The Notch response inhibitor DAPT enhances neuronal differentiation in embryonic stem cell-derived embryoid bodies independently of sonic hedgehog signaling

Authors


Abstract

During development of the neural tube, inhibition of the Notch response as well as the activation of the Sonic Hedgehog (Shh) response results in the formation of neuronal cell types. To determine whether Shh and Notch act independently, we tested the effects of the Notch inhibitor DAPT (N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester) on neuralized, embryonic stem (ES) cell-derived embryoid bodies (EBs), while varying the levels of Shh pathway activation. Shh-resistant EBs were derived from Smo null ES cells, while EBs with constitutive high level of Shh pathway activation were derived from Ptc1 null ES cells. Intermediate levels of Shh pathway activation was achieved by the addition of ShhN to the EB culture medium. It was found that DAPT-mediated inhibition of the Notch response resulted in enhanced neuronal differentiation. In the absence of Shh, more interneurons were detected, while the main effect of DAPT on EBs with an activated Shh response was the precocious loss of ventral neuronal precursor-specific markers. Developmental Dynamics 236:886–892, 2007. © 2007 Wiley-Liss, Inc.

INTRODUCTION

Notch signaling is involved at multiple times and locations during development to control cellular fate changes. Notch signaling is particularly well studied in Drosophila, both in the central and peripheral nervous system (Jan and Jan,1994). In this context, a fundamental role for Notch signaling has been determined; the resolution of a competition for neural cell fate, in which a single cell suppresses neural differentiation in its surrounding equivalent sister cells by activating the Notch response in these cells. This signal is mediated by cell surface bound ligands, such as Delta and Serrate signaling for which Notch is the receptor.

Activation of the Notch response is mediated by means of the Notch intracellular domain, which is cleaved away from the full-length receptor in a two-step proteolytic process, one of which is mediated by a presenilin–γ-secretase complex (Selkoe and Kopan,2003). This complex is also involved in the generation of the ameloid beta peptide, which has prompted a screen for small molecules inhibiting this complex. N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester (DAPT) efficiently blocks the presenilin–γ-secretase complex (Dovey et al.,2001) and, as a consequence, efficiently prevents activation of the Notch response (Sastre et al.,2001; Geling et al.,2002).

Notch signaling plays an important role in the developing vertebrate nervous system. Generally, activation of the Notch response favors the differentiation toward glial cell types, while cells without an activated Notch response differentiate toward neuronal fates (Louvi and Artavanis-Tsakonas,2006). For example, in hdac1 mutant zebrafish embryos, the Notch-induced gene her6 is ectopically expressed, resulting in a loss of motor neurons (Cunliffe,2004), supporting the model that an activated Notch response prevents neuronal differentiation. Sonic Hedgehog (Shh) is a strong inducer of ventral neuronal differentiation (Roelink et al.,1994,1995), and motor neuron induction in vertebrates requires Shh signaling (Chiang et al.,1996; Lewis and Eisen,2001). These observations raise the question of how the Shh and Notch responses are integrated to achieve neuronal patterning and differentiation.

An easily accessible in vitro system to study neuronal differentiation is embryonic stem (ES) cell-derived, neuralized embryoid bodies (EBs). EBs are re-aggregated ES cells that obtain a neural phenotype when cultured in defined, serum-free medium in the presence of retinoic acid (RA; Plachta et al.,2004). Addition of Shh to these cultured EBs results in the induction of motor neurons and other ventral cell types (Wichterle et al.,2002). By using EBs derived from Smo−/− and Ptc1−/− ES cells, we can examine the effects of RA and DAPT in the complete absence or constitutive activation the Shh response, respectively. The complementary heterozygous ES cells were used to generate Shh-responsive EBs, which were grown in the presence or absence of ShhN. Together, this allowed us to achieve various levels of Shh pathway activation. Here, we demonstrate that the Notch signaling inhibitor DAPT enhances the RA-dependent neuronal differentiation in the absence of Shh pathway activation, and the loss of neuronal precursor-specific protein expression when the Shh response is activated.

RESULTS AND DISCUSSION

A recurring problem using small molecule inhibitors and activators in developmental studies is the delivery to the embryo at relevant concentrations. In vitro cultures of embryonic explants have addressed this issue to a certain extent, but these experiments are often hampered by the difficulty of obtaining and culturing the tissues. This problem was alleviated by protocols turning ES cells into differentiated tissues in vitro. We determined the relative contributions of Shh signaling and the inhibition of the Notch response in the induction of neuronal differentiation, by varying the following parameters: (1) RA, which induces neuralization in EBs (Wichterle et al.,2002; Plachta et al.,2004); (2) DAPT, which blocks the Notch response (Sastre et al.,2001); and (3) Shh. To assess the role of Shh, these experiments were performed in EBs that are either lacking Smo and are thus completely unable to respond to Shh (Wijgerde et al.,2002), or lacking Ptc1 and consequently have a constitutively activated Shh response (Goodrich et al.,1997). The respective hemizygous EBs were cultured in the presence and absence of 12 nM soluble ShhN (Roelink et al.,1995). In general, we assayed the EBs 4 days after the addition of RA and DAPT, at which point the expression of progenitor markers has peaked, and the expression of neuron-specific markers is present but has not reached its maximum (Wichterle et al.,2002).

To validate this in vitro system, we examined Pax7. It is expressed in all cells of the dorsal neural tube, and Pax7 expression is efficiently inhibited by Shh signaling (Incardona et al.,1998). Pax7 was expressed under all culture conditions where the Shh response was not activated. The highest levels of Pax7 expression were observed in the Smo null EBs grown in the presence of RA (Fig. 1B,C), while lower levels were found in Smo and Ptc1 hemizygous EBs (Fig. 1F,G,J,K). In the absence of RA, some low-level Pax7 expression was also observed, likely representing non-neural expression. Nevertheless, this finding demonstrates that EBs grown under all conditions are responsive to activation of the Shh response. DAPT-mediated inhibition of the Notch response causes a reduction in the level of Pax7 expression (Fig. 1), which we think is a consequence of the formation of more mature neurons.

Figure 1.

Pax7. A–X: Embryoid bodies (EBs) were cultured under the following conditions: retinoic acid (RA) added, B,C,F,G,J,K,N,O,R,S,V and W; DAPT added, C,D,G,H,K,L,O,P,S,T,W, and X. EBs shown in A–D are derived from Smo−/−, and those in E–H and M–P are derived from Smo+/− cells. EBs shown in I–L and Q–T are derived from Ptc1+/− cells, and those shown in U–X are derived from Ptc1−/− cells. EBs shown in M–T are cultured in 12 nM ShhN. Under no condition is Pax7 expression observed where the Shh response is activated (M–X), while RA enhances the expression of Pax7. EBs are responsive to ShhN regardless of the presence of RA and DAPT. Relative expression levels are plotted on the right, and SEM is indicated. Scale bar = 500 μm, each panel has a surface of 3.7 mm2.

The near complete suppression of Pax7 expression by Shh signaling is consistent with previous studies, indicating that Pax7 is efficiently repressed by activation of the Shh pathway in both neural and non-neural tissues. To further determine whether these EBs have acquired a ventral fate in response to Shh, and whether this response is affected by Notch signaling, we examined the expression of Nkx2.2 and Olig2, which are expressed in ventral neuronal precursors (Briscoe et al.,2000). Nkx2.2 was induced in EBs in an RA- and Shh-dependent manner (Fig. 2N,R,V); however, the number of Nkx2.2-expressing cells was significantly curtailed by DAPT (Fig. 2O,S,W). We also observed Nkx2.2-expressing cells in EBs not treated with RA, when Smo is present, but this expression is not affected by DAPT (Fig. 2). Similarly, the expression of Olig2, which is expressed in motor neuron and oligodendrocyte precursors (Takebayashi et al.,2002), was detected only in EBs treated with RA, when the Shh response is activated, either with ligand in the Smo+/− and Ptc1+/− EBs, or due to the loss of Ptc1 (Supplementary Figure S1 N,R,V, which can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat). After 4 days of exposure to DAPT, the expression of Olig2 was significantly reduced (Supplementary Figure S1 O,S,W), similar to that of Nkx2.2. This result is consistent with the published observation that, in neural tube-specific conditional Notch1 null embryos, a significant loss of Nkx2.2 and Olig2 expression was observed (Yang et al.,2006).

Figure 2.

Nkx2.2. A–X: Embryoid bodies (EBs) were cultured under the following conditions: retinoic acid (RA) added, B,C,F,G,J,K,N,O,R,S,V and W; DAPT added, C,D,G,H,K,L,O,P,S,T,W, and X. EBs shown in A–D are derived from Smo−/−, E–H and M–P are derived from Smo+/− cells. EBs shown in I–L and Q–T are derived from Ptc1+/− cells, and those shown in U–X are derived from Ptc1−/− cells. EBs shown in M–T are cultured in 12 nM ShhN. Nkx2.2 is efficiently repressed by RA (B,F,J) and is efficiently induced by Sonic Hedgehog (Shh; N,R,V) and, in the presence of RA, Nkx2.2 is repressed by DAPT (O,S,W). In the absence of RA, Nkx2.2 is expressed in a Smo-dependent manner, unless the Shh response is highly activated. Relative expression levels are plotted on the right, and SEM is indicated. Scale bar = 250 μm, each panel has a surface of 0.92 mm2.

To determine how DAPT affects the induction or maintenance of the neuronal precursor population, we investigated the expression of Nkx2.2 over a 5-day period. To maintain a high level of Shh pathway activation, we performed this experiment in ptc1 null cells. DAPT had no effect on the initial induction of Nkx2.2, which starts being expressed between 1 and 2 days after the addition of RA (Fig. 3C,I). This finding demonstrates that inhibiting the Notch response has no effect on the initial induction of ventral neuronal precursors. However, a significant reduction in Nkx2.2 expression was observed in DAPT-treated cultures on days 4 and 5 of differentiation, while in the absence of DAPT, Nkx2.2 expression is maintained at relatively constant levels for the entire 5-day period (Fig. 3E,F). Blocking the Notch response allows widespread Nkx2.2 expression for only 2 days, after which Nkx2.2 expression is dramatically reduced and virtually disappearing by day 5 (Fig. 3K,L). These results indicate that the pool of Nkx2.2-expressing ventral neuronal precursors is prematurely lost in the absence of Notch signaling. This may either be caused by the precocious formation of more mature neurons, or the differentiation into non-neuronal cell types.

Figure 3.

Nkx2.2 induction time course. A–L: Embryoid bodies (EBs), derived from ptc1−/− embryonic stem (ES) cells, were cultured in the presence of retinoic acid (RA) to induce neural differentiation, either in the absence of (A–F) or in the presence of DAPT (G–L). D0 is the moment RA and DAPT (where indicated) was added to the culture medium. EBs were cultured for the indicated amount of time (1–5 days) and stained for Nkx2.2 expression. Nkx2.2 is induced between day 1 and 2, regardless of the presence of DAPT, but DAPT caused a premature loss of Nkx2.2 expression at day 4 and 5 (E,F,K,L). Scale bar = 120 μm, each panel has a surface of 0.21 mm2.

To determine whether DAPT affects non-neuronal differentiation in EBs, we examined the expression of FoxA2 (HNF3β), which is expressed in floor plate cells (Ruiz i Altaba and Jessell,1992; Ruiz i Altaba et al.,1993) and requires a strong activation of the Shh response (Roelink et al.,1995), and S100, which is enriched in glial cells (Gomez et al.,1990). In the presence of RA, we did not detect expression of FoxA2, regardless of the level of Shh pathway activation, or the presence of DAPT. However, in the absence of RA, widespread expression of FoxA2 was observed (Supplementary Figure S2), probably representing endodermal differentiation (Abe et al.,1996; Levinson-Dushnik and Benvenisty,1997). Similarly, we observed no significant effect of DAPT on the expression of S100 (Supplementary Figure S3). These results argue against a significant induction of non-neuronal cells in DAPT-treated cultures. This in turn is consistent with DAPT mediating the differentiation of mature neurons in the neural tube.

We tested if DAPT causes an increase in neuronal differentiation, first by assessing its activity to induce neurons in the absence of Shh signaling. Lim1/2 is expressed in dorsal and intermediate interneurons, and its expression is inhibited by Shh signaling (Ericson et al.,1995a,b). Consistent with this finding, we observed widespread Lim1/2 expression in RA-treated Smo−/−, Smo+/−, and Ptc1+/− EBs grown in the absence of Shh (Fig. 4B,F,J), while only occasionally Lim1/2-positive nuclei were observed in EBs grown in the presence of ShhN and in Ptc1−/− EBs (Fig. 4N,R,V). DAPT significantly enhanced Lim1/2 expression in neuralized EBs in the absence of Shh signaling (Fig. 4C,G,K), indicating that Notch signaling is inhibitory to the formation of Lim1/2-positive interneurons. The low level of Lim1/2 expression in EBs with an activated Shh response was essentially unaffected (Fig. 4O,S,W). This observation is consistent with the observation that Notch inhibition promotes neuronal differentiation in pluripotent stem cells (Lowell et al.,2006).

Figure 4.

Lim1/2. A–X: Embryoid bodies (EBs) were cultured under the following conditions: retinoic acid (RA) added, B,C,F,G,J,K,N,O,R,S,V, and W; DAPT added, C,D,G,H,K,L,O,P,S,T,W, and X. EBs shown in A–D are derived from Smo−/−, and those in E–H and M–P are derived from Smo+/− cells. EBs shown in I–L and Q–T are derived from Ptc1+/− cells, and those shown in U–X are derived from Ptc1−/− cells. EBs shown in M–T are cultured in 12 nM ShhN. Induction of Lim1/2 is significantly enhanced by RA (B,F,J), and DAPT further enhances this expression of Lim1/2 (C,G,K). Activation of the Shh response repressed Lim1/2 expression in RA-treated EBs, regardless of the presence of DAPT (N,O,R,S,V,W). Relative expression levels are plotted on the right, and SEM is indicated. Scale bar = 500 μm, each panel has a surface of 3.7 mm2

In the neural tube, Shh is a powerful inducer of ventral neuronal cell types, including motor neurons and a population of interneurons located dorsally to the motor neuron pool, the V2 interneurons. The motor neuron markers HB9 (Fig. 5N,R,V) and Isl1/2 (Supplementary Figure S4 N,R,V), as well as the V2 marker Lim3 (Fig. 6N,R,V) are expressed widely in neuralized EBs in which the Shh response is activated. The inclusion of DAPT in these cultures had no significant effect on HB9, and Lim3 expressing in EBs in which the Shh response was activated. Under conditions of very high activation of the Shh response (Ptc1 null and Ptc1+/− cultured with ShhN), we observed a significant increase in Isl1/2 expression. This finding indicates that, when the Shh response is activated, loss of Notch signaling can cause an increase in the number of motor neurons. However, because DAPT inhibits the expression of the ventral neuron precursor proteins Nkx2.2 and Olig2 (Fig. 2, Supplementary Figure S1), we conclude that, in the absence of Notch signaling, the Shh-induced neurons acquire a more mature phenotype precociously.

Figure 5.

HB9. A–X: Embryoid bodies (EBs) were cultured under the following conditions: retinoic acid (RA) added, B,C,F,G,J,K,N,O,R,S,V, and W; DAPT added, C,D,G,H,K,L,O,P,S,T,W, and X. EBs shown in A–D are derived from Smo−/−, and those in E–H and M–P are derived from Smo+/− cells. EBs shown in I–L and Q–T are derived from Ptc1+/− cells, and those shown in U–X are derived from Ptc1−/− cells. HB9 is highly expressed when the Sonic Hedgehog (Shh) response is activated in the presence of RA, regardless of the presence of DAPT (N,O,R,S,V,W). Relative expression levels are plotted on the right, and SEM is indicated. Scale bar = 100 μm, surface of each panel is 0.15 mm2.

Figure 6.

Lim3. A–X: Embryoid bodies (EBs) were cultured under the following conditions: retinoic acid (RA) added, B,C,F,G,J,K,N,O,R,S,V, and W; DAPT added, C,D,G,H,K,L,O,P,S,T,W, and X. EBs shown in A–D are derived from Smo−/−, and those in E–H and M–P are derived from Smo+/− cells. EBs shown in I–L and Q–T are derived from Ptc1+/− cells, and those shown in U–X are derived from Ptc1−/− cells. Lim3 is expressed when the Sonic Hedgehog (Shh) response is activated in the presence of RA (N,O,R,S,V,W), but is unaffected by DAPT. Relative expression levels are plotted on the right, and SEM is indicated. Scale bar = 100 μm, surface of each panel is 0.15 mm2.

The DAPT activity promoting neuronal differentiation is more pronounced in the absence of the neuron-inducing activity of Shh. The simplest explanation would be that Shh signaling inhibits the Notch response when promoting ventral differentiation, rendering DAPT ineffective. However, this model is not supported by other observations. Misexpressing a constitutively activated Smo allele in the cerebellum causes an up-regulation of Notch1 and Hes5 in the resulting tumors, which are susceptible to both the Shh response inhibitor cyclopamine, as well as to DAPT (Hallahan et al.,2004), indicating that activation of both pathways is required for tumor growth and that the Notch1 response is induced by Shh signaling. This idea is supported by the observation that, in granule neuron precursors, Shh induces the Notch responsive gene Hes1 (Solecki et al.,2001). Together, this indicates that Shh signaling actually activates the Notch response and that both are required for growth of cerebellar cells.

Our results demonstrate that the activities of Notch and Shh signaling on neuronal differentiation can be separated. Regardless whether the Notch is inhibited or not, Shh induces the expression of ventral neuronal markers, such as Isl1/2 and Lim3, while Notch signaling is required to maintain the expression of ventral precursors. These observations are consistent with a model that Notch signaling is involved in the maintenance of precursor cells, while, in parallel and independently, Shh induces ventral cell types (Fig. 7). When Notch signaling is inhibited, intermediate neurons appear in larger numbers in the absence of Shh signaling, while the neurons that appear in response to Shh signaling have lost some of their immature characteristics.

Figure 7.

Diagram summarizing the effects of Notch signal inhibition in combination with Sonic Hedgehog (Shh) response activation in neuralized embryoid bodies (EBs). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

EXPERIMENTAL PROCEDURES

ES Cell Culture and EB Formation

ES cell lines were kindly provided by Matt Scott (Ptc1−/− and +/−) and Andy McMahon (Smo−/− and +/−). The cell lines were maintained in ES medium (DMEM with 4.5 g/L D-glucose, L-glutamine, 110 mg/L sodium pyruvate (Gibco), and 3.7 g/L Na-bicarbonate (J.T. Baker), supplemented with 0.1 mM 2-mercapoethanol (Sigma), 15% fetal bovine serum (Gemini), 1% penicillin–streptomycin–glutamine, 30 mg/ml gentamicin, 1% nonessential amino acids (all Gibco), 1% ES cell nucleosides, and 1,000 U/ml recombinant murine LIF (both Chemicon). For EB differentiation, cells were trypsinized, washed, and diluted to a concentration of 50,000 cells/ml in DFNB (25% DMEM with 4.5 g/L D-glucose, L-glutamine, 110 mg/L sodium pyruvate (Gibco), and 3.7 g/L Na-bicarbonate (J.T. Baker), 50% Neurobasal, 25% Ham's F12 (both Gibco), supplemented with 80 μM β-mercapoethanol (Sigma), 1% penicillin–streptomycin–glutamine, 30 μg/ml gentamicin, and1% B-27 Supplement [all from Gibco]). The cells were grown in nonadherent bacterial-grade Petri dishes for 2 days to allow aggregation into EBs. On day 2, medium was changed to DFNB supplemented with appropriate combinations of 1 μM RA (Sigma), 1 μM DAPT (Calbiochem), and 12 nM ShhN (produced in Baculovirus-infected HiFive cells). EBs were grown for an additional 4 days in this supplemented DFNB with a medium change after 2 days. On day 6, EBs were fixed in 4% paraformaldehyde with 4% sucrose for 40 min on ice, then stained for immunofluorescence using established procedures.

Antibody Reagents and Microscopy

Mouse α-Lim1/2, α-Lim3, α-Pax7, α-FoxA2, α-Nkx2.2, α-Isl1/2 and α-HB9 were obtained from the Developmental Studies Hybridoma Bank, polyclonal goat-Olig2 was from R&D Systems. In all cases, Alexa488- or Alexa568-conjugated secondary antibodies from Molecular Probes were used. Images were collected using a Zeiss Pascal confocal microscope, and all panels in a single figure were processed identically in Adobe Photoshop. Figures were assembled and labeled in Adobe Illustrator and exported as a JPEG image.

Relative marker expression levels in EBs were quantified the following way; in the raw images, using ImageJ (NIH, public domain software), the cross-section surface of an EB was measured and taken as a representation of the number of cells in this plane. The average pixel intensity of this surface was determined (a number between 0 and 255 in the 8-bit images collected). Average pixel intensities measured from several EBs were averaged, and the SEM was calculated. These data were plotted, and the average pixel intensity is shown.

Acknowledgements

We thank Matt Scott for Ptc1 het and null ES cells and Andrew McMahon for Smo het and null ES cells. We also thank Christin Christensen for expert technical assistance.

Ancillary