain, spinal cord, liver, mesonephros, allantois, skeletal muscle, and gut. The presence of human cells in the regions of DiI fluorescence was confirmed using an anti-human cyclin E antibody (Fig. 3L,M). This antibody is specific for human cyclin E (Koff et al., 1991). Preliminary experiments confirmed that it stained the nuclei of proliferating human fibroblasts but did not stain any nuclei in the chicken embryo. In chick embryos where DiI fluorescence had indicated successful grafting of human-derived cells, nuclei of the same regions stained positive for human cyclin E. The identity of cyclin E-positive cells was confirmed in two tissues. Heart muscle cells that expressed cardiac troponin I also expressed cyclin E (Fig. 3M). In the head region a large cluster of cyclin E-stained cells contained subsets of neurofilament positive cells (not shown). These results indicate that human olfactory neurosphere-derived cells can differentiate into non-neural lineages in vivo when given the appropriate embryonic environment.

Mouse Olfactory Epithelial Cells Gave Rise to Cells of Non-neural Lineage In Vivo

The experiments above indicate that the human olfactory mucosa contains cells with multipotency after growth in vitro as neurospheres. It is possible that this multipotency arose only after manipulation in vitro and, therefore, may not reflect the potency of a “naive” olfactory stem cell as it exists in the olfactory epithelium. This hypothesis was tested by transplantation of mouse olfactory epithelial cells into the chick gastrula. Mouse cells were chosen for these experiments because the availability of a LacZ transgenic mouse allows unequivocal identification of cells of donor origin. Neural stem cells from the mouse forebrain produced multiple cell types after transplantation into this model (Clarke et al., 2000). The donor cells were from a transgenic mouse strain, which has the Escherichia coli Lac Z gene under the promoter of the HMGCoA Reductase gene (Tam and Tan, 1992). Chick gastrulae were transplanted in situ by microinjection of a few microliters of dissociated olfactory epithelium cells into the primitive streak as described (Clarke et al., 2000). Controls were (1) no interference, (2) sham injected, and (3) injected with a suspension of erythrocyte-free blood cells (leukocytes, bone marrow–derived stem cells, and progenitors).

Of the 387 embryos injected with mouse olfactory epithelial cells, 58 (15%) survived, and of these survivors, 16 (28%) stained positively for β-galactosidase activity indicated by blue X-gal staining (Fig. 5A,B). Of the 60 sham-operated and blood cell-transplanted embryos, 18 (30%) survived. None of these controls stained positive for bacterial β-galactosidase activity. In the olfactory cell-transplanted embryos, a variety of organs and tissues were blue: wing bud, leg bud, heart, pharyngeal arch, liver, notochord, somites, eye, hind brain, otic vesicle, forebrain, spinal cord, gut, amnion, and allantois (Fig. 6). Often the distribution of staining was asymmetric, for instance in only one eye. Some embryos were cleared in methyl salicylate to define better which cells expressed β-galactosidase (c.f. Fig. 5A vs. B). The specificity of the X-gal staining was confirmed using immunochemistry with an antibody to bacterial β-galactosidase (Fig. 5C vs. D,E). Thirteen X-gal–positive embryos were frozen, cryosectioned, and processed for immunohistochemistry for β-galactosidase and other phenotypic markers. β-Galactosidase immunoreactivity was observed in the heart atrium, heart ventricle, pharyngeal arch, liver, notochord, somites, optic vesicle, hind brain, otic vesicle, forebrain, midbrain, head mesenchyme, nasal pit, Rathke's pouch, spinal tube, dorsal root ganglia, intestine, amnion, allantois, trunk muscle, mesonephros, tail bud, and aorta. There was no X-gal staining or β-galactosidase immunoreactivity in control embryos, including those injected with blood cells, and no immunoreactivity in X-gal–positive sections in which the primary antibody was missing. Olfactory marker protein immunoreactivity was not found in any sections indicating that olfactory sensory neurons did not survive the transplantation process and did not differentiate from transplanted precursor cells.

Figure 5.

Mouse olfactory epithelial cells give rise to multiple lineages in the chick embryo. A.B: Mouse cells (blue X-gal staining) 4 days after transplantation of dissociated olfactory epithelial cells into chick gastrula identified in allantois (a), limb bud (l), heart (h), pharyngeal arch (p), and notochord (n). The embryo in A was cleared, the embryo in B was not. C: Fluorescence photomicrograph of section through chicken embryo 4 days after transplantation immunostained for mouse-derived β-galactosidase transgene (white). Prominent organs containing mouse-derived cells are heart (h), liver (l), and mesonephros (m). D,E: Section through the mesonephros (D) and liver (E) of the embryo shown in C, stained with X-gal (blue). Scale bars = 2 mm in A,B, 200 μm in C, 50 μm in D,E.

Figure 6.

Mouse olfactory epithelial cells differentiated into many tissues after transplantation into the chick gastrula. X-gal–positive tissues in chick embryos transplanted with mouse cells possessing the βgalactosidase transgene. A: Dorsal aorta. B: Pharyngeal arch. C: Mesonephros. D: Nephric canal. E,F: Heart ventricle. G,H: Liver. I: Otic vesicle. J: Dorsal root ganglion. Scale bars = 50 μm in A–E, 20 μm in F,H,I, 10 μm in G, 40 μm in J.

Within the host tissues, donor-derived cells had the normal morphology and expressed phenotypic markers typical of the host tissues in which they resided (Fig. 7). In neural tissues, neurons in many regions expressed both neurofilament and β-galactosidase, and some cells coexpressed S100 and β-galactosidase. In skeletal muscle, β-galactosidase colocalized with myosin and sarcomeric α-actin. In the ventricular wall of the developing heart, β-galactosidase colocalized with cardiac troponin I. In liver and in kidney mesonephros, β-galactosidase colocalized with ferritin. These results indicate that cells from olfactory mucosa have the ability to differentiate into cell types normally derived from endoderm, mesoderm, or ectoderm.

Figure 7.

Double labeling of mouse-derived cells in chick embryos. A: Dorsal root ganglion (4 day embryo) stained for neurofilament (NF), a neuronal protein (pink). B: The same dorsal root ganglion section subsequently stained for the mouse β-galactosidase transgene product (dark gray/black). Both A and B show residual X-gal staining (blue) identifying the enzymatic activity of the β-galactosidase transgene product. C: Control section adjacent to A and B stained with secondary antibodies only. Only remnant blue X-gal staining is visible. D: Control dorsal root ganglion (5-day embryo) double-stained for neurofilament (pink) and β-galactosidase (dark gray/black). This ganglion clearly contained no mouse-derived cells (no dark gray/black labeling). E: Mesonephros stained for β-galactosidase (dark gray/black). Some remnant blue X-gal staining is visible. F: The same mesonephros section subsequently stained for ferritin (pink), a kidney protein. G: Section containing a region of skeletal muscle stained for β-galactosidase (dark gray/black). H: The same section subsequently stained for sarcomeric α-actin (pink), a skeletal muscle protein. The same region of an adjacent section was positive for β-galactosidase and negative for neurofilament. I,J: Region of myocardium stained first for β-galactosidase (dark gray, I) and in subsequently stained for cardiac troponin I (pink, J). Scale bars = 50 μm in D, 10 μm in A–C,E–J.

Rat Olfactory Epithelial Cells Gave Rise to Cells of Hematopoietic Lineage In Vivo

Neural stem cells from the mouse forebrain produced hematopoietic cells after transplantation into bone marrow–irradiated mice (Bjornson et al., 1999). We sought to determine whether cells from the olfactory epithelium have a similar developmental potency. Accordingly cells dissociated from the olfactory epithelium were injected into the tail vein of irradiated rats, which were examined 9 months later. Cells dissociated from the olfactory lamina propria were injected into a separate group of animals. These experiments were performed on congenic strains of rat that allow unequivocal identification of cells deriving from donor when grafted into the host. Donor and host rats were from a congenic strain with RT7 leukocyte alloantigens (Kampinga et al., 1990). Donors were male, RT7.2-positive rats; hosts were female RT7.2-negative rats. The results were the same for animals injected with cells from either olfactory epithelium or lamina propria. Donor olfactory cell suspensions were negative for the CD45 RT7.2 antigen and for the CD34 antigen, indicating that they were free of leukocytes and free of hematopoietic stem cells (see below).

The leukocytes of all surviving host animals were immunopositive for the CD45RA antigen present on B lymphocytes in host and donor strains. This provided a positive control to confirm that all surviving host animals had B lymphocytes, a subset of which should express the donor RT7.2 antigen if the donor cells engrafted. The leukocytes of 9 host animals were immunopositive for the donor RT7.2 antigen: three injected with cells from lamina propria and six injected with cells from olfactory epithelium (Fig. 8). The proportion of donor-derived cells in leukocyte preparations from positive animals varied between 5 and 20%. This rate was not expected to be 100% because sublethal irradiation would not destroy all the endogenous hematopoietic stem cells. As another test of donor engraftment of hosts, genomic DNA was prepared from fixed leukocyte preparations and tested for the presence of the SRY gene using PCR. The leukocytes of the nine host animals positive for the donor RT7.2 antigen were also positive for the donor SRY gene (Fig. 9A). These results were confirmed using two independent sets of PCR primers. Products for both sets of primers from a positive host were sequenced and confirmed to be the SRY gene.

Figure 8.

Rat olfactory cells gave rise to leukocytes when transplanted into irradiated hosts. A: Donor-derived leukocytes in the blood of transplanted host animals (green cells immunostained for CD45 RT7.2 alloantigen superimposed on a differential interference contrast [DIC] image). B: Positive control. CD45 RT7.2-positive leukocytes in male donor blood (fluorescence image superimposed on DIC image). C: Negative control. CD45 RT7.2-negative leukocytes in female nontransplanted host blood (fluorescence image superimposed on DIC image). Scale bar = 20 μm in C (applies to A–C).

Figure 9.

Olfactory cells from male rat gave rise to leukocytes containing male SRY gene in female irradiated hosts. A: Polymerase chain reaction (PCR) for the male sex determining gene SRY in leukocyte preparations of male donor (positive control, lane 1), nontransplanted female (negative control, lane 2), and 13 female host animals transplanted with male olfactory cells (lanes 3–15). Nine are positive. Upper panel: glyceraldehyde-3-phosphate dehydrogenase (GAPDH, arrowhead, 265 bp). Lower panel: SRY (arrowhead, 310 bp). B–G: Transplanted male, RT7.2-positive donor rat olfactory cells gave rise to leukocytes in spleen of irradiated female hosts. B: Male donor-derived CD45 RT7.2-positive cells (green) in female host. C: Negative control. Lack of CD45 RT7.2-positive cells (green) in nontransplanted female. D: Positive control. CD45 RT7.2-positive cells (green) in male donor. E: In situ PCR showing male donor-derived SRY gene in spleen cells (red) in female host. F: Negative control. In situ PCR showing lack of SRY gene in spleen cells (red) in nontransplanted female. G: Positive control. In situ PCR showing SRY gene in spleen cells (red) in male donor. Not all cell nuclei were labeled in male controls probably because of variability in permeabilizing nuclei and exposing accessible regions of chromatin for the PCR template. The spleens did not contain olfactory marker protein, a protein of olfactory sensory neurons (Keller and Margolis, 1975) and, hence, a marker of donor olfactory epithelium (data not shown). Scale bars = 20 μm in D (applies to B–D), in G (applies to E–G).

The spleens of the positive hosts were examined for the presence of donor-derived hematopoietic cells (Fig. 9). Host spleens were assayed for the presence of male donor-derived cells using RT7.2 immunofluorescence and in situ PCR and hybridization for the SRY gene. Host and control spleens were cryosectioned at 8 μm and processed for in situ PCR for a target region of the SRY gene. In situ hybridization was then carried out using a probe designed from a segment of the SRY gene, which is internal to the region amplified by in situ PCR. This internal probe will not anneal to any false products of the in situ PCR process. For fluorescence detection, the probe was labeled with tetramethylrhodamine-5′-dUTP. Discrete regions of bright red fluorescence were detected in spleens of female hosts and male donor controls (Fig. 9). There was no fluorescence in female negative controls (uninjected females tested with in situ PCR and hybridization) or in male technical controls (no PCR, no probe, and nonsense probe controls). The in situ PCR and hybridization technique to identify SRY sometimes led to leakage of the PCR product from the nucleus. Nuclear localization of the signal was confirmed by counterstaining with the nuclear dye Hoechst Blue (Fig. 10). Detection of these markers was then combined using immunofluorescence to identify RT7.2 antigen-positive cells that were also SRY positive (Fig. 10). These observations demonstrate that host blood and spleen contained cells derived from the donor confirmed with two independent markers, a cell surface protein (RT7.2) and a nucleic acid (SRY), with some cells in host spleens that were double labeled for both markers.

Figure 10.

Olfactory cells from male rats gave rise to multiple types of leukocytes in the spleens of irradiated female hosts. Sections of female host spleens were triple labeled. AE: Immunolabeling for leukocytes. FJ: In situ polymerase chain reaction labeling for the SRY gene. KO: Nuclear labeling with Hoechst 33342. PT: Corresponding Nomarski images of the sections in each row. Leukocyte markers were RT7.2 (A), CD11b (B), CD3 (C), CD45RA (D), and CD34 (E). Scale bar = 10 μm in F (applies to A–K).

RT7.2 is an allele of CD45 expressed on leukocytes. We next examined whether donor-derived cells gave rise to cells expressing other CD markers indicative of cells of different hematopoietic lineages (Fig. 10). Donor-derived cells were identified using SRY in situ PCR and hybridization. In addition to SRY/CD45RT7.2 double-labeled cells, SRY was found in cells expressing CD45RA (an isoform found on B lymphocytes), CD3 (a marker of T lymphocytes), CD11b (a marker of myeloid cells), and CD34 (a marker of hematopoietic progenitor and endothelial cells). These antibodies are used widely and their specificity is well established (Woollett et al., 1985; Tanaka et al., 1989; Kampinga et al., 1990; Tamatani et al., 1993; Krause et al., 1994). These results indicate that the donor-derived cells gave rise to cells bearing markers of multiple leukocyte types.

Olfactory Cell Preparations Did Not Contain Hematopoietic Stem Cells

A potential confounding factor in these experiments is the presence of bone marrow–derived stem cells that may be present in the cell suspensions from olfactory epithelium and lamina propria. This possibility was tested directly by preparing cell suspensions of olfactory epithelium and lamina propria from eight male RT7.2 rats and testing these suspensions for the presence of antigens found on leukocytes or hematopoietic stem cells. As a positive control, leukocyte suspensions were prepared from the same animals. Cell suspensions were analyzed using fluorescence-activated cell sorting (FACS).

All CD antigens, including CD45 RT7.2 the donor animal leukocyte alloantigen, were present in the leukocyte suspensions. CD34 was present on 0.05% of control leukocytes. CD34 is a marker for hematopoietic progenitor cells and vascular endothelial cells. Only two CD antigens were present on cells from olfactory epithelium or lamina propria. CD11b is a marker for granular leukocytes, dendritic cells, and macrophages and was found on approximately 1% of cells from both olfactory epithelium and lamina propria. A total of 5% of cells from olfactory epithelium were immunopositive with the antibody Ox33 for a CD45RA isoform, which is expressed on B lymphocytes. The identities of these epithelial CD11b and CD45RA cells are not known. No other CD markers were found on cells from olfactory epithelium or lamina propria. No olfactory cells expressed the CD45 RT7.2 antigen.

FACS analysis was used to test cross-reactivity of the anti-mouse gamma globulin used as a secondary label. Cell fractions incubated with secondary antibody alone never exceeded the fluorescence of unlabeled cell fractions. None of 50,000 cells counted in the olfactory cell suspensions was positive for CD34, a marker of hematopoietic stem cells. These observations suggest that it is unlikely that the donor-derived leukocytes in the hosts were due to donor-derived hematopoietic stem cells present in the olfactory cell transplants. The possibility exists that the CD11b- or CD45RA-positive cells in the spleens of host animals might be cells that survived for 9 months from the donor cell suspensions. On the other hand, there may be nonhematopoietic cells in the olfactory epithelium that bear these markers.

The possibility of a hematopoietic origin for the stem-like properties of cells from the olfactory epithelium was investigated directly in a neurosphere-forming assay. Cells from rat olfactory epithelium and rat bone marrow were grown under identical conditions to produce neurospheres. Olfactory epithelial cultures produced neurospheres in 5 days. Bone marrow cultures failed to produce neurospheres in 22 days, although they remained viable and slowly proliferated.

Cell Fusion Assays

It is possible that the multipotency of olfactory cells might arise from fusion with other differentiated cells. This possibility was investigated in three conditions: coculture of mouse GFP-labeled olfactory cells with mouse DiI-labeled bone marrow cells, coculture of mouse GFP-labeled olfactory cells with DiI-labeled cells from chick embryo, and analysis of DNA content in human olfactory neurosphere cells after transplantation into the chick embryo.

In the first experiment, 1,129 green cells, derived from the mouse olfactory epithelium, were counted by visual inspection under the microscope. None were double labeled indicating no fusion with the red cells derived from the mouse bone marrow (Fig. 11). In the second experiment, 552 green cells, derived from the mouse olfactory lamina propria, were counted and none were double labeled red in the presence of the bone marrow cells in the cocultures. In the third experiment, cells derived from mouse olfactory epithelium (green) were cocultured with cell suspensions of chick gastrulae (red). A total of 1,104 green cells were counted and none were double labeled. In these experiments, three independent observers counted the cells. It is possible that fusion was not observed in these cultures because it was a very rare event (Ying et al., 2002). Accordingly, a fourth experiment was carried out using FACS. Cells from mouse olfactory mucosa were cocultured for 4 days in a 1:10 ratio of bone marrow cells (Fig. 11). None of the 25,000 cells examined were double labeled. Only noncellular highly autofluorescent doubly labeled particles were seen after gating and microscopic examination.

Figure 11.

Cell fusion assays were performed. Example image showing mouse bone marrow cell, labeled with 1,1′, di-octadecyl-3,3,3′,3′,-tetramethylindo-carbocyanine perchlorate (DiI, a fluorescent dye that incorporates into intracellular vesicular membranes after a period in vitro), after growing for 4 days in vitro with olfactory epithelial cell derived from green fluorescent protein (GFP) mouse. Cell fusion was not detected by visual examination or by fluorescence-activated cell sorting. Donor olfactory cells (GFP labeled) were cultured with DiI-labeled mouse bone marrow cells at a ratio of 1:10. A total of 25,000 cells were gated and DiI fluorescence vs. GFP fluorescence was plotted. From these, 90 “cells” registering high fluorescence in both channels were gated and collected for visual examination. Only noncellular highly autofluorescent particles were seen.

Behavior of human olfactory neurosphere cells was assessed in vivo after transplantation into chick gastrulae. The proliferation rate of these cells was calculated and the distribution of DNA contents in transplanted cells was compared with the same cells in vitro for the same period of time. In the first experiment, human neurosphere cells were labeled with CMFDA (a cytoplasmic green fluorophore). A total of 2,000 human cells were injected into each of four chicken gastrulae which were harvested 77 hr later and dissociated into single cells. After fixation and nuclear labeling with propidium iodide (PI), the cells from four embryos were pooled and analyzed using FACS. These embryos contained 45 million cells, of which 446,000 (0.97%) were derived from the green-labeled human olfactory neurosphere cells. Given the starting population of human cells, this finding indicates an approximate doubling time of 13.4 hr. This experiment was repeated twice more, once using CMFDA-labeled cells and transplanting for 5 days and once using GFP-labeled neurosphere cells and transplanting for 3 days (Table 2).

Table 2. Behavior of Cells Transplanted Into Chick Embryosa
 ModelbHuman cellsc (per embryo)Period transplantedTotal cells per embryo% human per embryo% green cells with PI = ChickG1d% Chick cells in G2% human cells in G2Doubling time
  • a

    FACS analysis reveals that chick and human cells had distinct behavior in embryo; human cells behaved similarly in vitro (Fig. 12).

  • b

    n = number of embryos. FACS, fluorescence-activated cell sorting; PI, propidium iodide.

  • c

    Number of human derived cells at harvest.

  • d

    Cells gated for green fluorescence were examined for PI fluorescence revealing their DNA content profile. A small percentage had the apparent DNA content of chick G1 cells. This small percentage was similar but less than in vitro.

Exp. 1CMFDA-human/ln chick (n = 4)11155077h115000000.975.329.113.513.4
Exp. 2CMFDA-human/ln chick (n = 2)643200120h335000001.920.831.69.214.5
Exp. 3GFP-human/ln chick (n = 12)14000077h250000000.56126.311.112.6
     Mean 1.15Mean 2.37Mean 29Mean 11Mean 13.5
  (per sample)in vitro SD 0.69SD 2.54SD 2.65SD 1.75SD 0.95
Exp. 1CMFDA-human/in vitro100000077h 1007.8 12.7 
Exp. 2CMFDA-human/in vitro1000000120h 1003.8 8 
Exp. 3GFP-human/in vitro100000077h 1003.1 8.1 
      Mean 4.9 Mean 9.6 
      SD 2.54 SD 2.69 
Figure 12.

Most human olfactory neurosphere-derived cells had normal DNA content 3 days after transplantation into 1-day-old chick gastrula. Host embryos were dissociated and subject to fluorescence-activated cell sorting analysis. A: Cells were gated for normal scatter. B: Cells gated in A were plotted for CMFDA fluorescence (human-derived, x-axis) and propidium iodide (PI, y-axis, DNA content). C: Histogram showing chick PI profile (pink; upper left quadrant of B gated) overlayed by human-derived PI profile (green; upper right quadrant of B gated). 1, chick G1 peak; 2, human G1 peak with chick G2 peak beneath; 3, human G2 peak. D: Comparison PI profile of human donor cells labeled with CMFDA and grown in vitro for 3 days.

PI fluorescence provided a measure of the DNA content of the cells, revealing two peaks coinciding with the G1 and G2 phases of the cell cycle (Fig. 12). The DNA profile of the green-labeled, human-derived cells was distinct from the DNA profile of the unlabeled cells in the chick embryo. The DNA profile of the green cells matched the DNA profile of human neurosphere cells grown for the same period in vitro (Fig. 12; Table 2). The DNA profile of the unlabeled cells matched the DNA profile of chicken embryos of similar age that were not transplanted with human olfactory cells. The DNA profiles of human and chicken cells were distinct, reflecting the smaller amount of DNA in the chicken genome. Another indication of the two distinct cell populations is the difference in the percentages of cells in G2 (Table 2), with the human cells behaving similarly in vivo and in vitro.

Fused cells (heterokaryons) may be multiploid, with genetic characteristics of both cell types. When fusion is provoked experimentally, the resulting heterokaryons share cellular components and membrane material. They are generally unstable and most chromosomes of one species or the other are eventually ejected, leaving nuclei that approximate one or other species cell type (Ruddle and Creagan, 1975). One indication of this process would be CMFDA- or GFP-containing cells with DNA content closer to the content in normal chicken cells. This was investigated by examining the green-labeled cells with small nuclei. Of the green cells, only 2.4% had nuclei the size of chicken G1 nuclei (mean of three experiments, using 18 embryos; Table 2), less than the proportion of these small nuclei in populations of human olfactory cells grown in vitro (mean, 4.9%).


We have shown in several complementary ways that the human adult olfactory mucosa contains cells with very broad developmental potency. Human olfactory mucosa yielded neurospheres that could be propagated as secondary and tertiary neurospheres. Primary, secondary, and tertiary neurospheres contained cells that differentiate into neurons and glia, depending on the culture conditions. Primary neurospheres also differentiated into cardiac, liver, and muscle cells in transwell induction assays. Neurospheres cloned from primary neurospheres demonstrated the same developmental potency as primary neurospheres. This developmental potency was demonstrated in vitro in the absence of differentiated cells with which to fuse. When transplanted into the early chicken embryo, human olfactory neurosphere cells integrated into many embryonic tissues, including developing cardiac atrium and ventricle.

The developmental potency of the stem-like cell in the mouse olfactory epithelium was investigated by transplantation of dissociated epithelial cells into the chicken embryo. Like human olfactory neurosphere cells, the mouse olfactory epithelial preparations contained cells that differentiated and integrated into numerous embryonic tissues, including developing nervous system, heart, liver, and kidney. Cells from rat olfactory epithelium and olfactory lamina propria appeared to repopulate the bone marrow of irradiated hosts, giving rise to numerous cells of the hematopoietic lineage. Several experiments indicate that the olfactory cell preparations do not contain hematopoietic stem cells that could account for the observations. Other experiments indicate that the observations are unlikely to be explained by fusion of olfactory donor cells with differentiated cells in vitro or vivo.

Neurospheres have been grown previously from human olfactory mucosa (Roisen et al., 2001; Zhang et al., 2004). The neurosphere-forming cells were analyzed. The cultures were composed of neurons and glia whose numbers varied with the culture conditions: 91–98% expressed β-tubulin III, 13–93% expressed A2B5 (a marker of immature glial cells), and 36–78% apparently expressed nestin (Zhang et al., 2004). An earlier study reported clonal “neuroblast” cultures generated from human olfactory mucosa with some cells expressing GFAP, some keratin, and some expressing various neuronal proteins, including proteins in olfactory sensory neuron transduction (Wolozin et al., 1992). These studies confirm the presence, in the human olfactory mucosa, of multipotent progenitors with a developmental trajectory favoring the neuronal lineage. In contrast, in the present study, in the absence of growth factors, development down the glial pathway was favored (Fig. 2), similar to mouse forebrain neurospheres (Reynolds and Weiss, 1992; Kilpatrick and Bartlett, 1995; Gritti et al., 1996). Human olfactory neurospheres, like mouse forebrain neurospheres also contained cells that were positive for nestin, β-tubulin III, and GFAP. Neurospheres with similar properties can also be generated from mouse and rat olfactory mucosa (unpublished observations).

Identity of the Olfactory “Stem Cell”

Our observations are not explained by hematopoietic stem cell contamination of the olfactory cell preparations. A priori, a significant contribution is unlikely because the olfactory epithelium is avascular and blood cells would wash from the lamina propria during the dissection procedures. On the other hand, macrophages, dendritic cells, and other leukocytes may migrate from, or be resident in, the olfactory epithelium or lamina propria. This was indicated by the small number of CD11b-positive cells, which could be macrophages or dendritic cells in olfactory cell preparations. There were also a small number of CD45RA-positive cells that may be B lymphocytes. It may be impossible to dismiss entirely the hypothesis that these cells are the “olfactory stem cells,” but given the role of the olfactory epithelium in protecting the central nervous system from infection (Mellert et al., 1992), it is not unexpected to find immune-type hematopoietic cells within the olfactory epithelium (Getchell et al., 2002). There were no CD34-positive hematopoietic stem cells in olfactory cell preparations and no cells positive for other CD markers. It is significant that when rat bone marrow cells and olfactory epithelial cells were grown in identical conditions, only the olfactory cultures produced neurospheres. This finding suggests that the culture conditions do not favor neurosphere formation from bone marrow cells. Additionally, when mouse leukocyte preparations were transplanted into the chicken embryo, under similar conditions that produce engraftment by olfactory epithelial cells, none of the leukocyte transplants led to engraftment. Taken together, these data indicate that the multipotency of the cells derived from the olfactory mucosa does not derive from any hematopoietic cells present.

The identity of the stem cell responsible for our observations is not known. Within the olfactory mucosa, there are two main candidates, both basal cells of the olfactory epithelium. The globose basal cells are a mixed population of cells. Some are the immediate neuronal progenitors that can differentiate into the olfactory sensory neurons (Caggiano et al., 1994; Goldstein et al., 1998; Huard et al., 1998; Newman et al., 2000). Recent transplantation experiments confirm that the globose basal population also contains a multipotent progenitor that can differentiate into the non-neuronal cells of the olfactory epithelium, namely the supporting cells and the cells of the ducts of the Bowman's glands, the specialized olfactory glands (Chen et al., 2004). They also appear to produce the other basal cell, the horizontal basal cell (Huard et al., 1998). When the horizontal basal cell is isolated and grown in vitro it can be induced to differentiate into both neurons and glia, suggesting that it may be a multipotent progenitor (Carter et al., 2004). It is possible that both these basal cells are multipotent, but the lineage relationship between them remains unclear if both can produce neurons and globose basal cells can produce horizontal basal cells.

Our experiments indicate that cell preparations from both olfactory epithelium and lamina propria may be multipotent, although the method of separation of these tissues does not guarantee that cells from the other tissue are not present. The epithelium and lamina propria are dissected by enzymatic treatment and gentle mechanical separation (Feron et al., 1999). This separation takes place along the basement membrane. If the stem cell is close to the basement membrane it could contribute to cell suspensions of both epithelium and lamina propria. Neural stem cells express nestin (Lendahl et al., 1990), although it is not exclusively a marker for neural stem cells. In the olfactory epithelium, the supporting cells are nestin-positive in their end-feet close to the basement membrane (Doyle et al., 2001). It is possible that these cells are the neural stem cell, but this lineage is unlikely because retroviral lineage studies in the olfactory epithelium were not able to demonstrate clones of cells containing supporting cells and other cell types (Caggiano et al., 1994; Huard et al., 1998). It is possible that olfactory ensheathing cells in the lamina propria are the neural stem cell, because they express nestin (Osada et al., 1995; Sonigra et al., 1999; Doyle et al., 2001) and their presence in epithelial suspensions cannot be excluded.

Multipotency of Olfactory “Stem Cells” Is Not Explained by Fusion

There is vigorous debate about the ability of adult, somatically derived cells to give rise to progeny of multiple lineages (Wurmser and Gage, 2002). There is an argument that adult cells with “stem-like” abilities result from fusion with other stem cells or with already differentiated cells (Alvarez-Dolado et al., 2003; Vassilopoulos et al., 2003; Wang et al., 2003; Ogle et al., 2004), although there is also evidence to the contrary (Cogle et al., 2004; Harris et al., 2004; Pochampally et al., 2004). Bone marrow–derived stem cells were shown to repair injured muscle in two ways: by fusion with established myotubes as well as by conversion to satellite cells (LaBarge and Blau, 2002), leading to a view that fusion may be stimulated by injury. Perhaps the most sophisticated methods to assess this phenomenon have been two studies using cre/lox recombinase and reporter constructs to test for fusion (Alvarez-Dolado et al., 2003; Harris et al., 2004). In both studies, fusion was a rare event. One study found fusion occurred in up to 1 of 10,000 liver cells examined, in up to 1 of 300,000 brain cells examined, and in up to 1 of 100 cardiomyocytes examined. No fused cells were found in other tissues (Alvarez-Dolado et al., 2003). The higher frequency of fusion in liver and cardiac cells might reflect the fusion that normally occurs in cells of these tissues. In the other study, there was little incidence of fusion amongst robust development of epithelial cells from bone marrow–derived cells in lung, liver, and skin (Wurmser et al., 2004).

Embryonic stem cells can fuse in vitro with adult bone marrow cells (Terada et al., 2002) and with adult forebrain neural stem cells (Ying et al., 2002). Fused cells (heterokaryons) may be multiploid, with genetic characteristics of both cell types. Heterokaryons are generally unstable, and most chromosomes of one species or the other are eventually ejected leaving nuclei that approximate one or other species cell type (Ruddle and Creagan, 1975). Stem cell fusion occurs at a low frequency in vitro (Ying et al., 2002) and appears to require the presence of embryonic stem cells (Terada et al., 2002; Ying et al., 2002).

Fusion is excluded as an explanation for the multipotency of human olfactory neurospheres in vitro, because they were not in contact with any other cells. In agreement is a recent in vitro study demonstrating that neural stem cells can differentiate along an endothelial lineage without any evidence of fusion (Wurmser et al., 2004). Multipotency is confirmed, therefore, for olfactory cells in vitro, but can fusion explain the “multipotency” of transplanted olfactory cells?

Examination of the DNA content of human cells after transplantation into the chicken embryo indicated that the transplanted human cells were a distinct population within the chick embryo. Their DNA profile was distinct from the chick cells and similar to human cells grown for the same period in vitro. The percentage of cells in G2 was similar in all the human cells and approximately 40% of the percentage among the chick cells. Consideration of the doubling times of the human cells in the chick embryo suggests that they are proliferating at a normal rate for mammalian cells. These doubling times indicate that it is unlikely that the green cells at the time of harvesting derived from fusion of human cells with a few founder chick cells, because the calculated doubling times would be much smaller to produce the same total number of green cells within the growing period. Thus, the distinct behavior and DNA profile of the human cells suggests that the majority are not fused with chick cells.

Another way to assess fusion in these embryos is to examine the proportion of small labeled cells of similar DNA content to the unlabeled chick cells in G1. Only 2.4% of the labeled, human-derived cells had a DNA content comparable to the unlabeled chicken cells in G1. By comparison, this proportion was 4.9% in the same human olfactory neurosphere cells grown for a similar period in vitro. These observations confirm that “fusion” is not required to explain the distribution of DNA content in the transplanted human cells. In agreement with this finding, a recent study of rat-chick embryo chimeras found no evidence of fusion after transplantation of bone marrow stem cells (Pochampally et al., 2004). Therefore, although fusion cannot be excluded in the chicken embryo, we found little evidence for it that cannot be explained by normal variation in DNA content.

Are Adult Olfactory Neural Stem Cells Pluripotent?

There is debate about whether adult stem-like cells are pluripotent and whether they can differentiate into cells outside the tissues from which they are derived (Alison et al., 2003). The reconstitution of blood by neural stem cell transplantation (Bjornson et al., 1999) was confirmed by one group (Shih et al., 2001) but not another (Morshead et al., 2002). The results of the present study are in agreement with the earlier studies (Bjornson et al., 1999; Shih et al., 2001) and indicate that hematopoietic contamination is unlikely to explain them (Morshead et al., 2002). The present results also confirm the engraftment of the chicken embryo by neural stem cells (Clarke et al., 2000). In contrast, when neural stem cells were transplanted into the mouse blastocyst, they gave rise only to glial cells (Greco et al., 2004). Differences in lineage restriction of neurosphere cells in published studies may be defined by extracellular environment, as observed for human neurospheres in the present study, or it may reflect differing sampling and culturing techniques. FACS, for example, is known to prevent nuclear cloning (Mombaerts, 2004).

It is interesting to consider the evidence for adult stem cell potential against the concepts of “specification” and “determination” that propose that progenitor cells in adult tissues are lineage restricted to tissue-type developmental pathways. These concepts arose from experiments where portions of tissue were explanted or transplanted rather than dissociated cells. Tissue explants carry with them a microenvironment, and after transplantation, this microenvironment would dominate the phenotypic response of resident stem cells. In contrast, adult stem cells are transplanted as dissociated or isolated cells that would be open to influence from the microenvironment into which they were transplanted. Certainly olfactory neurosphere cells can be directed along numerous developmental lineages by varying the culture conditions with defined media or tissue-conditioned medium. One could describe adult stem cells as “primitive programmable precursor cells” necessary for the redevelopment of their tissue niche after injury. There are now reports of “stem-like” cells in numerous adult tissues, including dermis, tooth pulp, hair cell follicle, gut, and adipose tissue (Kamimura et al., 1997; Deasy et al., 2001; Toma et al., 2001; Zuk et al., 2001; Bjerknes and Cheng, 2002; Mina and Braut, 2004). The present study adds the olfactory mucosa to this list and demonstrates this “stem-like” potency without any experimental manipulation in vitro other than dissociation into single cell suspension. In the mouse and rat transplants described here, there is no prolonged period in vitro that might alter lineage potential (Anderson et al., 2001).

The multipotency of olfactory cells was apparent in the high proportion of successful transplant experiments: 53% of surviving host rats had donor-derived leukocytes and 28% of surviving host chickens had mouse cells within their tissues. We cannot say from these experiments whether the multipotent olfactory cell transdifferentiates to overcome its normal, organ-specific lineage restrictions or whether it is developmentally equivalent to an embryonic stem cell (Watt and Hogan, 2000). It is possible to explain some aspects of our transplant experiments by invoking transdifferentiation rather than differentiation from a stem-like progenitor (Anderson et al., 2001), but the demonstration of multipotency in clonally derived neurosphere cells strongly supports the conclusion that the adult olfactory mucosa contains a multipotent, perhaps pluripotent, stem cell. These qualities suggest olfactory neural stem cells as candidates for autologous transplantation for tissue repair. The human olfactory mucosa is readily accessible by biopsy without affecting the sense of smell (Féron et al., 1998, 1999) and olfactory neurogenesis occurs in elderly persons (Murrell et al., 1996). Olfactory cells can be grown from representatives of healthy controls and various patient groups (Wolozin et al., 1992; Féron et al., 1998) potentially providing a source of stem cell lines for genomic and proteomic analysis to study disease etiology and for drug discovery and design.


Human Neurosphere Culture

Human nasal mucosa was obtained by biopsy during routine nasal surgery under general anesthesia, using an ethmoid forceps (Richard Wolf Medical, number 8211.551; Hoyland Medical; Féron et al., 1998). Fifty biopsies of approximately 1–2 mm2 were obtained from 25 individuals (11 males and 14 females aged 20–78 years of age), undergoing surgery for septoplasty or turbinectomy. All biopsy tissues were obtained with the informed consent of the patients, and the study carried out under a protocol that was approved by the ethics committees of the hospital and university according to guidelines of the National Health and Medical Research Council of Australia.

Biopsies were immediately placed on ice in DMEM/HAM F12 (Invitrogen) supplemented with 10% fetal calf serum, penicillin, and streptomycin and then incubated for 45 min at 37°C in a 2.4 units/ml Dispase II solution (Boehringer). Laminae propriae were carefully separated from the epithelium under a dissection microscope with a microspatula. Sheets of olfactory epithelium were mechanically dissociated while lamina propriae were cut into pieces of approximately 40 μm2 using a McIlwain chopper (Brinkmann) and incubated in a 0.25 mg/ml collagenase H solution (Sigma) for 10 min at 37°C. After mechanical trituration, the enzymatic activity was stopped using a 0.5 mM ethylenediaminetetraacetic acid solution (Invitrogen). Cell pellets of both tissues were resuspended in DMEM/HAM F12 culture medium containing 10% fetal calf serum plus penicillin/streptomycin and sequentially plated into flasks pretreated with poly-L-lysine (1 μg/cm2; Sigma). Eighteen hours after the initial plating, floating cells and undigested pieces of epithelium and lamina propria were transferred to other coated wells. This operation was repeated 24 hr later. Spheres of cells were harvested either collectively by aspiration of the culture medium and subsequent centrifugation or individually using a 2-μl pipette. They were then frozen in 90%serum/10% dimethyl sulfoxide for later use or immediately plated on glass or plastic dishes without or with a coating of collagen IV (5 μg/cm2), fibronectin (10 μg/cm2), laminin (3.5 μg/cm2), poly-L-lysine (2 μg/cm2), or poly-ornithine (10 μg/cm2). Various growth factors were tested: FGF2 (50 ng/ml), TGFα (10 ng/ml), CNTF (25 ng/ml), NGF (50 ng/ml), NT3 (50 ng/ml), and retinoic acid (0.1 and 1 μg/ml).

Nestin mRNA expression.

Biopsies and neurospheres were collected and purified, total RNA was isolated using guanidinium thiocyanate phenol chloroform extraction followed by DNase digestion of genomic DNA before being converted into cDNA with the SuperScript Choice system (Invitrogen) using an oligo-dT primer. Primers used were as follows: forward, GAGAGGGAGGACAAAGTCCC; reverse, TCCCTCAGAGACTAGCGCAT. Cycle profile was 3 min at 94°C, 40 cycles of 30 sec at 94°C, 60 sec at 59°C, 60 sec at 72°C, and 6 min at 72°C. No RT controls confirmed the absence of any contaminating genomic DNA.


For cell counting procedures, immunochemistry was performed using peroxidase and alkaline phosphatase conjugated secondary antibodies (Sigma) with diaminobenzidine and Fast red as chromogens. Otherwise, fluorescent secondary antibodies were used. The following antibodies were used: anti-nestin (1:1,000, gift of Dr. Lendahl, University of Lund), anti–neuron-specific β-tubulin type III (5 μg/ml, Sigma), anti-MAP5 (75 μg/ml, Sigma), anti-neurofilament (1:400, Sigma), anti-GFAP (8 μg/ml, DAKO), anti-O4 (18 μg/ml, Chemicon), anti-galactocerebroside (1:50, Sigma), anti-ferritin (1:200, Sigma), anti-human serum albumin (1:500, Sigma), anti-skeletal myosin (1:200, Chemicon), anti-striated tropomyosin (1:50, Sigma), anti-human cardiac troponin I (16 μg/ml, Chemicon), anti-human cyclin E (1:100, Santa Cruz), and anti-sarcomeric α-actin (1:200, Sigma). Cells were not permeabilized for O4 staining. The specificity of all primary antibodies used was confirmed using controls incubated with nonimmune sera of the species used to raise those antibodies. Positive and negative controls were performed for each antibody. In some experiments, cyclin E detection was enhanced using a Tyramide Signal Amplification kit (NEN Life Science Products) and fluorescein.

Transwell induction assay.

Neonate rats were killed by CO2 asphyxiation. Heart, liver, and leg skeletal muscle were removed, washed, and chopped into 1-mm slices using a McIlwain chopper (Brinkmann). Sections were placed on polycarbonate inserts (pore size, 3 μm) and inserted in transwell culture dishes (Costar). Medium was added level with the tissue slices: DMEM-HAM F12 supplemented with insulin/transferrin/selenium or 5% FCS. Individual neurospheres were carefully plated on the wells underneath but separated from the overlying sections. Four days later, the inserts containing the sections were removed and the cells were fixed and processed for immunochemistry. For experimental controls, neurospheres were plated in wells without inserts and cultured in DMEM-HAM F12 and 5% FCS.

Transplantation of Human Neurosphere Cells Into Chick Embryo

Human olfactory neurospheres were enzymatically dissociated and labeled with DiI or CMFDA (Molecular Probes) according to the manufacturer's instructions. Fertile white Leghorn eggs were incubated at 38°C for approximately 20–24 hr. Shells above the air space region were removed and dissociated olfactory cells were injected into the region of the primitive streak using a micromanipulator (Narshige Model MN151) and drawn glass micropipette. Two microliters of cells (1,000 cells/μl) were injected. Eggs were then resealed and opened for processing 2–5 days later.

Transplantation of Mouse Olfactory Epithelial Cells Into Chick Embryo

The olfactory mucosa was dissected from donor animals and the olfactory epithelium was separated from the lamina propria as described for the human biopsies above. Dissociated epithelial cell suspensions were prepared and tested for viability by trypan blue uptake. The donor animals were from a transgenic mouse strain, which has the E. coli Lac Z gene under the promoter of the HMGCoA Reductase gene (Tam and Tan, 1992).

Chick gastrulae (20–24 hr incubation, Hamburger Hamilton stage 4 to 6) were transplanted in situ by microinjection of a few microliters of dissociated olfactory epithelium cells (1,000 cells/μl) into the primitive streak as described (Clarke et al., 2000). Controls were (1) no interference, (2) sham injected, and (3) injected with a suspension of erythrocyte-free blood cells (leukocytes, bone marrow–derived stem cells, and progenitors; also 1,000 cells/μl; prepared using a Vitalyse erythrocyte lysing kit [Bioergonomics] according to the manufacturer's instructions) from the same transgenic donor mice. Two to 5 days later, the embryos were fixed and processed for X-gal staining to identify bacterial β-galactosidase activity.

Embryos were fixed at 4°C for 30 min in 4% paraformaldehyde in PME buffer (PME: 100 mM sodium phosphate buffer, pH 7.8, 2 mM MgCl2, 5 mM ethyleneglycoltetraacetic acid). Embryos were washed in wash buffer (0.01% sodium deoxycholate, 0.02% Nonidet P40 in PME) twice for 20 min at room temperature, followed by one wash for 60 min at 50°C. Embryos were then incubated in staining buffer (5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 1 mg/ml X-gal in Wash buffer) for 20 hr at 32°C. Embryos were washed three times in wash buffer and post-fixed for 3 hr in 4% paraformaldehyde. They were then washed and dehydrated through a graded series of alcohols to prevent background staining occurring. Embryos could then be stored in 100% ethanol at 4°C until further processing. Some embryos had excess blue precipitate cleared in methyl salicylate after dehydration.

Before sectioning, the embryos were rehydrated through a graded series of alcohols to phosphate buffered saline (PBS) and allowed to equilibrate in 30% sucrose in PBS overnight. The embryos were then frozen in OCT and sectioned at 8 μm on a cryostat. Processing of sections for detection of β-galactosidase, skeletal myosin heavy chain, cardiac troponin I, ferritin, neurofilament, S100, olfactory marker protein, and cyclin E was undertaken using standard methods for single and double labeling. Secondary antibodies used were conjugated for fluorescence detection, Fast Red conversion by alkaline phosphatase, or diaminobenzidine conversion by horse radish peroxidase. Antibodies and their respective dilutions were as follows: mouse anti-Nestin monoclonal (MAB353, Chemicon, 1:100), mouse anti–E. coli β-galactosidase monoclonal (Gal40, Sigma, 1:100), rabbit anti-bacterial β-galactosidase polyclonal (AB986, Chemicon, 1:1,000), rabbit anti-human ferritin polyclonal (F5012, Sigma, 1:200), rabbit anti-human skeletal myosin polyclonal (AB998, Chemicon, 1:100), mouse anti-human cardiac troponin I monoclonal (MAB3150, Chemicon, 1:200), rabbit anti-cow S100 polyclonal (Z0311, Dako, 1:200), mouse anti-neurofilament monoclonal (NO142, 1:400), goat anti-rodent olfactory marker protein polyclonal antibody (a generous gift of Dr. Frank Margolis, 1:5000; Keller and Margolis, 1975), goat anti-rabbit IgG fluorescein isothiocyanate (FITC) conjugate (AP132F, Chemicon, 1:200), rabbit anti goat IgG(H&L) FITC conjugate (FI-5000, Vector, 1:200), goat anti-mouse IgG rhodamine conjugate (AP132R, Chemicon, 1:200), goat anti-rabbit alkaline phosphatase conjugate (AL13405, Bioscience, 1:200), goat anti-mouse alkaline phosphatase conjugate (AM13405, Bioscience, 1:200), goat anti-Mouse IgG(H&L) horseradish peroxidase conjugate (172-1011, BioRad, 1:200), and goat anti-rabbit IgG(H&L) horse radish peroxidase conjugate (170-6515, BioRad, 1:200).

Transplantation of Rat Olfactory Mucosal Cells Into Irradiated Rat

Donor and host rats were from a congenic strain with RT7 leukocyte alloantigens (Kampinga et al., 1990). Donors were male, RT7.2-positive rats whose olfactory mucosa was dissected from the donors and enzymatically divided into olfactory epithelium and underlying lamina propria and cell suspensions were prepared as described above. Cells from 10 donors were injected into the tail veins of 20 virgin female hosts—10 received cells from olfactory epithelium and 10 from lamina propria. Approximately 106 donor cells (60–70% viable) were injected into each host animal. Hosts were female, RT7.2-negative rats, which were irradiated with 700 rads using a Phillips Deep X Ray unit (250 Kilovolts/15 milliamps), sublethal irradiation that destroys most of the bone marrow. At 4 days before and 10 days afterward, the rats received subcutaneous injections of gentamicin (0.1 mg/kg) twice daily. Nine months after transplantation, 17 of 20 host animals remained alive. These were killed with an overdose of sodium pentobarbitone, their blood was collected, and their spleens were dissected, fixed, and frozen for later analysis. Donor-derived cells were distinguished from host cells after transplantation by the expression of the RT7.2 allele of CD45 antigen on the surface of leukocytes and by the presence of the male sex-determining gene SRY (Koopman et al., 1991).

Leukocyte immunochemistry and PCR.

Leukocytes were prepared from the blood by using Vitalyse (Bioergonomics) erythrocyte lysing kit according to the manufacturer's instructions. Control and host animal leukocytes were subjected to indirect immunofluorescence detection of CD45 RT7.2 (the leukocyte common antigen on all RT7.2 donor animals' leukocytes) and CD45 RA (an antigen on all B lymphocytes). At the time of transplant, aliquots of freshly prepared donor cells were subjected to indirect immunofluorescence detection of CD45RT7.2 and CD34 and then fixed using the Vitalyse fixing buffer. Primary antibodies were His 41 (Serotec, anti RT7.2, 1:10), Ox 33 (Pharmingen, anti-CD45RA, 1:100), and ICO115 (Santa Cruz, anti-CD34, 1:50). The secondary antibody used was Oregon green 488 conjugated anti-mouse IgG A&M preabsorbed against other species (Molecular Probes, 1:200). For immunodetection, cells were allowed to settle on poly-L-lysine–coated slides and viewed with an Olympus BX50 fluorescence microscope fitted with an Apogee KX85 digital camera and digital software (PMIS, Apogee). For FACS experiments, cells were labeled in like manner, fixed, and filtered (70 μm) to remove any cell clusters.

Genomic DNA was prepared from fixed leukocyte cell suspensions using proteinase K digestion followed by phenol chloroform extraction according to standard methods. PCR primers were designed from a published 459-bp sequence from the rat SRY cDNA (accession no. X89730). The following primer pairs proved reliable: SRYb forward, 5′-CTACAGCCTGAGGACATATTA-3′; SRYb reverse, 5′-TCCGTATATAATAGTGTGTAG-3′; and SRYc forward, 5′-GGAGCAGTGACAGTTGTCTAG-3′ and SRYc reverse, 5′-GAGGCAACTTCACGCTGCAA-3′. The program used for testing leukocyte DNA consisted of a preliminary denaturing step of 95°C for 3 min, followed by 45 cycles of 95°C denaturing for 30 sec, an annealing step of 60°C for 60 sec, and 72°C extension for 60 sec. A 6-min extension step at 72°C was included at the end. Amplification of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) under the same conditions was used as a positive control and measure of DNA quality. GAPDH primers were as follows: forward, 5′-ACAGTCCATGCCATCACTGCC-3′; and reverse, 5′-CCTGCTTCACCACCTTCTTG-3′. These experiments used a MJ Research MiniCycler model PTC-0150. Two differently primed PCR products from a test animal were sequenced using ABI Big Dye. As well, a product was cloned into pGem T-easy (Promega) for probe generation. Suitable quantities of a 310-bp product were gel purified and labeled with tetramethylrhodamine-5′-dUTP using a Random Hexamer Labeling Kit (Roche).

Spleen immunochemistry and PCR.

Spleens were fixed in PLP fixative (2% paraformaldehyde, 67 mM lysine, 10 mM sodium periodate in phosphate buffer; McClean and Nakane, 1974) for 2 hr at 4°C. Spleens were washed in PBS and equilibrated in 30% sucrose in PBS before freezing on a cryostat in OCT medium (Sakura). Eight micrometer sections were cut and mounted on Super Frost Plus precoated microscope slides. Sections were blocked in 10% goat serum, 1% anti-mouse IgG, A&M in PBS. Blocking solution was removed and primary antibodies to rat CDs (CD45RT7.2, CD45RA, CD3 [IF4, Serotec, 1:50], CD11b] WT.5, Pharmingen, 1:50], and CD34) were applied in 10% goat serum in PBS. Sections were washed in PBS and secondary Oregon Green anti-mouse conjugate applied in a like manner. Sections were washed in PBS and 1% Tween 20 and were viewed for fluorescence as above. For testing for the presence of olfactory marker protein, anti-rodent olfactory marker protein, goat polyclonal antibody (a generous gift of Dr. Frank Margolis, 1:5,000; Keller and Margolis, 1975) was used. In this case, the secondary was rabbit anti-goat IgG (H&L) fluorescein conjugate (Vector, 1:200) and blocking was carried out using rabbit serum.

Spleen sections were subjected to 1μg/ml proteinase K digestion for 10 min. At this point, in colocalization experiments, immunofluorescent detection of CD markers took place as described above, after which sections were post-fixed in 4% paraformaldehyde for 5 min at 4°C. Sections were then membrane permeabilized in 100% methanol for 30 min at room temperature. A Hybaid Omnigene fitted with an in situ humidified PCR block was then used for the in situ PCR procedure. Sections were circumscribed with PAP pen, and 20 μl of reaction mix prewarmed to 82°C was applied to each. Round coverslips were applied to each section followed by an overlay of mineral oil and a rectangular coverslip. The reaction mix consisted of 1 μM each of primers SRYb forward and SRYb reverse, 200 μM DNTPs, 4.5 mM MgCl2, 1× reaction buffer, and 7.5 units of Amplitaq (Roche). The following PCR program was used: 1 cycle of 94°C, 3 min followed by 55°C, 2 min; and then 29 cycles of 94°C, 1 min, followed by 55°C, 2 min. Rectangular coverslips were removed in xylene; slides were then immersed in isopropanol followed by a graded series of ethanol until round coverslips could be removed in PBS. DNA in sections was denatured by immersion in 0.1 N NaOH for 90 sec. Slides were washed in 2× standard saline citrate (SSC) and 200 ng/ml of probe (SRY or nonsense probe) applied in hybridization buffer (40% DI formamide, 10% dextran sulphate, 1× Denhardt′s solution, 5×SSC, and 100 μg/ml fragmented denatured herring sperm DNA). Sections were hybridized overnight at 37°C. Sections were washed in 2×SSC, 50°C for 5 min twice. Slides were then counterstained in Hoechst blue (Hoechst), mounted, and photographed.

Preparation of Rat and Mouse Bone Marrow Cells

The long bones of the hind limbs were dissected clear of attached muscles and soft tissues. The ends of the femurs and tibias were excised. Each bone was then flushed with 5% fetal calf serum in PBS to remove the bone marrow. A cell suspension was then created by repeated gentle aspiration and flushing through a 22-gauge syringe into a Petri dish. The suspension was then filtered through a 70-μm cell filter. The cell suspension was washed and resuspended in appropriate medium. These cell suspensions were used as source material for neurosphere-forming assays, and as source material for the fusion assays (below).

Cell Fusion Assays

Experiments were undertaken to test whether olfactory cells might fuse with other cells: coculture of mouse olfactory cells with mouse bone marrow cells, coculture of mouse olfactory cells with cells from chick embryo, and analysis of DNA content in human olfactory neurosphere cells after transplantation into the chick embryo.

Coculture of mouse olfactory cells with mouse bone marrow cells.

The mouse olfactory cells were derived from a GFP mouse (GFP-A8, base strain: BALB C, a generous gift from Dr. Klaus Matthaei, Australian National University), which has EGFP expression under the H2A.Z promotor (Faast et al., 2001) that directs EGFP expression to several tissues, including the olfactory mucosa. Cell suspensions of mouse olfactory epithelium and lamina propria were prepared as described above, and then cocultured for 4 days with mouse (BALB/C) bone marrow cells at high density at a ratio of 1:10. The bone marrow cells were labeled with 1 μg/ml DiI (Molecular Probes), a membrane-bound red fluorophore, according to the manufacturer's instructions, such that any fused cells should be identifiable by their double fluorescence: green (olfactory cells) and red (chick embryo or bone marrow cells). Cells were then examined visually or by FACS for any GFP/DiI double labeling.

Coculture of mouse olfactory epithelial cells with chick embryo cells.

Cell suspensions of mouse olfactory epithelium were prepared as described above, and then cocultured for 4 days with cell suspensions prepared from the chick embryo at high density at a ratio of 1:10. Chicken gastrulae (Hamburger Hamilton stage 4–5) were dissociated by a brief trituration and incubation in 0.25% trypsin in Earle's balanced salt solution (EBSS) at 37°C. The chick embryo cells were labeled with DiI such that any fused cells should be identifiable by their double fluorescence: green (olfactory epithelial cells) and red (chick embryo cells). Cells were then examined visually or by FACS for any GFP/DiI double labeling. Bone marrow and chick embryo cells remained viable for the culture period as assessed by membrane integrity (trypan blue) and intact nuclei (Hoechst blue).

Retroviral labeling of olfactory cells for transplant.

Transduction experiments used a pFB-hrGFP, Viraport (Stratagene) retroviral control supernatant that contains an MMLV replication defective retrovirus. As a precaution, cells targeted for transduction were tested using a reverse transcriptase assay in case of endogenous replication competent retrovirus. When confirmed to be clear of endogenous retrovirus, they were transduced according to the manufacturer's instructions. PCR was then used to confirm genomic integration of the viral insert. Neurospheres obtained from GFP-transduced cultures were tested capable of producing both neurons and glia before use in transplant experiments.

Dissociation of chick–human chimeric embryos for FACS analysis.

Chick embryos were harvested at 4 or 6 days development and dissected in EBSS to 3-mm cubes of tissue. Tissue was washed in the same medium then incubated in 0.25% trypsin in EBSS at 4°C overnight. Excess trypsin was then removed, and tissue was triturated at 37°C until microscopically clearly dissociated. The reaction was stopped in 10% FCS, and the cells were washed twice before fixing and permeabilization for 15 min in 70% ethanol in PBS. Samples were then stained with 1 μg/ml PI (Molecular Probes), according to the manufacturer's instructions.


Flow cytometry used a FACS Vantage (BD Biosciences) cell sorter equipped with a 488-nm argon ion laser, and sample data were analyzed using WinMDI version 2.8 software.


Photographic and electronic images were processed using Adobe Photoshop 7.0.


The authors thank U. Lendahl for the anti-human nestin antibody, F. Margolis for the anti-rodent olfactory marker protein antibody, M. Schirmer for assistance with irradiation, S. Tan, K. Matthei, and David Tremethick for transgenic mice, K. Field and I. Hayward for assistance with FACS analysis, and J. Kan for technical assistance.