ABSTRACT
Endogenous signals capable of inducing neuroectodermal differentiation are expressed by differentiating P19 EC cells in vitro. The present study demonstrates that at least two discrete signals are required. One is expressed by isolated primitive streak mesoderm-like cell lines and has the capacity to induce the expression of Pax-3 but, alone, induces neural differentiation inefficiently. The second signal is not expressed by the primitive streak mesoderm-like cell line but is present in conditioned media from differentiating P19 EC cells following DMSO treatment. This signal does not induce either Pax-3 expression or morphological differentiation and does not commit stem cells to a neuroectodermal fate. Rather, it acts synergistically with the signal derived from the primitive streak mesoderm-like cells to increase the efficiency with which stem cells respond initially by Pax-3 expression and subsequently by differentiation towards neural lineages. The activity of this second signal can be replaced by forskolin and 3-isobutyl-1-methyl-xanthine suggesting that its effects are transduced by a cyclic nucleotide-dependent pathway.
INTRODUCTION
Demonstration of a biological activity capable of inducing neural tissue was first described using amphibians (Spemann, 1938). During gastrulation in amphibians dorsal mesoderm forms due to inductive signals from the endoderm adjacent to the position where the blastopore will form (the Nieuwkoop center; Nieuwkoop et al., 1985). Transplantation of mesoderm from the dorsal lip of the blastopore (Spemann’s organizer) to ectoderm of the opposite, or ventral, side of a second embryo results in the organization of a secondary axis in which ectoderm normally fated to form epiderm is redirected to a neu-roectodermal fate.
Even the earliest dorsal mesoderm of the organizer expresses neural-inducing activity. However, the competence of the ectoderm to respond to neuroectoderm-inducing signals changes both temporally and spatially during development (reviewed in Slack and Tannahill, 1992). Ectoderm rapidly loses its ability to respond to mesoderm-inducing signals during early stages of gastrulation but remains responsive to neural induction late into gastrulation. Eventually, ectoderm that does not receive a neuralizing signal from dorsal mesoderm, either due to its location in the embryo or when isolated in vitro, differentiates towards epiderm. In addition to temporal changes, the competence of ectoderm to differentiate towards neural lineages differs spatially. Dorsal ectoderm expresses neural markers more strongly than does ventral ectoderm in response to signals from mesoderm (Sharpe et al., 1987).
Despite structural differences, gastrulation in mammals (reviewed in Beddington, 1986) exhibits many similarities to that of amphibians. In mammals, gastrulation begins in the egg cylinder, at which point the embryo consist of an outer layer of extraembryonic endoderm and an inner mass of cells termed the primitive ectoderm. The initial event in gastrulation is the differentiation of cells from the primitive ectoderm into mesoderm and the migration of this mesoderm along the primary axis of the embryo, between the primitive ectoderm and endoderm, to form the primitive streak. The neural plate forms from primitive ectoderm located adjacent to the axial mesoderm. Further, based on the developmental fate (Beddington, 1981; Beddington and Smith, 1993) and comparison of the expression patterns of the homeodomain-containing gene goosecoid (Blum et al., 1992), it has been suggested that the anterior primitive streak mesoderm of the mouse is equivalent to Spemann’s organizer of amphibians and that extraembryonic endoderm adjacent to the initial site of primitive streak formation expresses signals equivalent to the Nieuwkoop center. Additionally, cells from this general region of the mouse embryo, but not other regions, are sufficient to induce a secondary axis when grafted to Xenopus (Blum et al., 1992) or mouse (Beddington, 1994) embryos.
Expression of endogenous signals capable of inducing neu-roectodermal differentiation in a mammalian system has also been demonstrated in vitro using the P19 embryonal carcinoma (EC) model (Pruitt, 1992). The ability of LIF to block endodermal and mesodermal differentiation without inhibiting neural differentiation (Pruitt and Natoli, 1992) was utilized to assay for the expression of endogenous neural inducers in the absence of spontaneously differentiating endodermal or mesodermal lineages. In addition to the effect of LIF, these experiments relied on use of a P19 EC cell line (EH3), isolated using a gene trap approach, in which the endogenous Pax-3 gene was marked by incorporation of the lacZ gene. Pax-3 is expressed in the dorsal neuroepithelium shortly after its formation (Goulding et al., 1991). Using this cell line as a reporter, it was then possible to induce differentiation of P19 cells with DMSO (which gives rise to a high proportion of mesodermal lineages; McBurney et al., 1982), remove the chemical inducer and determine whether any of the differentiating cells present in these cultures expressed an endogenous activity capable of activating Pax-3 or neural differentiation when mixed with EH3 stem cells in the presence of LIF. Results from such experiments demonstrated that DMSO-treated P19 cells express an endogenous activity capable of inducing expression of the Pax-3/lacZ fusion gene and that these cells also express neural markers. Further, it has also been possible to derive cell lines from P19 EC cells which constitutively express high levels of the mesoderm-specific gene Brachyury and exhibit other properties expected for primitive streak mesoderm (Pruitt, 1994). Mixing experiments with the Pax-3/lacZ marked reporter cell line in the presence of LIF demonstrate that each of these lines also express high levels of Pax-3-inducing activity (Pruitt, 1994).
The present study demonstrates that, although the primitive streak mesoderm-like cells express one key factor which is required for neural induction, a second soluble factor which is expressed in excess by some cell type present in differentiating P19 EC cell cultures following DMSO treatment is also required for efficient neural differentiation. This second factor, alone, has no ability to induce Pax-3 expression or neural differentiation. Rather, it acts synergistically with the primitive streak mesoderm-like-cell-derived factor to allow a higher pro-portion of the target stem cells to respond by Pax-3 expression initially and subsequently by neural differentiation. These observations suggest that the second factor may act to induce stem cells to become competent to respond to the primitive streak mesoderm-like-cell-derived signal. Finally, the competence-inducing factor can be substituted for by forskolin plus 3-isobutyl-1-methyl-xanthine suggesting that neural competence is mediated by activation of a cyclic nucleotide dependent signal transduction pathway in P19 EC cells.
MATERIALS AND METHODS
Cell culture, induction and preparation of conditioned media
P19O1A1S18, EH3 and GCLB cells were cultured and maintained as described previously (McBurney et al., 1982, Pruitt and Natoli, 1992; Pruitt, 1992, 1994). To prepare conditioned media from DMSO-treated P19O1A1S18 cells (D-CM), approximately 1×106 cells were plated to 50 ml of culture media containing 1% DMSO as aggregates in bacterial grade Petri dishes, DMSO-containing media was changed on day 2 of culture and cells were plated to 15 cm diameter tissue culture grade dishes in the absence of DMSO on day 4. On day 5, following attachment of the aggregates to plates, media was removed and replaced with 25 ml of culture media. Media was harvested 18 hours later, centrifuged twice at 1000 g for 5 minutes and frozen at −70°C prior to use. In experiments requiring D-CM, this preparation was utilized in a ratio of 1/10 with non-conditioned culture media. To prepare conditioned media from GCLB cells (G-CM), approximately 1×106 cells were plated to tissue culture grade dishes in 50 ml of culture media containing 100 u/ml LIF. Cells were cultured for 4 days, media was harvested as described above and used at a ratio of 1/4 to 1/2 with non-conditioned media. Stock solutions (1000×) of forskolin (Sigma), 3-isobutyl-1-methyl-xanthine (Sigma) and phorbal 12-myristate 13-acetate (Sigma) were prepared in DMSO or ethanol and used at final concentrations of 10 nM, 100 nM and 5 ng/ml respectively. EH3/GCLB cell mixtures were prepared at a 1:4 ratio and cultured in the presence of 100 units/ml LIF as described previously (Pruitt and Natoli, 1992; Pruitt, 1992, 1994).
Immunofluorescence and β-galactosidase assays
Cell fixation, immunofluorescence staining, β-galactosidase histochemical staining and β-galactosidase assays were performed as described previously (Pruitt and Natoli, 1992; Pruitt, 1992, 1994). The unit definition for β galactosidase activity used here is the A420×106. In cases were individual cells were assayed, cells were trypsinized in phosphate-buffered saline, plated to tissue culture dishes (coated with 0.01% poly-L-lysine, Sigma) in culture media and allowed to attach for a period of 2 hours prior to fixation. Counts of individual cells following staining for either β-galactosidase or neurofilament M were performed from photomicrographs.
Northern blot analysis and probes
RNAs and probes were prepared, and northern blot analysis was performed, as described previously (Pruitt, 1994).
RESULTS
Two endogenous signals are required for efficient neural induction of P19 EC stem cells
To assay for the expression of neural-inducing activity by the primitive streak mesoderm-like cell line GCLB (Pruitt, 1994), GCLB cells were mixed in a 4:1 ratio with the reporter cell line EH3 (Pruitt, 1992) which carries the lacZ gene incorporated into one allele of the endogenous Pax-3 gene. Cell mixtures were incubated as aggregates in the presence of LIF for a period of 4 days followed by plating to tissue culture grade Petri dishes for various periods of time in the continued presence of LIF. The morphology of β-galactosidase-expressing cells on day 7 following mixing is shown in Fig. 1, where panel A shows several mixed aggregates at lower magnification and panels B and C show a single mixed aggregate at higher magnification. Although many of the cells express the Pax-3/lacZ fusion gene, these cells seldom exhibit morphological differentiation. These observations are in contrast to the efficient morphological differentiation observed previously when DMSO-treated P19 cells were mixed with the reporter cell line (Pruitt, 1992).
Effect of D-CM on the morphology of β-galactosidase-expressing cells in EH3/GCLB mixed aggregates. Mixtures of EH3 and GCLB (1:4) cells were prepared and grown as aggregates in the presence of LIF for a period of 4 days, plated to tissue culture grade dishes and grown for an additional 3 days (day 7 of the experiment) in the continued presence of LIF, fixed and stained for β-galactosidase activity as described in the text. (A-C) D-CM omitted; (D-F) D-CM included in a 1/10 ratio with non-conditioned media. Photomicrographs were taken at magnifications of 40× (A,D), or 100×, (B,C,E,F). Bright-field optics were used in A, C, D and F and phase-contrast optics were used in B and E.
Effect of D-CM on the morphology of β-galactosidase-expressing cells in EH3/GCLB mixed aggregates. Mixtures of EH3 and GCLB (1:4) cells were prepared and grown as aggregates in the presence of LIF for a period of 4 days, plated to tissue culture grade dishes and grown for an additional 3 days (day 7 of the experiment) in the continued presence of LIF, fixed and stained for β-galactosidase activity as described in the text. (A-C) D-CM omitted; (D-F) D-CM included in a 1/10 ratio with non-conditioned media. Photomicrographs were taken at magnifications of 40× (A,D), or 100×, (B,C,E,F). Bright-field optics were used in A, C, D and F and phase-contrast optics were used in B and E.
One possible explanation for the failure of Pax-3-expressing target cells to differentiate following mixing with the mesoderm-like cell line GCLB is that GCLB cells express only one of two signals required for neural differentiation. In this case an additional signal, expressed by some cell type induced to differentiate following DMSO treatment of P19 EC cells, would be predicted. To test this possibility, mixtures of GCLB and EH3 cells were aggregated as previously except in the presence of 1/10 volume of conditioned media prepared from differentiating P19 EC cells following DMSO treatment (D-CM). The presence of D-CM is sufficient to cause the Pax-3/lacZ-expressing target cells present in EH3/GCLB cell mixtures to differentiate. Differentiation is first detected by a change in morphology and migration of the Pax-3/lacZ-expressing target cells from the aggregates as is shown for cells on day 7 of the experiment in Fig. 1, panels D (several aggregates at lower magnification) and E and F (a single aggregate at higher magnification). To confirm that migration and changes in morphology observed in D-CM-treated EH3/GCLB mixtures are accompanied by the expression of appropriate markers, β-galactosidase-expressing cells were assayed for the expression of smooth muscle actin (SMA) and neurofilament M (NFM). Expression of SMA is observed in P19 EC cells differentiating towards neuroectoderm following treatment with retinoic acid and its expression at very early stages of differentiation along a variety of neuroectodermal and mesodermal lineages formed by P19 cells has been demonstrated (St-Arnaud et al., 1989; Rudnicki et al., 1990). Similarly, β-galactosidase-positive, migrating, cells in D-CM-treated EH3/GCLB mixed cultures express SMA (Fig. 2C,D). In contrast, Pax-3/lacZ-expressing target cells present in GCLB/EH3 cell mixtures not treated with D-CM fail to express this early differentiation marker (Fig. 2A,B).
Effect of D-CM on smooth muscle actin and neurofilament M expression in EH3/GCLB mixed aggregates. (A-D) Mixed aggregates of EH3 and GCLB cells were prepared as in Fig. 1; (A,B) cells grown in the absence of D-CM and (C,D) cells grown in the presence of D-CM through day 7 of the experiment. Cells were photographed using bright-field optics in A and C and the same fields were stained for the expression of smooth muscle actin by immunofluorescence in B and D. (E-H) Mixed aggregates of EH3 and GCLB cells were prepared as above except that cells were cultured for two additional days (day 9 of the experiment) as monolayer cultures; (E,F) cells grown in the absence of D-CM and (G,H) cells grown in the presence of D-CM. Cells were photographed using bright-field optics in E and G and the same fields were stained for the expression of neurofilament M by immunofluorescence in F and H. All photomicrographs were taken at a magnification of 100×.
Effect of D-CM on smooth muscle actin and neurofilament M expression in EH3/GCLB mixed aggregates. (A-D) Mixed aggregates of EH3 and GCLB cells were prepared as in Fig. 1; (A,B) cells grown in the absence of D-CM and (C,D) cells grown in the presence of D-CM through day 7 of the experiment. Cells were photographed using bright-field optics in A and C and the same fields were stained for the expression of smooth muscle actin by immunofluorescence in B and D. (E-H) Mixed aggregates of EH3 and GCLB cells were prepared as above except that cells were cultured for two additional days (day 9 of the experiment) as monolayer cultures; (E,F) cells grown in the absence of D-CM and (G,H) cells grown in the presence of D-CM. Cells were photographed using bright-field optics in E and G and the same fields were stained for the expression of neurofilament M by immunofluorescence in F and H. All photomicrographs were taken at a magnification of 100×.
In D-CM-treated EH3/GCLB mixtures, Pax-3/lacZ-expressing cells exhibiting a neural morphology are detected beginning on approximately day 8 and increase in number through day 12. These cells also express the neural markers NFM (Fig. 3G,H) and HNK-1 (data not shown). The efficiency with which neural differentiation occurs in D-CM-treated EH3-GCLB mixed cultures is similar to that observed previously when EH3 cells were mixed with DMSO-treated P19 cells or following treatment of the target stem cells with 1× 10−6 M RA (Pruitt, 1992). In these cases approximately 80% of the β-galactosidase-expressing foci were associated with cells expressing neural markers. In the experiment shown in Fig. 3, approximately 84% (n=102) of the β-galactosidaseexpressing foci present in the mixed EH3/GCLB aggregates treated with D-CM are closely associated with cells staining for NFM. Additionally, when individual cells are assayed following trypsinization, approximately 1.9% of the cells from D-CM-treated EH3/GCLB mixed aggregates express NFM (Table 1). In comparison, approximately 1.3% of cells from EH3 aggregates treated with 1×10−6 M RA express NFM in similar experiments (data not shown). In contrast, even when followed through 10 days of culture, EH3/GCLB mixed cultures grown in the absence of D-CM show little neural differentiation (Fig. 2E,F) and fewer than 2% (n=104) of the β-galactosidase-expressing foci are associated with cells staining for the neural marker NFM. Additionally, when individual cells are assayed following trypsinization, fewer than 0.05% (no NFM-expressing cells in 2302 scored) stain for NFM (Table 1). Similar results have been obtained for mixing experiments where the GCLB line was substituted for a second primitive streak mesoderm-like cell line (LD3; Pruitt, 1994). These observations demonstrate that a factor present in conditioned media from DMSO-treated P19 cells increases the efficiency with which target stem cells are induced to differentiate towards neural lineages following mixing with the primitive streak mesoderm-like cell line.
Effect of D-CM and forskolin/IMX treatment on the proportion cells expressing β-galactosidase and neurofilament M

Effect of D-CM on β-galactosidase expression in EH3 or mixed EH3/GCLB aggregates. EH3 and GCLB cells were pretreated as monolayer cultures with 1/10 dilution of D-CM in the presence of LIF for 2 days. Following pretreatment, EH3, or EH3 plus GCLB, aggregates were prepared in the presence or absence of D-CM as indicated in the figure. Cells were cultured and harvested on day 7 of the experiment for determination of total β-galactosidase expression as described previously. The average results from two experiments are shown where the range is indicated by error bars.
Effect of D-CM on β-galactosidase expression in EH3 or mixed EH3/GCLB aggregates. EH3 and GCLB cells were pretreated as monolayer cultures with 1/10 dilution of D-CM in the presence of LIF for 2 days. Following pretreatment, EH3, or EH3 plus GCLB, aggregates were prepared in the presence or absence of D-CM as indicated in the figure. Cells were cultured and harvested on day 7 of the experiment for determination of total β-galactosidase expression as described previously. The average results from two experiments are shown where the range is indicated by error bars.
The proportion of target cells that respond to the mesoderm-like-cell-derived signal is increased by conditioned media from DMSO-treated P19 cells
In addition to increasing the efficiency of neural differentiation, D-CM also increases the total level of Pax-3/lacZ expression in EH3-GCLB mixed cultures by a factor of 5- to 7-fold when assayed by determination of total β-galactosidase activity (e.g. Fig. 3). This increase is apparently not accounted for by an increase in the level of Pax-3 expression occurring in individual cells assayed in situ since the intensity of β-galactosidase expression in responding cells is similar between D-CM-treated and untreated cultures (e.g. Fig. 1). The proportion of aggregates that contain at least some β-galactosidase-expressing cells is also similar suggesting that D-CM does not affect the efficiency with which the EH3 and GCLB cells form mixed aggregates (data not shown). Rather, the increase in β-galactosidase expression can be accounted for by an increase in the proportion of target stem cells which are expressing the Pax-3/lacZ fusion gene. Assays of individual cells following trypsinization of mixed aggregates demonstrates that approximately 3.7% of the total cells present express β-galactosidase in the absence of D-CM and this proportion increases to approximately 25.9% when D-CM is present (Table 1). These observations demonstrate that, in addition to increasing the efficiency of neuroectodermal differentiation, D-CM also contains an activity capable of increasing the proportion of individual target stem cells that respond by induction of Pax-3 expression initially. However, since D-CM does not induce Pax-3/lacZ expression from target stem cells in the absence of primitive streak mesoderm-like cells, this activity apparently acts either to increase the efficiency with which the mesoderm-like-cell-derived signal is expressed or the efficiency with which the target stem cells respond to the signal.
An activity contained in conditioned media from DMSO-treated P19 cells acts on the target stem cells
To attempt to distinguish whether D-CM acts on the target stem cells or the mesoderm-like cells present in mixed aggregates, the effect of pretreating either EH3 cells or GCLB cells with D-CM for 2 days prior to mixing the cells in the absence of D-CM was determined. Pretreatment of neither cell type is sufficient to obtain the full response observed when D-CM is present continuously. However, pretreating the EH3 target stem cells with D-CM is sufficient to increase the level of β-galactosidase expression to approximately half of the level observed when D-CM is present throughout the incubation as aggregates (Fig. 3). Further, many of the β-galactosidase expressing cells show the altered morphology characteristic of D-CM-treated cells at this stage of differentiation (data not shown). In contrast pretreating the GCLB cells has a smaller effect on the level of β-galactosidase expression (Fig. 3) and does not induce morphological changes (data not shown). These observations suggest that D-CM acts, at least in part, on the target stem cells to increase their ability to respond to signal(s) provided by the primitive streak mesoderm-like cell line.
If D-CM acts on the target stem cell line, it should be possible to observe its effects when only the signal(s) from the primitive streak mesoderm-like cells, but not the cells them-selves, are present. To test this possibility, the effect of D-CM on the ability of target stem cells to respond to conditioned media from GCLB cells (G-CM) was determined. Fig. 4 demonstrates that, although G-CM alone can induce a low level of Pax-3/lacZ expression, the response is increased by a factor of approximately 10 in the presence of D-CM. Additionally, Pax-3/lacZ-expressing cells induced by G-CM alone show little morphological differentiation or expression of SMA (Fig. 5E,F) while efficient differentiation and expression of SMA occurs when both D-CM and G-CM are present (Fig. 8B,C). These observations suggest that D-CM acts on the target stem cell to enhance their ability to respond to the signal from the primitive streak mesoderm-like cells.
Effect of D-CM and G-CM on expression of β-galactosidase in EH3 cells. EH3 cell aggregates were prepared and incubated in the presence of either a 1/10 dilution of D-CM, a 1/2 dilution of G-CM or both for 4 days in the presence of LIF. Samples were plated to monolayer culture and incubated for an additional 4 days prior to fixation, staining for β-galactosidase expression in situ and counter staining with eosin (A) or harvesting for total β-galactosidase activity determination (B) as previously.
Effect of D-CM and G-CM on expression of β-galactosidase in EH3 cells. EH3 cell aggregates were prepared and incubated in the presence of either a 1/10 dilution of D-CM, a 1/2 dilution of G-CM or both for 4 days in the presence of LIF. Samples were plated to monolayer culture and incubated for an additional 4 days prior to fixation, staining for β-galactosidase expression in situ and counter staining with eosin (A) or harvesting for total β-galactosidase activity determination (B) as previously.
Effect of D-CM and G-CM on morphology and smooth muscle actin expression in EH3 cells. Photomicrographs of cells from the experiment shown in Fig. 4. (A,B) Treated with D-CM; (E,F) treated with G-CM; (C,D) treated with both. (A,C,E) Bright-field photographs of the same fields as shown in B, D and F following immunofluorescent staining of smooth muscle actin. All photomicrographs were taken at a magnification of 100×.
Effect of D-CM and G-CM on morphology and smooth muscle actin expression in EH3 cells. Photomicrographs of cells from the experiment shown in Fig. 4. (A,B) Treated with D-CM; (E,F) treated with G-CM; (C,D) treated with both. (A,C,E) Bright-field photographs of the same fields as shown in B, D and F following immunofluorescent staining of smooth muscle actin. All photomicrographs were taken at a magnification of 100×.
Activities present in D-CM do not commit cells to a neuroectodermal lineage
One mechanism by which a factor present in D-CM could act to enhance the efficiency with which mesoderm-like cells induce neural differentiation is by committing the cells to an early neuroectodermal lineage. To test this possibility, the EH3 stem cell line was pretreated for two days with D-CM in the presence of LIF as previously and the cells were challenged to differentiate towards mesodermal lineages by removal of LIF and treatment with 1.0% DMSO as aggregates. If D-CM treatment results in commitment to neuroectodermal lineages, the mesodermal lineages which are normally formed by these cells following DMSO treatment would not be expected to differentiate. Cells were either harvested on day 4 of the induction and assayed for the mesodermal marker Brachyury (Fig. 6) or plated to tissue culture grade dishes in the absence of inducer and stained for cardiac- and skeletal-specific myosin on day 12 of the experiment (Fig. 7). D-CM pretreatment has no effect on either the level of Brachyury expression on day 4 or in the efficiency with which cardiac and skeletal myoblast differentiate subsequently on day 12. These observations demonstrate that, although a factor present in D-CM increases the efficiency with which stem cells differentiate towards neuroectoderm in the presence of the GCLB-derived factor, this factor does not commit the cells to neuroectodermal lineages.
Effect of D-CM pretreatment on Brachyury expression in DMSO-treated EH3 cells. EH3 cells were grown in monolayer cultures for 2 days under the pretreatment conditions indicated where LIF and D-CM were used as described previously and DMSO was used at a concentration of 1.0%. Cells were then trypsinized and plated to aggregates under the treatment conditions indicated and cultured for 4 days. RNAs isolated from samples of each culture on day 4 were utilized in northern blot analysis and probed for the expression of Brachyury (12 hour exposure) or triosephosphate isomerase (TPI; 6 hour exposure) messages.
Effect of D-CM pretreatment on Brachyury expression in DMSO-treated EH3 cells. EH3 cells were grown in monolayer cultures for 2 days under the pretreatment conditions indicated where LIF and D-CM were used as described previously and DMSO was used at a concentration of 1.0%. Cells were then trypsinized and plated to aggregates under the treatment conditions indicated and cultured for 4 days. RNAs isolated from samples of each culture on day 4 were utilized in northern blot analysis and probed for the expression of Brachyury (12 hour exposure) or triosephosphate isomerase (TPI; 6 hour exposure) messages.
Effect of D-CM pretreatment on cardiac/skeletal-specific myosin expression in DMSO-treated EH3 cells. Samples of EH3 cells from the experiment shown in Fig. 6 were plated to monolayer cultures on day 4 of the experiment and cultured in the absence of LIF, D-CM, or DMSO for an additional 8 days (day 12 of the experiment) prior to fixation and immunofluorescent staining for cardiac and skeletal muscle-specific myosin using the MF-20 mAb. The samples shown are cells treated initially with: (A) pretreatment = no LIF, no D-CM; treatment = no LIF, no D-CM, plus DMSO; (B) pretreatment = plus LIF, plus D-CM; treatment = no LIF, no D-CM, plus DMSO; (C) pretreatment = plus LIF, no D-CM; treatment = no LIF, plus D-CM, no DMSO. (D) A parallel experiment where EH3 plus GCLB aggregates grown in LIF were treated with the D-CM preparation used in this experiment and stained for neurofilament M expression by immunofluorescence to demonstrate that this preparation of D-CM was efficient is promoting neural differentiation in EH3/GCLB mixed aggregates. All photomicrographs were taken at a magnification of 100×.
Effect of D-CM pretreatment on cardiac/skeletal-specific myosin expression in DMSO-treated EH3 cells. Samples of EH3 cells from the experiment shown in Fig. 6 were plated to monolayer cultures on day 4 of the experiment and cultured in the absence of LIF, D-CM, or DMSO for an additional 8 days (day 12 of the experiment) prior to fixation and immunofluorescent staining for cardiac and skeletal muscle-specific myosin using the MF-20 mAb. The samples shown are cells treated initially with: (A) pretreatment = no LIF, no D-CM; treatment = no LIF, no D-CM, plus DMSO; (B) pretreatment = plus LIF, plus D-CM; treatment = no LIF, no D-CM, plus DMSO; (C) pretreatment = plus LIF, no D-CM; treatment = no LIF, plus D-CM, no DMSO. (D) A parallel experiment where EH3 plus GCLB aggregates grown in LIF were treated with the D-CM preparation used in this experiment and stained for neurofilament M expression by immunofluorescence to demonstrate that this preparation of D-CM was efficient is promoting neural differentiation in EH3/GCLB mixed aggregates. All photomicrographs were taken at a magnification of 100×.
Effects of activating protein kinase C and cyclic nucleotide-dependent signal transduction pathways on Pax-3 and neural induction
In Xenopus, the enhanced competence for neural induction of dorsal, relative to ventral, ectoderm is mediated by elevated levels of protein kinase C-α (PKC-α; Otte and Moon, 1992). Additionally, Xenopus ectoderm that has been made competent by activation of PKC will differentiate towards neural lineages more efficiently in response to activation of the adenylate cyclase pathway (Otte et al., 1989). To test the possibility that PKC or adenylate cyclase activation mediates the signal derived from either primitive streak mesoderm-like cells or the competence factor present in D-CM, the ability of phorbal 12-myristate 13-acetate (PMA), which activates PKC, or forskolin plus 3-isobutyl-1-methyl-xanthine (F/I), which activates cyclic nucleotide-dependent signal transduction pathways, to substitute for either of the endogenous inducing activities was tested. PMA had no stimulatory effect either alone or in combination with F/I and was unable to substitute for either the endogenous signal provided by primitive streak mesoderm-like cells or the factor present in D-CM (data not shown). F/I also failed to stimulate Pax-3/lacZ expression from target cells in the absence of primitive streak mesoderm-like cells even when D-CM was present (data not shown). However, in mixtures of EH3 and GCLB cells, addition of F/I stimulated β-galactosidase expression by a factor of approximately 10 (Fig. 8). This stimulation is reduced by approximately 50% when PMA is also present (data not shown). The effect of pretreating either target stem cells or the primitive streak mesoderm-like cells with F/I was also determined. Pretreatment of GCLB cells results in no enhanced Pax-3/lacZ expression in mixed aggregates, while some stimulation is observed when EH3 cells are pretreated, similar to the effect of D-CM.
Forskolin/IMX treatment mimics the effect of D-CM on β-galactosidase expression in EH3/GCLB mixed aggregates. EH3, or EH3 plus GCLB, cell aggregates were prepared and incubated as described previously under the conditions indicated in the figure. Forskolin was used at a concentration of 10 nM and 3-isobutyl-1-methyl-xanthine (IMX) was used at a concentration of 100 nM. Where pretreatments were used [GCLB(F/I) and EH3(F/I)] cells were grown as monolayer cultures for 2 days in the presence of these concentrations of forskolin and IMX prior to preparation of aggregates. Cells were harvested on day 7 of the experiment, extracts were prepared and total β-galactosidase expression was determined as described previously.
Forskolin/IMX treatment mimics the effect of D-CM on β-galactosidase expression in EH3/GCLB mixed aggregates. EH3, or EH3 plus GCLB, cell aggregates were prepared and incubated as described previously under the conditions indicated in the figure. Forskolin was used at a concentration of 10 nM and 3-isobutyl-1-methyl-xanthine (IMX) was used at a concentration of 100 nM. Where pretreatments were used [GCLB(F/I) and EH3(F/I)] cells were grown as monolayer cultures for 2 days in the presence of these concentrations of forskolin and IMX prior to preparation of aggregates. Cells were harvested on day 7 of the experiment, extracts were prepared and total β-galactosidase expression was determined as described previously.
The effect of F/I on morphological differentiation and SMA expression at 7 days (Fig. 9A,B), and the differentiation of cells exhibiting a neural morphology and NFM expression at 9 days (Fig. 9C,D), in mixed aggregates was determined for the experiment shown in Fig. 8. Similar to the effect of D-CM, F/I has no ability to induce Pax-3 expression, morphological differentiation, or expression of neural markers in target stem cells in the absence of a signal from the primitive streak mesoderm-like cells (Fig. 8 and data not shown). Additionally, in mixed aggregates the presence of F/I results in morphological changes and the expression of specific markers which are virtually identical to those observed in the presence of D-CM. The efficiency of neural induction between D-CM- and F/I-treated cultures is also similar. For the experiment shown in Fig. 8 only 2% (n =93) of the Pax-3/lacZ-expressing foci in mixed EH3/GCLB aggregates were associated with NFM-staining cells in absence of either D-CM or F/I, similar to previous observations. This proportion increased to 93% (n=107) in the presence of D-CM and 97% (n=95) in the presence of F/I. Additionally, assays of NFM expression in individual cells following trypsinization of mixed aggregates demonstrated that fewer than 0.05% (no NFM-expressing cells in 2302 scored) of the cells expressed NFM in the absence of either D-CM or F/I and this proportion increased to 1.9% or 2.6%, respectively, in their presence (Table 1).
Forskolin/IMX treatment mimics the effect of D-CM on differentiation in EH3/GCLB mixed aggregates. EH3 plus GCLB mixed aggregates were incubated in the presence of LIF plus F/I for a period of 4 days and plated to monolayer cultures in the absence of F/I but the continued presence of LIF as described previously. Cells were fixed and stained in situ for β-galactosidase expression and smooth muscle actin on day 7 (A,B) or neurofilament M on day 9 (C,D) of the experiment as described previously. (A,C) Bright-field photomicrographs of the fields shown in B and D. All photomicrographs were taken at a magnification of 100×.
Forskolin/IMX treatment mimics the effect of D-CM on differentiation in EH3/GCLB mixed aggregates. EH3 plus GCLB mixed aggregates were incubated in the presence of LIF plus F/I for a period of 4 days and plated to monolayer cultures in the absence of F/I but the continued presence of LIF as described previously. Cells were fixed and stained in situ for β-galactosidase expression and smooth muscle actin on day 7 (A,B) or neurofilament M on day 9 (C,D) of the experiment as described previously. (A,C) Bright-field photomicrographs of the fields shown in B and D. All photomicrographs were taken at a magnification of 100×.
DISCUSSION
Previous studies have demonstrated that P19 cells differentiating towards mesodermal lineages (following treatment as aggregates with DMSO) express activities capable of inducing both Pax-3 and neuroectoderm differentiation in adjacent stem cells. Further, Pax-3 expression and neural differentiation were closely associated (Pruitt, 1992). Expression of Brachyury following DMSO treatment of P19 cell cultures parallels the expression of the Pax-3 and neural-inducing activities, consistent with the possibility that primitive streak mesoderm-like cells are the source of the inducing activities (Pruitt, 1994). Further, two different cell lines that exhibit properties of primitive streak mesoderm, including constitutively high levels of expression of Brachyury, (but not a number of other non-mesodermal lines tested) express an activity capable of inducing Pax-3 expression in adjacent stem cells (Pruitt, 1994). However, the present study demonstrates that very few fully differentiated neural lineages are found in association with these Pax-3-expressing cells. This observation suggests that the primitive streak mesoderm-like cell lines do not express all of the signals necessary for efficient neuroectodermal differentiation. The major observation reported here is that a second factor, which is present in excess in conditioned media of DMSO-treated P19 cells (D-CM), is sufficient to allow efficient neural differentiation of the target stem cells induced to express Pax-3 following mixing with the primitive streak mesoderm-like cells. This observation demonstrates that at least two discrete endogenous signals are required for the induction of neuroectoderm differentiation in P19 EC cells.
Roles of neural-inducing activities in P19 stem cells
Several observations suggest that the role of the D-CM-derived factor is to change the response of target stem cells to the signal(s) derived from primitive streak mesoderm-like cells. First, D-CM has no ability to induce either Pax-3 expression or neural differentiation in the absence of the signal from primitive streak mesoderm-like cells and treatment of stem cells with D-CM does not prevent their subsequent differentiation towards mesodermal lineages. These observations suggest that the activity present in D-CM is not sufficient, alone, to commit the cells to a neuroectodermal lineage. Second, target stem cells can respond to the signal derived from primitive streak mesoderm-like cells even in the absence of D-CM; however, in the presence of D-CM this response is modified in two ways: (1) a much higher proportion of the target stem cells in the mixed aggregates respond to the mesoderm-derived signal by expression of Pax-3 initially and (2) the target stem cells that respond begin morphological differentiation leading to mature neurons much more efficiently. Finally, the effects of D-CM do not require that the primitive streak mesoderm-like cells are present during D-CM treatment. This is observed both in the elevated response of target stem cells that have been pretreated with D-CM prior to mixing with the primitive streak mesoderm-like cells and in the ability of D-CM to cause an enhanced response of target stem cells to conditioned media derived from the primitive streak mesoderm-like cell line. These results suggest that D-CM acts on the target stem cells directly. Together these observations are consistent with the possibility that D-CM acts to induce changes in the target stem cells that modify their response to the signal(s) from the primitive streak mesoderm-like cells. One result of this modification is that the target stem cells become competent to differentiate to mature neuroectodermal cell lineages and in this sense D-CM can be considered a neural competence-inducing factor.
Roles of PKC and cyclic nucleotide-dependent transduction pathways in mediating neural-inducing activities
Previous studies have implicated a role for the activation of both the PKC and adenylate cyclase-dependent signal transduction pathways during neural induction in Xenopus (Otte et al., 1988, 1989, 1990, 1991; Otte and Moon, 1992). The possibility that activation of these pathways, either individually or in combination with each other, is sufficient to induce neural differentiation in mammals has also been studied using the human teratocarcinoma stem cell line NT2/D1. Both adenylate cyclase and PKC activities are increased during neural differentiation of these cells (Abraham et al., 1991). However, treatment of NT2/D1 cells with phorbal ester, di-butyryl cAMP, or combined treatment is not sufficient to induce neuronal markers (Andrews et al., 1986; Abraham et al., 1991). Results on P19 cells presented here are consistent with these previous studies and demonstrate that activation of neither the PKC or adenylate cyclase pathways, using PMA and forskolin/IMX, respectively, nor combined treatment is sufficient to induce either Pax-3 expression or neural differentiation.
The possibility that activation of either the PKC or cyclic nucleotide-dependent signal transduction pathways could substitute for either of the endogenous activities required for neural induction of P19 cells was also tested. These studies demonstrated that PKC activation was unable to substitute for either endogenous activity but that the activity present in D-CM could be replace by forskolin plus IMX. Forskolin/IMX treatment was sufficient to mimic each of the effects of D-CM, including enhancing the level of Pax-3 expression and induction of mature neural differentiation, when target stem cells were mixed with primitive streak mesoderm-like cells. Similar to the effect of D-CM, forskolin/IMX has no effect in the absence of the signal from primitive streak mesoderm-like cells. The ability of forskolin/IMX to substitute for D-CM suggests that the factor present in D-CM functions by activation of a cyclic nucleotide-dependent pathway in the target stem cell.
Several ligands have been shown to induce neural differentiation in Xenopus including noggin (Lamb et al., 1993), bFGF (Kengaku and Okamoto, 1993) and follistatin (Hemmati-Brivanlou et al., 1994). It is possible that one or more of these ligands also has a role in neural induction in mammals. However, the differences observed between the signal transduction pathways capable of mediating neural induction in the Xenopus (Otte et al., 1988, 1989, 1990, 1991; Otte and Moon, 1992) and mammalian (Andrews et al., 1986; Abraham et al., 1991; and the present study) systems implies that there may also be differences in the ligands responsible for neural induction between these systems.
Relationship between the induction of Pax-3 expression and neural differentiation
Two effects of D-CM have been demonstrated. It enhances the ability of target stem cells to respond to the signal from primitive streak mesoderm-like cells by increasing the proportion of cells that express Pax-3and it increases the efficiency with which the responding cells initiate differentiation resulting in mature neural lineages. That competence for each of these responses can be mimicked by forskolin/IMX suggests that activation of a single transduction pathway, possibly resulting in the activation of PKA, mediates the induction of competence for both responses in the target stem cell. However, Pax-3 expression and neural induction can be uncoupled. Pax-3 expression occurs in non-neuroectodermal lineages in both the mouse (Goulding et al., 1991) and the P19 (EH3) EC cell in vitro model (Pruitt, 1992). Additionally, neuroectoderm is induced in Splotch homozygotes in which the Pax-3 gene is disrupted (Epstein et al., 1991). These observations demonstrate that Pax-3 expression does not commit cells to a neuroectodermal lineage and is not required for neural induction. Additionally, the present study demonstrates that Pax-3 expression can occur in the absence of neuroectodermal differentiation in stem cells that have not received the neural competence signal. However, it is less clear whether neuroectoderm can be induced without activating, at least transiently, Pax-3 expression. In both the mouse and chick, Pax-3 is expressed only in dorsal regions of the neuroectoderm (Goulding et al., 1991; 1993). However, notochord ablation and transplantation studies demonstrate that this localization is due to a repression of Pax-3 expression when the neuroectoderm is ventralized since, in the absence of signals from the notochord, its expression domain extends throughout the neural tube (Goulding et al., 1993). Neural differentiation in the absence of associated Pax-3-expressing cells has not been observed in vitro using P19 (EH3) stem cells (unpublished observation). The demonstration that each of the factors required for efficient Pax-3 and neural differentiation in the P19 (EH3) cell model can be obtained in a soluble form will allow these possibilities to be distinguished.
ACKNOWLEDGEMENTS
This work was supported by grants from the American Cancer Society (DB 38) and the American Heart Association.