Locally Born Olfactory Bulb Stem Cells Proliferate in Response to Insulin-Related Factors and Require Endogenous Insulin-Like Growth Factor-I for Differentiation into Neurons and Glia

Isolation and characterization of stem cells from the E12.5–E14.5 olfactory bulb

To isolate local olfactory bulb stem cells, a cell suspension was prepared from mouse E14.5 OB (Fig.1A) and plated under tissue culture conditions in the presence of FGF-2 plus EGF and insulin. One day after plating, many cells in the culture had immature neuronal morphology (Fig. 1B, arrowheads) bearing several neurites. A smaller proportion of the cells showed a rounded morphology (arrows). Daily visualization of the cultures indicated that the rounded cells divided, forming cell aggregates or spheres (Reynolds and Weiss, 1996) (Fig.1B) that grow to confluence in the presence of the mitogens by day 4 (Fig. 1D,E). Because rostral migration of SVZ-derived proliferative neuroblasts to the mouse OB may occur from E14.5 but not at earlier ages (Hinds, 1968b), cell suspensions were also prepared from the E13.5 and E12.5 OB. After plating, the E13.5- and E12.5-derived cells proliferated and formed spheres in response to FGF-2 plus EGF, with the same temporal pattern as the E14.5 OB-derived cultures (Fig.1F,G). During the first days after the original cell suspension was plated, FGF-2 but not EGF promoted cell proliferation, although the combination of both factors gave the maximum number of BrdU-positive cells (Fig. 1H). After passages 1–2, cells could be grown in either factor alone, but the yield of proliferative cells was still greater when both factors were added together (data not shown). These results support the concept that FGF-2-responsive stem cells are born early in embryonic development and that they give rise to EGF-responding stem cells at later stages (Ciccolini and Svendsen, 1998; Tropepe et al., 2001).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Cells that respond to FGF-2 plus EGF by proliferating and forming spheres can be isolated from the embryonic olfactory bulb. Cells prepared from the OB of E12.5, E13.5, and E14.5 mice were plated in the presence of 20 ng/ml FGF-2, 20 ng/ml EGF, and 10 μg/ml insulin. A, Dorsal view of a E14.5 mouse brain. The lines mark the plane of the OB dissection.B, One day after plating, some cells with rounded morphology (arrows) were surrounded by cells with immature neuronal morphology (arrowheads).C, Day 2, the rounded cells divided and formed small aggregates or spheres, which detached from the bottom of the plate and floated. D, E, Days 3–5, the spheres increased in size as a result of additional cell division and aggregation. F, G, Large spheres were also obtained from E13.5 (F) and E12.5 (G) cell cultures at days 4–5. There were small variations between cultures in the timing of sphere formation. H, Effect of various concentrations of FGF-2 and EGF, or FGF-2 plus EGF, on BrdU incorporation by E14.5 OB precursor cells. Results are the mean ± SEM of data from four cultures. Scale bar, 40 μm.

Under these standard growth conditions, cells could be passaged every 3–5 d for at least 5.5 months (40 passages), representing ∼150 cell doublings (Fig. 2A). The average doubling time of these cells was approximately 26 hr. The x-fold increase in cell number varied between passages, but no tendency was observed toward lower numbers in the last passages. If all cells had been passaged, a theoretical estimation of the total yield of nestin-positive cells resulted in 10²⁵cells at 3 months and 10⁴⁵ at 5.5 months. Cell proliferation required FGF-2 plus EGF even after long periods in culture, indicating that they were not transformed into growth factor-independent cells (data not shown). In cultures growing in the presence of FGF-2 plus EGF and insulin for up to 3 months (passage 23), 99.2% of cells expressed nestin (Fig.2B,C), a marker of neuroepithelial cells, and <0.1% expressed antigen markers for neurons (β-III-tubulin), astrocytes (GFAP), or oligodendrocytes (O4) [for references on cell markers, see Vicario-Abejón et al. (1995) andJohe et al. (1996)]. On average, 86.5% of cells incorporated BrdU (Fig. 2B,D); this value was 96.2% at passage 15, 81.1% at passage 23, and 77.6% at passage 36 (Fig.2B, inset). Similarly, E12.5 and E13.5 cultures proliferated extensively: they were composed of 99.5–100% nestin-positive cells, whereas ≤0.4% of the cells were positive for neuronal or glial markers. Together, these results indicate that OB cultures were composed predominantly of highly proliferative neural precursor cells that could be expanded for long periods of time. In addition, the near absence of β-III-tubulin-positive immunostaining, a marker of SVZ-derived migrating neuroblasts (Menezes and Luskin, 1994), and of GFAP-positive immunostaining, a marker of adult SVZ stem cells (Doetsch et al., 1999) in the cultures, further supports the conclusion that the source of the proliferative precursors is the OB neuroepithelium and not the SVZ-rostral migratory stream (RMS).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Antigen expression and BrdU labeling of expanded olfactory bulb stem cells. E14.5-derived cells were plated in the presence of FGF-2 and EGF (20 ng/ml each) and 10 μg/ml insulin; 5 μm BrdU was added to some cultures. A, The x-fold increase in cell number in each passage from 2 to 40 was measured by Trypan blue dye exclusion. Cells were passaged every 3–5 d. Other cultures were maintained to passage 20–23; their growth pattern was similar to the example shown. B, Antigen expression and BrdU labeling in OBSCs at different passages. In themain graph, for each cell marker, results show the mean ± SEM of data from five cultures of four experiments (passages 3, 4, 15, and 23). Inset, Results are the mean of data from two cultures analyzed at the passages (Ps) indicated. C–H, Images showing nestin, BrdU, IGF-I, IGF-IR, (pro)insulin, and InsR expression by OBSCs. IGF-I, (pro)insulin, and InsR immunostainings were performed on fixed cells that had been cultured previously for 3 d in the absence of IGF-I and insulin in the medium. Cultures were photographed under a fluorescence microscope (C, D), or images were captured using a confocal microscope (E–H). Scale bars:C, D, 19 μm; E,G, H, 6 μm; F, 10 μm.

To test whether these proliferative precursor cells had stem cell features (multipotentiality and self-renewal capacity) (Anderson, 2001;Temple, 2001), clonal analysis experiments were performed (Fig.3; Table1). Two hundred microliters of a E14.5-derived cell suspension containing two cells obtained from primary spheres (passages 4–16) were plated in 96 wells; the next day, 22.2 ± 2.5% (n = 5 experiments) of the wells contained a single cell. Only those wells were fed with FGF-2 plus EGF and maintained for further analysis. After 7–10 d, 42.5 ± 4.6% (n = 5 experiments) of the single cells formed spheres. Twelve clonally derived secondary spheres and six clonally derived tertiary spheres were analyzed by triple immunostaining. Of the spheres, 61.1% generated neurons, astrocytes, and oligodendrocytes after differentiation, indicating that the founder cells were multipotent (Fig. 3, Table 1). Other clonally derived spheres gave rise to neurons and astrocytes (5.6%), astrocytes and oligodendrocytes (5.6%), neurons (5.6%), or astrocytes only (22.2%). Dissociation and subclonal analysis of clonally derived secondary spheres confirmed the presence of cells that were multipotent with self-renewal capacity. In fact, four of six tertiary spheres gave rise to neurons, astrocytes, and oligodendrocytes.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Neurons, astrocytes, and oligodendrocytes can be derived from single-plated E14.5 olfactory bulb precursor cells. A cell suspension containing two cells in 200 μl of DMEM/F12/N2–OBSC conditioned medium (1:1) was plated into each well of 96-well plates.A, Wells containing only a single cell were marked.B, Clonally derived spheres were then transferred to polyornithine-coated glass coverslips, maintained for 2 d in FGF-2 plus EGF, and then (C) induced to differentiate. Cells were fixed 7–10 d after and triple immunostained using antibodies against neurons (D,TuJ1), astrocytes (E,GFAP), and oligodendrocytes (F,O4). Scale bars: A,C, 8.5 μm; B, 17 μm;D–F, 6 μm.

Olfactory bulb stem cells express IGF-I, (pro)insulin, and their cognate receptors, and these factors cooperate with FGF-2 plus EGF to promote proliferation

Expression of IGF-I, (pro)insulin, the IGF-IR, and the InsR was detected in the majority of OBSCs using previously characterized antibodies (Fig. 2E–H) (Herrera et al., 1985; Quiroga et al., 1995; Alarcón et al., 1998;Dı́az et al., 2000). On average (mean of two experiments), IGF-I and (pro)insulin were expressed in 99.8 and 96.5% of cells, respectively. IGF-IR was expressed in 89.2 ± 4.4% (n = 3) and InsR in 97.3 ± 0.9% (n = 3) of cells. Specific RT-PCR for proinsulin I and II showed that the mRNA expressed by stem cells was proinsulin II (C. Hernández-Sánchez, M. J. Yusta-Boyo, C. Vicario-Abejón, F. de Pablo, unpublished observations). We thus tested the action of exogenous IGF-I, insulin, and proinsulin alone or in combination with FGF-2 plus EGF on the proliferation and survival of OBSCs, growing at low density (5000–6000 cells per square centimeter, proliferative conditions). Proinsulin and insulin alone were ineffective in promoting the proliferation of olfactory bulb cells and had only a minimal effect on cell survival, even when high concentrations of these proteins were added to the cultures (Fig.4; Table2). In contrast, 100 ng/ml IGF-I elicited a significant increase both in the total number and in the number of proliferative cells, compared with values obtained with insulin or proinsulin. In the IGF-I-treated cultures, however, only 33.2% of cells were BrdU positive (Table 2), and some cells expressed neuronal or glial markers (data not shown). Similar results were obtained in the IGF-I plus insulin-treated cultures. In the absence of (pro)insulin or IGF-I, FGF-2 plus EGF promoted cell proliferation to a suboptimal level (Fig. 4, Table 2). The classical stem cell mitogens, FGF-2 and EGF alone, were thus insufficient to induce optimal cell proliferation rates, and under these conditions, maintenance and expansion of the cells were inefficient and variable.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Effect of IGF-I, insulin, and proinsulin on olfactory bulb precursor cell proliferation. E14.5-derived cells growing in medium containing FGF-2 plus EGF and 10 μg/ml insulin were passaged mechanically and plated in four-well chamber slides, where they were maintained for 3–4 d in the presence of the indicated growth factor concentration. A, Effect of insulin (INS), IGF-I, FGF-2, and EGF or their combinations on BrdU incorporation and total cell number. B, Effects of proinsulin (PROINS) compared with insulin, FGF-2, and EGF or their combinations on BrdU incorporation and total cell number. The concentration of FGF-2 and EGF was always 20 ng/ml. The concentration of IGF-I was always 100 ng/ml. Letters (a–f) indicate statistically significant changes (p < 0.05) between the two values compared, using Student's t test. Results are the mean ± SEM of data from four cultures of two experiments (passage 3).

We then tested whether the insulin-related growth factors cooperated with FGF-2 plus EGF to promote OBSC proliferation. As shown (Fig.4A,B), either 100 ng/ml insulin or IGF-I or proinsulin in addition to FGF-2 plus EGF yielded significantly higher numbers (1.8- to 2.7-fold; p < 0.05) of dividing and total cells than FGF-2 plus EGF alone. Cooperation was additive between IGF-I and FGF-2 plus EGF, and synergistic between (pro)insulin and FGF-2 plus EGF. The proportion of BrdU-positive cells increased progressively from the FGF-2 plus EGF-treated cultures to FGF-2 plus EGF plus high insulin-treated cultures (Table 2). Additionally, because BDNF promoted survival and proliferation of SVZ precursor cells (Benraiss et al., 2001; Pencea et al., 2001), the action of this neurotrophin was tested on OB precursors. BDNF produced no effect on cell proliferation or survival under any assay conditions (data not shown). Maximum levels of both dividing and total cells were found when FGF-2 and EGF were added to a medium containing high insulin (10 μg/ml) (Table 2), and these were chosen as standard culture conditions for proliferation. Proinsulin (10 μg/ml) also stimulated proliferation in the presence of FGF-2 plus EGF but was slightly less efficient than insulin (Fig. 4B). All of these results show that exogenous insulin-related growth factors are necessary together with FGF-2 plus EGF to promote optimal OBSC proliferation in culture.

Because BrdU can be incorporated by stem cells and other dividing progenitors, we tested whether IGF-I was necessary for sphere formation (a fact that attests to the presence of stem cells) (Arsenijevic et al., 2001). As seen in Figure 5,A and B, the addition of IGF-I for 3 d to the FGF plus EGF-treated cultures allowed generation of markedly larger spheres than those generated in FGF plus EGF alone. Under the latter conditions, however, the size of the spheres increased after 5–7 d in culture (Fig. 5C). (At that point, most spheres in the IGF-I-containing cultures were very confluent; data not shown). To perform a quantitative analysis of the IGF-I effects, cell aggregates formed by at least six to eight cells were considered to be spheres and counted after 3 d. Figure 5D shows that the number of spheres was significantly higher in the IGF-I-containing cultures than in the cultures containing only FGF plus EGF, and this effect was independent of the embryonic age of the tissue. All of these results show that IGF-I is necessary for optimal sphere formation. In addition, they show that spheres can be formed in the presence of FGF plus EGF alone, although with significantly slower kinetics, as the BrdU-incorporation analysis also indicated (Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Effect of IGF-I on sphere formation. Cells growing in medium containing FGF-2 plus EGF and 10 μg/ml insulin were passaged mechanically and plated in 24-well plates in the absence of insulin, where they were maintained for 3 d (A,B, D) or 7 d (C). A, E14.5-derived sphere growing in the presence of FGF-2 plus EGF. B, E14.5-derived sphere growing in the presence of FGF-2 plus EGF plus IGF-I. C, E12.5-derived sphere growing in the presence of FGF-2 plus EGF. D, The number of E12.5-E14.5-derived spheres growing in the absence or presence of IGF-I was counted. Results are the mean ± SEM of data from five cultures (passages 3–6). Growth factor concentrations were as in Figure 4. Scale bar, 40 μm. *p < 0.05.

Neurons and glia generated from olfactory bulb stem cells respond differently to IGF-I: cooperation among IGF-I, insulin, and BDNF

Our clonal analysis shows that a single proliferative cell from the olfactory bulb has the potential to give rise to three main CNS cell types: neurons, astrocytes, and oligodendrocytes (Fig. 3, Table1). We studied the differentiation of E14.5-derived stem cells using high-density culture conditions (100,000 cells per square centimeter) and analyzed the effect of IGF-I, insulin, and BDNF on these processes. Removal of FGF-2 and EGF, although maintaining insulin, initiated OBSC differentiation into neurons (TuJ1 and/or MAP2ab-positive cells), astrocytes (GFAP-positive cells), and oligodendrocytes (O4-positive cells) (Fig. 6A), as reported for hippocampal and striatal/subependymal stem cells (Gritti et al., 1996; Johe et al., 1996; Vicario-Abejón et al., 2000). The increase in TuJ1-positive cells was more pronounced during the first 2 d of culture, whereas the number of astrocytes augmented gradually up to day 4. One day after mitogen withdrawal, very few oligodendrocytes could be detected in the cultures. Under these conditions, neuronal generation therefore preceded that of astrocytes and oligodendrocytes; this is consistent with the in vivosituation (Jacobson, 1991) and with previous work in culture (Qian et al., 2000). β-III-tubulin expression in neurons preceded MAP-2ab expression (Fig. 6A), a pattern reminiscent of other neuronal systems (Menezes and Luskin, 1994).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Olfactory bulb stem cell differentiation into neurons, astrocytes, and oligodendrocytes: time course and proportions. E14.5-derived cells cultured in FGF-2 plus EGF and insulin were passaged mechanically and plated in four-well chamber slides or glass coverslips for 3–4 d in the absence of the mitogens but maintaining insulin. After fixation, cells were immunostained with the indicated antibodies. A, Time course of TuJ1-, MAP-2ab-, GFAP-, and O4-positive cells generated from stem cells. Results are the mean ± SEM of data from four to seven cultures of three experiments (passages 4 and 8). Numbers of MAP-2ab+ cells are the mean of data from two cultures of one experiment. B, Average numbers of each type of cell generated from stem cells after 3–4 d of differentiation. Results are the mean ± SEM of data from 14–15 cultures of eight experiments (passages 4–8).C–H, Morphological characteristics of neurons, astrocytes, and oligodendrocytes derived from olfactory bulb stem cells. C, TuJ1-positive cells, passage 5.D, MAP-2ab-positive cells, passage 40. E, Calretinin-positive cells, passage 4. F, GABA-positive cells, passage 4. G, GFAP-positive cells, passage 40.H, O4-positive cell, passage 15. Scale bar (shown in C): C, 15 μm;D, 7 μm; E, F, 10 μm;G, H, 12 μm.

Notably, the proportion of neurons was greater than that of astrocytes and oligodendrocytes at all time points studied. The average cell percentages obtained at 3 and 4 d after mitogen withdrawal in the presence of insulin was 52% neurons, 26% astrocytes, and 4% oligodendrocytes (Fig. 6B). Neuronal and glial differentiated progeny were obtained from stem cells both in early (passages 4–8) and late passages (15, 23, 40) (Fig.6B,C). To show that the majority of differentiated cells had originated from dividing cells in the culture, BrdU was added to the FGF-2 plus EGF-treated cultures for 22 hr. Once differentiated and fixed, double immunostaining of cultures for BrdU and the neuronal or glial markers showed 87.4 ± 1.8, 82.8 ± 2.0, and 64.3 ± 6.2% double BrdU-positive and TuJ1-, GFAP-, or O4-positive cells, respectively (n = 4–5 cultures; passages 6–23). In accordance with the clonal analysis experiments, these results indicate that OBSCs are able to generate the major CNS cell types after proliferating for long periods of time in culture. In addition, neuronal subtypes such as calretinin-positive cells and, in lower proportion, GABA- or GAD-positive cells, were present in the cultures (Fig. 6E,F).

The above results were obtained in the presence of a high insulin concentration in the medium. We next studied the specificity of the effects of exogenous IGF-I, in the absence of insulin, during the differentiative phase of stem cells. In the absence of IGF-I and insulin in the medium (BSA conditions), the number of neurons and glial cells was very low after 3–4 d of culture (Fig.7). Addition of 100 ng/ml IGF-I caused a marked increase in the absolute numbers of neurons, astrocytes, and oligodendrocytes. Moreover, the proportion of TuJ1- plus GFAP- plus O4-positive cells was 4.5-fold greater in IGF-I-treated cultures compared with the control BSA value. Under our culture conditions for OBSC differentiation (total absence of insulin and serum), it was not possible to distinguish an exclusive effect of exogenous IGF-I on differentiation from a combined differentiation plus survival effect. In fact, IGF-I treatment for only 15 hr produced larger numbers of TuJ1-positive cells, as well as of total cell number, compared with control cultures (data not shown). These conditions differ from those described by Arsenijevic and Weiss (1998) on striatal stem cells, which included 1% FBS in the differentiation medium. The minimum IGF-I dose augmenting the numbers of neurons, astrocytes, and oligodendrocytes was similar (Fig. 7A); however, the saturating concentration of IGF-I was 40 ng/ml for neurons and astrocytes and ∼200 ng/ml for oligodendrocytes (Fig. 7A,B). The strong IGF-I dependence of stem cell-generated neurons and astrocytes was further corroborated in experiments in which insulin or BDNF was added to cultures alone or in combination with IGF-I. As depicted (Fig.7B,C), 100 ng/ml insulin or 100 ng/ml BDNF produced only small increases in the numbers of stem cell-derived neurons and glia, compared with the effects elicited by 100 ng/ml IGF-I or 10 μg/ml insulin. Moreover, the numbers of TuJ1- or GFAP-positive cells were similar in the cultures treated with IGF-I alone or with IGF-I plus insulin or IGF-I plus BDNF. Interestingly, the numbers of O4-positive cells were two- to 2.5-fold greater (p < 0.01) in cultures treated with the combination of insulin plus IGF-I or BDNF plus IGF-I versus those treated with 100 ng/ml IGF-I (Fig.7B,C). OBSC-generated neurons and astrocytes are therefore markedly dependent on exogenous IGF-I for differentiation and survival, whereas oligodendrocytes are dependent on factors from various families.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

IGF-I effects on numbers of neurons and glia derived from olfactory bulb stem cells: cooperation with insulin and BDNF. E14.5-derived cells growing in the presence of FGF-2 plus EGF and insulin were plated in four-well chamber slides or glass coverslips for 3–4 d in the absence of insulin and mitogens. A, Dose–response curve of IGF-I effects on total cell number; TuJ1-, GFAP-, and O4-positive cells generated from OBSCs. Insetshows data for O4⁺ in an expanded scale. Differences between O4 values with 40 and 100 ng/ml IGF-I were statistically significant (*p < 0.05). B, Effects of IGF-I, insulin (INS), and their combination on total cell number; TuJ1-, GFAP-, and O4-positive cells. Insetshows data for O4⁺. Statistical analysis for O4-positive cells: 100 ng/ml insulin plus 100 ng/ml IGF-I versus 100 ng/ml IGF-I (*p < 0.01); 200 ng/ml IGF-I versus 100 ng/ml IGF-I (*p < 0.05); 10 μg/ml insulin versus 100 ng/ml IGF-I (*p < 0.01).C, Effects of IGF-I, BDNF, and their combination on total cell number; TuJ1-, GFAP-, and O4-positive cells.Inset shows data for O4⁺. Statistical analysis for O4-positive cells: 100 ng/ml BDNF plus 100 ng/ml IGF-I versus 100 ng/ml IGF-I (*p < 0.01). All of the results are the mean ± SEM of data from 4–10 cultures (passages 4–8).

Differentiation of stem cell-derived neurons and glia but not stem cell proliferation was compromised in cultures from the olfactory bulb ofIgf-I^−/−mice

The above results indicate that IGF-I and (pro)insulin have distinct, important roles and probably cooperate in vivo to control the number of OBSCs and their differentiated products. To determine whether endogenous IGF-I was absolutely required for stem cell proliferation and differentiation, cell cultures were prepared from the OB of E14.5 mutant mice lacking both copies of theIgf-I gene (null mutant) and of wild-type mice. Cultures were expanded in the presence of FGF-2 plus EGF plus insulin. Under these conditions, no differences in the number of nestin- or BrdU-positive cells were found betweenIgf-I^−/−and Igf-I^+/+ cultures (data not shown). Moreover, the cell proliferation rate was not affected by the mutation, even when cells were cultured in the absence of insulin, maintaining EGF plus FGF-2 (Fig.8A), or in the absence of insulin, EGF, and FGF-2 (data not shown); this indicates that endogenous IGF-I is not absolutely necessary to maintain OB stem cell proliferation. Cells were then induced to differentiate in the absence of insulin and mitogens for 3 d (Fig.8B,C). The percentages of neurons, astrocytes, and oligodendrocytes in theIgf-I^−/−cultures were reduced by 2.2-, 3.3-, and 4.8-fold, respectively, compared with the Igf-I^+/+cultures (Fig. 8B). In accordance, the total proportion of positive TuJ1- plus GFAP- plus O4-labeled cells was 14.3% inIgf-I^−/−cultures and 34.9% in wild-type cultures. Nonetheless, total cell number was not significantly lower in cultures from null compared with those from wild-type mice (Fig. 8C). These results indicate that the progeny of OBSCs require endogenous IGF-I for differentiation in culture.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8.

Differentiation of neurons and glia is impaired in olfactory bulb stem cell cultures fromIgf-I^−/−mice. Cultures from individual E14.5 embryos were expanded in a medium containing FGF-2 plus EGF and 10 μg/ml insulin. Cells were then allowed to proliferate (A) or, after mitogen withdrawal, to differentiate (B,C) for 3–4 d in the absence of insulin. Cells were immunostained using BrdU, TuJ1, GFAP, or O4 primary antibodies. Results are expressed as percentage of BrdU-positive cells (A), percentage of TuJ1-, GFAP-, or O4-positive cells (B) (inset shows data for O4⁺ in an expanded scale), and total cell number (C) and are the mean ± SEM of data from 6–12 cultures of three to six embryos per value (passages 5–9). Statistically significant differences (*p < 0.001) betweenIgf-I^−/−and Igf-I^+/+ cells.

Absence of IGF-I causes neuronal and glial defects in the olfactory bulb in vivo

To further establish that IGF-I is necessary for OB cell differentiation in vivo, sections obtained from E15.5-E18.5Igf-I^−/−and Igf-I^+/+ mice were analyzed. At E15.5, most of the mitral neurons have already generated in the mouse OB (Hinds, 1968a,b). At this age, two cell layers are prominent in the OB (Fig.9A–D), one composed mostly of neuroepithelial cells, surrounding the ventricle, and a second layer that stains positively for calretinin. The calretinin-positive cells would correspond to immature mitral neurons (and in a lower proportion, to tufted neurons) (Bastianelli and Pochet, 1995). As shown, H&E and calretinin staining revealed no clear effect of the Igf-I deletion in OB cell organization at E15.5. Notably, 1 d later, at E16.5 (Fig.9E,F), the wild-type OB shows a distinct mitral cell layer (MCL), apparently absent in theIgf-I null OB. Immunostaining with an antibody against neurotensin, a mitral neuron marker (Kiyama et al., 1991; Bulfone et al., 1998), revealed the presence of positive cells in theIgf-I null mouse OB, although they did not form a distinct MCL as observed in the wild-type mouse (data not shown). By the end of gestation, at E18.5 (Fig. 9G,H), there was a marked decrease in calretinin expression in the MCL inIgf-I^{− / −} embryos, whereas calretinin-positive cells were abundant in the MCL of wild-type embryos. No clear differences in calretinin expression were seen in the olfactory nerve layer or in the olfactory receptor neurons (data not shown) between the two genotypes, suggesting a specific effect of the Igf-I mutation on mitral neurons. GABA and tyrosine hydroxylase immunostaining of E18.5 sections did not reveal clear and consistent differences in the number and distribution of GABAergic and dopaminergic interneurons, respectively, betweenIgf-I^+/+ andIgf-I^{− / −} mice at this age (data not shown). Nestin protein (Fig.9I,J) was found primarily in neuroepithelial cells surrounding the lateral ventricle and in radial glia in the E15.5, E16.5 (data not shown), and E18.5 wild-type embryos. The MCL, in contrast, was nestin negative (Fig. 9I,arrowheads). In the Igf-I^+/+ embryos, nestin expression displays a radial pattern, with branched processes of radial glia forming two dense plexus, one superficial to the MCL and a second deep to MCL, characteristic of olfactory bulb radial glia (Bailey et al., 1999; Puche and Shipley, 2001). This pattern was distorted inIgf-I^{− / −} mice, both in radial distribution and in the extent and morphology of the plexus, especially the superficial one (Fig. 9J), indicating that radial glia differentiation was impaired as a consequence of the lack of IGF-I. The use of an antibody against GFAP did not reveal OB astrocytes at E18.5, although GFAP-positive cells were found in the hippocampal fimbria and in the spinal cord (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 9.

Mitral cell layer formation and radial glia differentiation are disrupted in the olfactory bulb ofIgf-I^−/− mice. Sections from E15.5, E16.5, and E18.5 Igf-I^+/+andIgf-I^−/−mouse olfactory bulb were stained with H&E (A,B, E, F) or immunostained with antibodies against calretinin (C,D, G, H) or nestin (I, J). In the absence of IGF-I, formation of the mitral cell layer (MCL) (E–H) and radial glia differentiation (I, J) are disrupted. No major differences in the cellular organization of the olfactory bulb between the two genotypes were observed at E15.5 (A–D). Similar phenotype was observed in two to four animals per genotype. NE, Neuroepithelial cells;ONL, olfactory nerve layer. Scale bar (shown inA): A-F, 25 μm;G, H, 10 μm; I,J, 30 μm.

Multipotent stem cells can be isolated from the mouse embryonic olfactory bulb

Here we report the isolation and characterization of stem cells locally born in the early embryonic mouse OB. Two facts support the conclusion that the OB neuroepithelium and not the SVZ–RMS is the source of the stem cells. First, cells having stem cell features can be isolated from the E12.5 and E13.5 OB, two ages at which no SVZ is observed in the OB, and not only from the E14.5 OB, the age at which the SVZ is first detected in the OB. The germinal zone surrounding the E14.5 OB ventricle in fact consists primarily of a layer of neuroepithelial cells, whereas the SVZ becomes prominent after E17–E18 (Hinds 1968b; Altman and Bayer, 1995; our observations). Second, under proliferative conditions, β-III-tubulin-positive cells, a characteristic feature of SVZ–RMS-derived neuroblasts (Menezes and Luskin, 1994), and GFAP-positive cells, a feature of adult SVZ stem cells (Doetsch et al., 1999), were nearly absent from the cultures.

Embryonic OBSC cells can divide in culture for at least 5.5 months, and clonal analysis demonstrates that 61.1% of differentiated clones contain neurons, astrocytes, and oligodendrocytes. Clonal analysis also confirms the existence within the spheres of fate-restricted proliferative precursor cells, which was observed previously in stem cell cultures from other brain regions (Davis and Temple, 1994;Reynolds and Weiss, 1996; Arsenijevic et al., 2001). Remarkably, embryonic OBSCs gave rise to greater numbers of neurons than of astrocytes or oligodendrocytes, at least up to passage 23 (3 months). Even after passage 40 (5.5 months in culture), they maintain the potential to differentiate into neurons and glia. These results indicate that under our culture conditions, OBSCs retain their differentiative potential for long periods of time. Astrocytes were nonetheless the major cell population that originated from adult OB stem cells (Gritti et al., 2002). It would be important to determine whether the distinct potential of embryonic and adult OB stem cells reflects a change in the cells' ability to integrate extracellular and intracellular signals responsible for cell fate decisions. Our results support the existence of stem cells in the mouse embryo OB with the potential to generate various types of glial as well as neuronal cells, the latter indicated by the presence of calretinin- and GABA-positive cells in the cultures. The finding of stem cells in the embryonic and adult OB opens the question as to whether they are lineage related. Whether the radial glia is a cell type intermediate between embryonic and adult stem cells (Alvarez-Buylla et al., 2001) that could function as a repository pool needs to be investigated.

Cooperative interplay of growth factors stimulates OBSC proliferation in culture

We report that in the absence of insulin-related growth factors, the proliferative and sphere formation potential of FGF-2 and EGF on olfactory bulb stem cells diminishes markedly, and that this potential can be rescued by addition of IGF-I and insulin, as well as by proinsulin. These results support cooperative interplay between IGF-I and (pro)insulin together with FGF-2 plus EGF, which is critical for optimal levels of OBSC proliferation in culture. The fact that IGF-I and (pro)insulin increase the percentage of proliferative cells as well as the total cell number in FGF-2 plus EGF-treated cultures suggests that these effects may be mediated by mechanisms accelerating cell entry into the S phase of the cell cycle or preventing apoptosis as reported in other cell systems (Dı́az et al., 2000; Jiang et al., 2001).

Arsenijevic et al. (2001) recently reported that in the absence of IGF-I, EGF or FGF-2 was unable to stimulate sphere formation of embryonic striatal cells; however, removal of IGF-I and (pro)insulin from OBSC cultures slowed down but did not block sphere formation promoted by FGF-2 plus EGF. Thus, stem cells from dorsal (OB) and ventral (striatum) telencephalic regions have slightly different growth factor requirements for proliferation. The data from both groups nevertheless reinforce a role for IGF-I in regulating stem cell number in culture, and our data extend this role to insulin and its precursor, proinsulin. In support of a role for IGF-I in stem and precursor cell proliferation, this factor promotes neurogenesis in the adult dentate gyrus (Aberg et al., 2000; Trejo et al., 2001). Our results demonstrate, however, that stem cell proliferation can progress in the absence of IGF-I. Ongoing studies should clarify whether other insulin-related growth factors can compensate for the lack of IGF-I in maintaining cell proliferation. The fact that OBSCs express (pro)insulin and IGF-II at the mRNA and protein levels, in addition to IGF-I, supports this possibility. These results together with data from Arsenijevic et al. (2001) reporting that endogenous IGF-I was necessary for EGF- but not for FGF-2-promoting neurosphere formation, and from Cheng et al. (2001) reporting that cell proliferation in the dentate gyrus was not reduced in Igf-Inull mice, should encourage further studies on the function of IGF-I and other insulin-related growth factors during stem cell proliferationin vivo.

The observations that IGF-I has a significant effect on OB cell proliferation whereas insulin has no effect, even at a very high concentration, and that effects of insulin plus IGF-I are similar to those of IGF-I (Fig. 4), suggest that in OBSCs, signals from the insulin-related growth factors are mediated primarily by the IGF-IR. A possible upregulation of IGF-IR levels by FGF-2 (Hernández-Sánchez et al., 1997) would facilitate (pro)insulin interaction with the IGF-IR. Two observations nonetheless suggest implication of the InsR, as a typical dimer or forming molecular hybrids with the IGF-IR, during OBSC proliferation. First, insulin and proinsulin elicit a synergistic response when added to FGF-2 plus EGF, whereas IGF-I only elicits an additive response; this may be interpreted as the activation of the InsR or hybrids in the first case and the IGF-IR in response to IGF-I. Second, immunocytochemical data and preliminary RT-PCR data (Hernández-Sánchez, Yusta-Boyo, Vicario-Abejón, de Pablo, unpublished observations) indicate that OBSCs express InsR as well as IGF-IR. In any case, the signaling cascade should converge, because addition of 100 ng/ml insulin does not increase the effects of 100 ng/ml IGF-I. Whether the InsR plays a specific role in oligodendrocyte differentiation, as suggested by the distinct response of this cell population to insulin and IGF-I, remains to be studied.

Endogenous IGF-I is required for differentiation of OB stem/precursor cell-derived neurons and glia in culture and in vivo

Short-term IGF-I treatment, after mitogen withdrawal from OBSCs, produced greater numbers of TuJ1-positive cells and total cells compared with control cultures. This fact suggests that exogenous IGF-I may act as a differentiation as well as a survival factor. Nonetheless, the data showed that the percentages of TuJ1-, GFAP-, and O4-positive cells were markedly lower in theIgf-I^−/−cultures compared withIgf-I^+/+, whereas total cell numbers were similar. This supports a specific role for endogenous IGF-I in OBSC differentiation, probably acting in an autocrine/paracrine manner. In accordance, formation of the MCL and morphological differentiation of radial glia is impaired in theIgf-I^{− / −} mice.

Although the numbers of OBSC-derived neurons, astrocytes, and oligodendrocytes are all IGF-I dependent, we found cell-specific characteristics in these responses. In particular, higher concentrations of exogenous IGF-I are necessary for maximal numbers of oligodendrocytes, in comparison with those of neurons or astrocytes. This may explain the synergistic effects of insulin and IGF-I or, more markedly, of BDNF and IGF-I in promoting oligodendrocyte development. The effect of BDNF on OBSC-derived oligodendrocytes is specific to this cell population, because BDNF alone or in combination with IGF-I has only a minimum effect on neuron or astrocyte numbers. This differs from the reported action of BDNF in promoting survival and differentiation of hippocampal (Vicario-Abejón et al., 1995, 2000) and striatal neuronal precursors (Ahmed et al., 1995; Arsenijevic and Weiss, 1998). In contrast, IGF-I is the most potent factor in promoting differentiation of neurons generated from OBSCs.

The in vivo analysis further supports the strong dependence of OB cells on IGF-I for complete differentiation; indeed, postmitotic mitral neurons appear to be generated in Igf-I null mice, but their allocation into an organized MCL is disrupted. A previous work reported specific depletion of mitral neurons in the OB of adult IGF-I null mice, but the reason for this was not described (Cheng at al., 1998). Our data suggest that mitral neuron depletion in theIgf-I^−/−may be secondary to the alteration of cell allocation. The MCL disorganization may be the cause or the consequence of the altered radial glia morphology, because formation of the two plexus of radial glia in the OB appears to concur with the presence of a distinct MCL (Bailey et al., 1999; Puche and Shipley 2001). Although the molecular events involved in MCL formation and radial glia differentiation, as well as the role of radial glia, are poorly understood, it is noteworthy that in the cortex, radial glia differentiation is dependent on the secreted protein reelin (Super et al., 2000). Because mitral neurons express reelin mRNA (Alcántara et al., 1998) and reelin plays a role in OB development (Kim et al., 2002), it is tempting to speculate that reelin, secreted by mitral neurons, may support the differentiation of radial glia in the OB and that absence of IGF-I may cause deregulation of the reelin pathway. Alternatively, an initial defect in radial glia differentiation, a consequence of the lack of IGF-I, may produce an alteration in mitral cell allocation into a precise layer, because it is suggested that OB radial glial may provide scaffolds for neuronal migration (Puche and Shipley, 2001). The OB is unique among brain structures expressing IGF-I high levels throughout life, particularly the mitral neurons (Bondy, 1991). Our results show that the OB has developed to acquire an exquisite dependence on IGF-I from early neurogenesis onward. The presence of stem cells in the OB opens up the possibility of studying whether OBSCs are a suitable source of diverse cell types for transplantation-based therapies of neurodegenerative diseases or CNS lesions.