Different effects of Notch intracellular domain and ESR 1 on MyoD activity in ectodermal cells
Notch signaling is generally thought to affect cellular differentiation and cell fate determination indirectly by biasing a choice between two fates, although it can also be instructive in some tissues (Artavanis-Tsakonas et al. 1999; Gaiano & Fishell 2002; Sato et al. 2002). In order to address the role of Notch signaling on myogenesis in frogs, we started by asking how Notch modifies MyoD function in ectodermal cells. RNA encoding MyoD (1 ng), or MyoD with ICD (0.4 ng, 0.8 ng, 1 ng) was injected into animal poles, and animal caps were collected and subjected to RT–PCR analysis using primers for MA (skeletal muscle actin). As shown in Figure 1A, co-injection of ICD inhibited the expression of MA induced by MyoD (Fig. 1A), suggesting that Notch signaling inhibits myogenesis in ectodermal cells.
Figure 1. Different effects of Notch and ESR 1 on myogenesis. (A–C), RNAs as indicated were injected into animal poles of two cell embryos. Animal caps were explanted at late blastula stage and cultured to stage 20. RNA was extracted and processed for RT–PCR assays. Note that while ICD inhibited MyoD function (A), ESR 1, a downstream effector of Notch signaling (B), did not (C). Amounts of RNA injected: (A) MyoD, 1 ng; ICD (lanes 6–8), 0.4 ng, 0.8 ng and 1 ng, respectively; (B) ICD, 1 ng; (C) ESR 1, 0.8 ng; MyoD, 1 ng; ESR 1 (lanes 6–7), 0.8 ng and 1 ng, respectively.
Download figure to PowerPoint
While Notch signaling inhibits myogenesis in mouse cell lines (Kopan et al. 1994; Kato et al. 1997; Luo et al. 1997; Jarriault et al. 1998; Nofziger et al. 1999) as well as in chicken embryos (Hirsinger et al. 2001), its role in frog myogenesis is rather confusing. For example, expression of an activated mutant Xotch causes an increase in muscle tissue (Coffman et al. 1993), but the same manipulation with a mouse mutant Notch (mNotchIC) leads to repression of myogenesis (Kopan et al. 1994), and yet in embryos where Notch signaling is blocked with a Delta mutant (DeltaStu) the muscle differentiates normally (Jen et al. 1997). In agreement with these reports, as shown below, activation of Notch signaling in the frog embryo, as well as in explanted tissues, often displays different, sometimes opposite phenotypes, which appear to be dependent upon cell types as well as how strongly Notch signaling is activated.
To investigate the mechanism underlying the Notch activity, we expressed ESR 1, a downstream effector of Notch signaling, in an animal cap assay to ask how ESR 1 would affect MyoD function (Fig. 1B; see also Wettstein et al. 1997; Schneider et al. 2001). RNA encoding MyoD, or MyoD with ESR 1 (0.8 ng or 1 ng) was injected into animal poles of 2-cell embryos, and RT–PCR was performed. As shown in Figure 1C, in contrast with the observation that ICD strongly inhibits the function of MyoD, co-expression of ESR 1 did not significantly affect MyoD activity (Fig. 1C).
The ESR 1 effect on myogenesis was investigated by expressing its RNA (0.5 ng) in embryos. In such embryos, expansion of MyoD expression was often seen (62% (55), not shown). This suggests that while considered as a transcription repressor, ESR 1 can enhance expression of some genes, such as MyoD (and Delta; see below), although the mechanism is currently not known. This finding further implies that ESR 1 is not the effector that mediates the Notch-triggered inhibitory effect on myogenesis and that there must be another downstream component in the Notch signaling pathway.
Hairy inhibits myogenesis
We considered Hairy to be an excellent candidate for being downstream of Notch and negatively regulating MyoD activity. This is because the promoter of the Hairy 2 gene contains potential binding sites for Su(H) (Davis et al. 2001), and Hairy 1 severely inhibits myogenesis (Umbhauer et al. 2001). Like Hairy 1, Hairy 2 also shows strong antimyogenic activity, as injection of its RNA (0.1–0.2 ng) downregulated expression of endogenous MyoD (55% (31)), the general mesoderm marker Xbra (83% (23); arrows in Fig. 2A), as well as Xwnt8 (74% (23); data not shown). Interestingly, like Hairy 1, Hairy 2 did not inhibit other dorsal markers such as goosecoid or chordin (data not shown; Umbhauer et al. 2001).
Figure 2. Hairy inhibits myogenesis. (A), Embryos were injected into marginal zone of 2-cell embryos with β-gal, alone or with Hairy 2. At gastrula stage the embryos were fixed and stained with Red Gal and MyoD or Xbra, respectively, by in situ hybridization. (B), Embryos (the anterior is up), injected with Tex Red Dextran alone or with Hairy 2, were fixed at stage 25 and stained with the muscle antibody 12/101. (C), RT–PCR analysis shows Hairy inhibits MA expression induced by MyoD. Amounts of RNA injected: (A–B) Hairy 2, 0.1–0.2 ng; β-gal, 0.2 ng; Tex Red Dextran, 10 ng; (C) Hairy 2, 0.2 ng; MyoD, 1 ng; Hairy 2 (lanes 6–8), 0.05 ng, 0.1 ng, and 0.2 ng, respectively.
Download figure to PowerPoint
Consistent with the finding that the early mesoderm markers were blocked by expression of Hairy 2, differentiated muscle, as stained with a monoclonal antibody 12/101, was also lost on the injection side of the embryo (81% (38)), suggesting that muscle failed to form in the affected area (arrow, Fig. 2B). In agreement with this, co-expression of Hairy 2 RNA inhibited MA induction by MyoD in ectodermal cells (Fig. 2C). Therefore Hairy is a candidate for being in the Notch signaling pathway and mediating Notch-induced inhibition on myogenesis.
Since Hairy 2 and Hairy 1 share functional redundancy, and Hairy 2 is more abundantly expressed than Hairy 1 in frog embryos (see below), we have used Hairy 2 rather than Hairy 1 in our experiments.
Hairy is under the regulation of Notch signaling
We then tested whether Hairy 2 is under the regulation of the Notch pathway in frog embryos. While previous studies have suggested that Hes 1 may be an effector for Notch signaling in some cells, it functions independently of the Notch cascade in other cells. For example, Hes 1 expression is not disturbed in Notch 1 mutant embryos (de la Pompa et al. 1997).
In situ hybridization with an antisense probe for Hairy 2 was performed in embryos injected unilaterally with either β-galactosidase (β-gal; 0.2 ng) alone (Fig. 3A) or with ICD (0.5 ng; Fig. 3B,C). β-gal was used to trace the injected cells. While expression of β-gal never induced expression of Hairy 2 (48), ICD not only strongly expanded its endogenous expression domain, but also induced a significant amount ectopically (79% (86); Fig. 3B).
Figure 3. Regulation of Hairy by Notch. (A–C) RNA encoding β-gal alone (A), or together with ICD (B,C), was injected into embryos at the 2-cell stage, and at stage 20 the embryos were stained with Red Gal and processed with a Hairy 2 probe. (C), Close-up of an embryo stained with Red Gal and Hairy 2. Arrows mark cells with both Red Gal and Hairy 2 stain. Arrowhead indicates a cell with Red Gal but less Hairy 2 induction. The inset (C′), shows cells with Red Gal alone. The diffuse blue stain in the upper half of panel (C) was from the endogenous Hairy 2. (D–E), Animal cap assay, performed as in Fig. 1, shows induction of Hairy 2 by ICD (D) and Neurogenin (E). (F–H), Induction of Hairy 2 by Neurogenin. Note the ectopic stain of Hairy 2 in the Neurogenin injected side (arrow, G) as compared with the uninjected side (F). (H), Close-up of an embryo showing induction of Hairy 2 by Neurogenin. The arrow points to a pair of cells with Red Gal and Hairy 2, respectively. (I), Neurogenin activates the neurogenic pathway, which includes among others (such as NCAM and Nrp 1, not shown in the diagram) NeuroD and N-tubulin, in cell 1 (Koyano-Nakagawa et al. 1999), at the same time, together with NeuroD, induces expression of Delta (Chitnis & Kintner 1996; Koyano-Nakagawa et al. 1999) and therefore activates Notch signaling in cell 2. Consequently, in contrast to Notch, Neurogenin has a non-cell autonomous effect in inducing Hairy. Embryos face down in panels A, B, F and G. Amounts of RNA injected: (A–C) ICD, 0.5 ng; β-gal, 0.2 ng; (D, lanes 6–8) ICD, 0.5 ng; (lanes 9–11), 1 ng; (E) DeltaStu, 0.2 ng; Ngn, 10 pg; DeltaStu (lanes 6–8), 0.1 ng, 0.2 ng, and 0.4 ng, respectively; (F–H) Ngn, 10 pg.
Download figure to PowerPoint
An RT–PCR analysis was performed to test this further in ectodermal cells. As shown in Figure 3D, while there was some level of Hairy 2 RNA expressed in uninjected animal caps, injection of ICD increased the RT–PCR signal by an average of 2.5-fold (for 0.5 ng of ICD RNA) and 4-fold (for 1 ng of ICD RNA), as determined by densitometry analysis of the bands (Fig. 3D). This suggests that, as in the PSM (Jouve et al. 2000) and the neural progenitor cells (Ohtsuka et al. 1999), Hairy (both Hairy 2 and Hairy 1; data not shown) may act as an effector for Notch signaling.
If Notch lies upstream of Hairy 2 and upregulates its expression, Neurogenin should also do so. Neurogenin is a bHLH protein and plays a critical role in promoting neuronal differentiation in Xenopus embryos (Ma et al. 1996). Like other bHLH proteins such as XASH-3, NeuroD, and ATH-3, Neurogenin can induce neuronal differentiation in the neural epithelium of the neural plate as well as in the non-neural ectoderm (Lee et al. 1995; Chitnis & Kintner 1996; Ma et al. 1996; Takebayashi et al. 1997). At the same time, Neurogenin induces Delta expression to activate Notch signaling in a negative feedback loop that inhibits neuronal differentiation (Ma et al. 1996; Koyano-Nakagawa et al. 1999). An elevated level of Delta may also activate Notch signaling by upregulating Notch expression (Huppert et al. 1997).
Consistent with these observations, Neurogenin (10 pg) upregulated expression of Hairy 2 transcripts in animal caps (Fig. 3E), and this induction appears to be dependent on Notch signaling, since coexpression of DeltaStu (0.1 ng, 0.2 ng, 0.4 ng), which blocks the Notch pathway (Jen et al. 1997), reduced its expression to the background level (Fig. 3E).
Induction of Hairy 2 transcripts by Neurogenin was confirmed by in situ hybridization. Like Notch, Neurogenin (10 pg) injection not only expanded the endogenous Hairy 2 domain, but also induced Hairy 2 ectopically (88% (44); arrow, Fig. 3G). Since Hairy 2 is induced by Neurogenin through Notch signaling, its transcripts should be present in the cell (cell 2) that neighbors Neurogenin-expressing cell (cell 1, Fig. 3I).
The cell autonomous induction of Hairy 2 by Notch was confirmed by in situ hybridization as shown in Figure 3C (arrows). Similarly, in Neurogenin RNA injected embryos, we were also able to identify some Hairy 2 positive cells at an ectopic site, which were in close proximity to Neurogenin expressing cells (arrow, Fig. 3H), suggesting that the Hairy 2 transcripts in these cells might have been induced by Neurogenin non-cell autonomously.
Taken together, our data are consistent with the previous report in that Hairy is regulated, at least in part, by Notch signaling in frog embryos (Davis et al. 2001).
Mutual regulation of Hairy and ESR 1
Since Hairy is a downstream component in Notch signaling, we wished to know how it relates to ESR 1. As assayed by RT–PCR, overexpression of ESR 1 (0.5 ng, 1 ng) in animal caps led to an upregulation of Hairy 2 (Fig. 4A), suggesting that Hairy 2 may be downstream of ESR 1. However, it is also possible that ESR 1 and Hairy 2 are both downstream to Notch in parallel pathways. Moreover, co-injection of an RNA for a dominant negative ESR 1 with ICD never downregulated expression of Hairy 2 (data not shown), suggesting that Hairy 2 is under regulation of both Notch and ESR 1.
Figure 4. Mutual regulation of Hairy and ESR 1. RT–PCR was performed in A, B, D, and E to show that ESR 1 upregulated expression of Hairy 2 (A), that Hairy inhibited expression of ESR 1 in Neurogenin-injected caps (B), and that HΔW, a deletion mutant form of Hairy 2 (C), rescued the repressed expression of ESR 1 by Hairy 2 in VMZ (D) and de-repressed expression of Delta and ESR 1 in normal caps (E). (F), Hairy, induced by Neurogenin through Notch signaling, inhibits ESR 1 in cell 2. At the same time Hairy may determine the fate of cell 1 or other cells surrounding cell 2 by regulating Delta. Amounts of RNA injected: (A) ESR 1 (lanes 4–5), 0.5 ng, 1 ng; (B) Nog, 10 pg; Hairy 2, 0.2 ng; Ngn/nog (lanes 6–8), 10 pg/10 pg; Hairy 2 (lanes 7–8), 0.1 ng, and 0.2 ng, respectively; (D) HΔW, 0.2 ng; ICD/HΔW, 1 ng/0.2 ng; ICD, 1 ng; ICD/Hairy 2 (lanes 7–9), 1 ng/0.4 ng; HΔW (lanes 8–9), 0.2 ng and 0.4 ng, respectively; (E) HΔW, 0.2 ng.
Download figure to PowerPoint
To test how Hairy regulates ESR 1, we overexpressed Neurogenin in neuralized caps (co-injected with 10 pg of noggin) to strongly induce ESR 1 expression (Koyano-Nakagawa et al. 1999), since ICD in these cells often showed a weak induction. As shown in Figure 4B, co-injection of Hairy 2 (0.1 ng, 0.2 ng) with Neurogenin (10 pg) led to a strong reduction in ESR 1 expression (Fig. 4B). This suggests that Hairy inhibits Neurogenin-induced ESR 1 expression. This inhibition is presumably not due to a direct inactivation of the Neurogenin protein since if that were true, Hairy should have inhibited all genes that are induced, including Nrp 1 (see Fig. 7A).
Figure 7. Hairy and Notch on Neurogenesis. (A), An RT–PCR assay was used to test the effect of Hairy on expression of neural genes induced by Neurogenin in animal caps. (B), Expressions of Nrp 1, Sox2, N-tubulin, NeuroD and ESR 1 were examined in embryos expressing ICD and Hairy 2 by in situ hybridization. In the NeuroD panel (k), black dots denote the midline of an ICD-expressing embryo. (C–D), Double-staining in situ hybridization was performed to compare the spatial expression patterns of NeuroD/Hairy 2 (C) and ESR 1/Hairy 2 (D). In the right panel of (C), the embryo was cut into two parts, which were probed for NeuroD (Red) and Hairy 2 (purple), respectively. (E), Misexpression of HΔW led to an expansion of the endogenous ESR 1 (arrows; the control was not shown to save the space). Arrows indicate either an ectopic induction or persisted expression of the endogenous gene; arrowheads mark a loss of the endogenous transcripts. All embryos face down or to the reader except those in the right panels of C and D which are to the left. Amounts of RNA injected: (A) Nog, 10 pg; Hairy 2, 0.2 ng; Ngn/nog, 10 pg/10 pg; Hairy 2 (lanes 7–8), 0.1 ng and 0.2 ng, respectively; (B) ICD, 0.3–0.5 ng; Hairy 2, 0.2 ng; (E) HΔW, 0.5 ng.
Download figure to PowerPoint
We engineered a mutation in the Hairy 2 construct, and used it to inhibit activity of the wild-type protein. Hairy, like other members in the bHLH-WRPW family, has three domains: a bHLH domain, which is at the N-terminal, an Orange domain in the center and a WRPW domain at the C-terminal (Davis & Turner 2001; Fig. 4C). The WRPW domain is required for mediating the inhibitory effect of the protein by interacting with the Groucho family of general transcription repressors (Paroush et al. 1994). Therefore, by deleting this domain, we created a Hairy mutant called HΔW, in the hope that it would interfere with activity of the wild-type protein when they heterodimerize (Davis & Turner 2001). As expected, while Hairy 2 (0.4 ng) inhibited induction of ESR 1 by ICD in VMZ tissues (Fig. 4D), addition of HΔW (0.2 ng, 0.4 ng) to ICD/Hairy 2 (1 ng/0.4 ng) injected tissues overcame the inhibitory effect of Hairy 2 and led to a complete rescue of ESR 1 expression (Fig. 4D). The stronger induction of ESR 1 in ICD/Hairy2/HΔW groups than in the ICD sample suggests that ESR 1 is repressed by the endogenous Hairy, therefore upon alleviating repression, the cells are more responsive to the ICD signal (see below). Similarly in animal caps, HΔW rescued, although partially, the inhibitory effect of Hairy on MyoD-induced MA expression (data not shown). Therefore HΔW acts as a Hairy interfering mutant. To further test the specificity of this mutant, we coexpressed RNAs encoding HΔW (1 ng) and ESR 1 (0.5 ng) in VMZ tissues, to ask if HΔW interferes with the ability of ESR 1 to activate MA (Fig. 6E). Our data confirmed that while HΔW strongly inhibited the activity of Hairy 2, it failed to do so with ESR 1 (Fig. 4D; data not shown).
Figure 6. Hairy regulates Notch activity. RT–PCR was performed to compare Notch signaling on MyoD-induced myogenesis in animal caps and VMZ (A), and how a modified level of Hairy in these cells affected Notch signaling (B–C). (D), Embryos injected with ICD (not shown), HΔW or both were stained with a MyoD probe. Arrow indicates enhanced MyoD expression in an ICD/HΔW injected embryo. (E–F), Isolated VMZ were stained with MA (E) and ESR 1 probes (G), or processed with RT–PCR (F). (H), ESR 1 and Hairy, both downstream to Notch signaling, have opposite effects on myogenesis. The relative levels of these two mediators may decide the general effect of Notch. Amounts of RNA injected: (A) ICD, 1 ng; MyoD, 1 ng; ICD (6–8, 12–14), 0.4 ng, 0.8 ng, and 1 ng, respectively; (B) Hairy 2, 0.4 ng; MyoD, 1 ng; MyoD/Hairy 2, 1 ng/0.2 ng; ICD, 0.8 ng; MyoD/ICD, 1 ng/0.8 ng; Hairy 2 (lanes 9–11), 0.1 ng, 0.2 ng, and 0.4 ng, respectively; (C) HΔW, 0.2 ng; MyoD, 1 ng; MyoD/HΔW, 1 ng/0.2 ng; ICD, 0.8 ng; MyoD/ICD, 1 ng/0.8 ng; HΔW (lanes 9–11), 0.1 ng, 0.2 ng, and 0.4 ng, respectively; (D) ICD, 0.5 ng; HΔW, 0.2 ng; (E–G) ICD, 1 ng; HΔW, 0.4 ng; ESR 1, 0.5 ng.
Download figure to PowerPoint
We then asked how modified Hairy 2 activity in the cell would affect the expression of ESR 1. As shown in Figure 4E, consistent with the observation that addition of Hairy 2 to Neurogenin-injected cells inhibited ESR 1 expression (Fig. 4B), injection of HΔW, which interferes with activity of the wild-type Hairy, increased the ESR 1 signal by 5.7-fold in uninjected animal caps (Fig. 4E), suggesting that the endogenous Hairy negatively controls expression of ESR 1 (see also Fig. 4D).
Significantly, Hairy 2 not only controls the expression of ESR 1, but also regulates Delta. As shown in Figure 4B,E, addition of Hairy 2 led to a reduction of Delta expression in Neurogenin-expressing caps (Fig. 4B), and inhibition of the endogenous Hairy activity by HΔW (0.2 ng) led to a de-repression of Delta in normal animal cap cells (2.6-fold increase; Fig. 4E). Significantly, since the Delta level controls Notch-mediated lateral inhibition (Kunisch et al. 1994), we speculated that Hairy may regulate this process through modifying Delta (Fig. 4F).
Endogenous Hairy controls Notch-mediated Delta inhibition
An animal cap assay was performed in which we tested Delta expression in the cells overexpressing Notch ICD (1 ng) alone, with Hairy 2 (0.2 ng, 0.4 ng) or HΔW (0.2 ng, 0.4 ng). These cells were considered as a single cell in this assay since they all inherited the injected RNAs. Therefore, the Delta level in this cell should indicate if it has strong or weak lateral inhibition were it in contact with another cell(s).
As shown in Figure 5A, expression of HΔW in animal caps led to a robust expression of Delta (and ESR 1); addition of HΔW to ICD also resulted in Delta and ESR 1 upregulation (4–10-fold increase; Fig. 5A). However, when the similar assay was performed in VMZ, we noted that overexpression of Hairy 2, or co-injection of Hairy 2 with ICD did not lead to a significant repression on Delta expression, while the same manipulation led to a strong inhibitory effect on ESR 1. Similarly, coexpression of HΔW with ICD did not lead to an upregulation of Delta, while there was a significant increase in the ESR 1 induction (Fig. 5A).
Figure 5. Hairy controls Notch mediated Delta inhibition. (A) Animal cap or VMZ assay was performed by RT–PCR to test the effect of modulated levels of Hairy on Delta or ESR 1 expression. In (B), expression of Hairy 2/1, ESR 1 and Delta in animal caps or VMZ was compared at different stages. (C), A mid-gastrula stage embryo, as viewed from different directions, shows Hairy 2 expression. The embryo, in the panels viewed from the vegetal pole (vegetal) and side (side), was oriented so that the anterior was to the left. (D), An animal cap assay was performed showing that ESR 1 upregulated the expression of Delta. (E), Hairy likely mediates the inhibitory effect of Notch signaling on Delta. However, how Notch regulates Delta, and hence the fate of cell 2, and ESR 1 depends on the expression level of Hairy in cell 1. Question marks indicate possibilities so that the strength of Notch signaling in cell 2 correlates negatively with the endogenous Hairy level in cell 1. Note it may require a higher level of endogenous Hairy for Notch to inhibit Delta than to inhibit ESR 1 and other genes. Amounts of RNA injected: (A) HΔW, 0.2 ng; ICD, 1 ng; HΔW (lanes 6–7, 16–17), 0.2 ng and 0.4 ng, respectively; Hairy 2, 0.2 ng; Hairy 2 (lanes 11–12), 0.2 ng and 0.4 ng, respectively; (E) ESR 1, 0.5 ng.
Download figure to PowerPoint
Notably, while overexpression of ICD in animal caps led to a nearly complete repression of Delta expression (90% reduction; lanes 3 and 5, Fig. 5A), the same injection did not have a profound effect in VMZ (5% reduction; lanes 8 and 10, Fig. 5A). This suggests that Notch could mediate either a strong (in animal caps) or weak (in VMZ) lateral inhibition in different cells.
Since manipulation of Hairy activities could affect Delta expression, this cell context-dependent Delta inhibition mediated by Notch signaling may be controlled by levels of Hairy in these cells. Another possibility is that Notch signaling can generate different amounts of Hairy in ectoderm and VMZ, which has been ruled out by our comparison analysis (data not shown). We then assayed for Hairy expression in these two tissues by RT–PCR. Tissues were prepared at stage 10 and either processed for the assay immediately or left to age to stage 13 or 20. As shown in Figure 5B, in all stages examined, there was a marked difference in Hairy 2 expression in these tissues (1.5–3.6-fold difference). Hairy 2 was expressed consistently at higher levels in ectodermal cells than in VMZ. Hairy 1 expression followed the same pattern although at the early neurula stage, there was no significant difference in the two cells. Therefore, the different levels of Hairy in ectodermal versus VMZ likely accounts for the different effect from Notch signaling in these cells. Consistent with the observation that Hairy inhibits expression of Delta (Fig. 4B), in ectodermal cells where there was high Hairy expression, Delta is expressed at low levels. Conversely, in VMZ where there was a low level of Hairy, Delta is high. The difference for Delta expression between these two cells was much larger at early stages (3.2-fold difference at stage 10, and 2.6-fold at stage 13), and then the difference reduced to minimal at later stages (Fig. 5B). Interestingly, ESR 1, which represents the readout of Notch signaling, was expressed at higher levels in ectodermal cells at early gastrula/neurula stages, but this pattern was reversed at tadpole stages. This suggests that in addition to Notch, other signaling cascades or mechanisms may be involved in regulating the expression of Hairy 2 (Davis et al. 2001).
Hairy 2 expression at midgastrula stage was examined by in situ hybridization. As shown in Figure 5C, the transcripts were abundantly present in the ectodermal and dorsal mesodermal cells, but expressed less in the ventral side, confirming the results from our PCR analysis.
Importantly, activation of the Notch pathway in the embryo leads to upregulation of Hairy (Fig. 3); however, this induced Hairy does not compromise the Notch cell context-dependent effect. For example, ICD injection results in much stronger induction of ESR 1 in VMZ than in ectodermal cells (lanes 10 and 5, Fig. 5A), suggesting that the endogenous Hairy is an important regulator.
Furthermore, whether Notch can inhibit Delta is also dependent on the endogenous Hairy activity. Only when it is high, such as in ectodermal cells (cell 1, Fig. 5E), can Delta be downregulated by Notch activation. In contrast, in VMZ cells where the endogenous Hairy activity is low, Delta expression is resistant to Notch regulation. Interestingly, while ESR 1 upregulates Delta (Fig. 5D), its own expression is subjected to regulation by Hairy (Fig. 4B,E). This suggests that while ESR 1 plays a part in regulating Delta, Hairy, by inhibiting both Delta and ESR 1, is more important in this regard. Nevertheless, conceivably, ESR 1 may exhibit its effect in some developmental processes such as myogenesis when the endogenous Hairy level is low (Fig. 5E).
Hairy determines Notch activity on myogenesis
We compared the effects of Notch signaling on MyoD-induced myogenesis in ectodermal cells and VMZ. Consistently, while ICD (0.4 ng, 0.8 ng,1 ng) displayed a strong inhibition on MA expression induced by MyoD (1 ng), it did not seem to inhibit MA in VMZ (Fig. 6A). This suggests that for myogenesis, Notch is an inhibitory signal in the ectodermal cells for this dose range, while it is permissive or neutral in isolated VMZ. Since Hairy is expressed at different levels, that may account for the different effect of Notch signaling in these two cells.
To test this, we manipulated the levels of endogenous Hairy in these cells, asking how that would affect Notch signaling. In a VMZ assay, we expressed RNA for MyoD (1 ng), either alone or with ICD (0.8 ng), in the presence of increased Hairy 2 RNA (0.1 ng, 0.2 ng, 0.4 ng). MA expression was assayed as shown in Figure 6B. Consistent with its effect on myogenesis, the addition of Hairy 2 switched Notch from being permissive to inhibitory, as evidenced by the downregulation of MA expression (lanes 10, 11 over lane 8, Fig. 6B).
The activity of endogenous Hairy in ectodermal cells was brought down by expression of HΔW and the effect of Notch signaling on myogenesis was examined in this compromised background. As shown in Figure 6C, while ICD (0.8 ng) strongly inhibited MyoD-induced MA induction in normal ectodermal cells (lane 8 over lane 5, Fig. 6C), MA induction recovered significantly in cells when Hairy 2 activity was decreased with HΔW (0.1 ng, 0.2 ng, 0.4 ng; 2-fold increase in lane 11 over lane 8, Fig. 6C). While comparison of MA levels in these samples indicated that a lower level of Hairy in animal cap cells still could not change Notch from being inhibitory to enhancing, HΔW RNA injection clearly made Notch embark on this track. The incomplete rescue of the ICD inhibitory effect on MyoD, echoed by the similarly incomplete rescue of ICD inhibition on Delta expression in ectodermal cells (Fig. 5A), suggests that the cellular microenvironment established by the high level of Hairy defies exogenous efforts to compromise it. VMZ cells are even more refractory, that is, addition of Hairy could not in any way restore the ability of ICD to inhibit Delta (Fig. 5A). These findings highlight an important role for the endogenous Hairy in establishing a cell context.
We examined the effect of Notch signaling on myogenesis in frog embryos by in situ hybridization. While expression of ICD (0.5 ng; data not shown) or HΔW (0.2 ng) did not have any effect on MyoD expression, co-injection of these RNAs led to a significant increase in MyoD transcripts in a majority of the injected embryos (55% (38); arrow, Fig. 6D). This was further confirmed by RT–PCR, where co-injection of ICD with HΔW enhanced MyoD expression by 23% over the control, and more than 70% increase over the ICD group (10 individual embryos were analyzed; data not shown). This suggests that when Hairy level is low enough, Notch signaling enhances the myogenic pathway.
In order to test this in a more defined context, VMZ were isolated from embryos which had been injected with RNA for HΔW (0.4 ng), ICD (1 ng), or both, and assayed for MA expression by in situ hybridization and RT–PCR analysis. As shown in Figure 6E, while ICD RNA injection induced expression of MA in some VMZ, coexpression of ICD with HΔW led to its induction at a much higher level, which was confirmed by RT–PCR (Fig. 6F).
ESR 1 was able to induce MA in VMZ by itself (Fig. 6E,F), therefore a higher level of MA induction by ICD/HΔW injection might be due to the enhanced expression of ESR 1, which was repressed by the endogenous Hairy (Fig. 4E). Consistent with this hypothesis, co-injection of ICD and HΔW led to a higher expression of ESR 1 over the ICD group in VMZ (Fig. 6G; also Fig. 5A).
Collectively, these data suggest that the level of Hairy in a cell determines whether Notch signaling is inhibitory, permissive, or enhancing for myogenesis in the frog embryo. It may do so, at least partially, by regulating the level of ESR 1 (Fig. 6H).
More importantly, since the endogenous Hairy appears to selectively regulate a large number of genes, we were prompted to consider Hairy as a cell context signal, which controls how a cell responds to an external signal, such as Notch. As a cell context signal, its own expression is regulated by different mechanisms. In support of this, while the Hairy 2 promoter contains binding sites for Su(H) (Davis et al. 2001) and its expression can be upregulated by Notch, misexpression of DeltaStu does not significantly decrease its endogenous transcripts (Fig. 3E), and the spatial expression pattern in the embryo does not always correlate with that of ESR 1 (Fig. 5B; see also Davis et al. 2001). Together, our data suggest that Hairy, regulated by Notch and other mechanisms, provides a cell (or tissue) specific setting which may bias the cell towards a particular decision in response to Notch, or other external stimuli. This may explain why Notch often displays opposite effects in different contexts.
Hairy and Notch in neurogenesis
To test the Hairy–Notch relationship in other systems, we chose to examine neurogenesis, since Notch is known to regulate neural development. In Xenopus, an early landmark for neurogenesis is the expression of N-tubulin, whose expression marks neurons (Oschwald et al. 1991; Chitnis et al. 1995). Previous studies have shown that the generation of neurons from neural progenitors is promoted by the activity of a variety of neural bHLH proteins and a balanced inhibition by Delta–Notch signaling. Neurogenin, and other bHLH proteins, not only promotes expression of genes involved in neuronal differentiation such as NCAM (a general neural marker; Kintner & Melton 1987), NeuroD and N-tubulin (markers of differentiated neurons; Lee et al. 1995; Ma et al. 1996), it activates the lateral inhibition machinery at the same time (Chitnis 1995; Chitnis & Kintner 1996; Koyano-Nakagawa et al. 1999). We therefore tested if Hairy could affect Neurogenin-induced neurogenesis in ectodermal cells.
Consistent with our observation that a high level of Hairy makes Notch signaling inhibitory, co-injection of RNA for Hairy (0.1 ng, 0.2 ng) with Neurogenin (10 pg) in ectodermal cells led to inhibition of some neural genes such as N-tubulin, NCAM, and NeuroD, although Nrp 1 was essentially not affected (Fig. 7A). An alternative interpretation is that a high level of Hairy itself is inhibitory, thus inhibiting the Neurogenin-induced neurogenesis.
We compared the spatial expression patterns of these neural markers in embryos expressing either Notch ICD or Hairy 2 in order to further study the relationship between them. Expression of Nrp 1 as well as Sox2 and Sox3 in either ICD or Hairy 2-injected embryos (not shown) was not dramatically reduced; occasionally there was some enhancement instead (Nrp 1, 14% (21); Sox2, 12% (32) for ICD-injected embryos; arrows, Fig. 7Bb and 7Bd). Since Nrp 1 is a pan-neural marker, and Sox2 and Sox3 are important for neural competence (Kishi et al. 2000), these findings suggest that there is an intrinsic link between neural competence and Notch/Hairy signaling.
We further examined the expression of N-tubulin, NeuroD and the Notch downstream gene ESR 1 in ICD- and Hairy 2-injected embryos. When ICD (0.3 ng–0.5 ng) was injected unilaterally into embryos, N-tubulin was inhibited uniformly throughout the injected side (94% (31); arrowhead, Fig. 7Bf). However, NeuroD was more severely inhibited in the head area of the embryo than in the trunk (52% (29); arrows/arrowhead, Fig. 7Bj and 7Bk); in the injected embryos, the highly expressed anterior domain of NeuroD was often completely lost while the weaker, posterior strips often persisted, although at a significantly reduced level. This suggests that for NeuroD expression, injection of the same amount of Notch can elicit different responses, which are cell context-dependent. Consistent with our observation, injection of NeuroD causes formation of ectopic neurons of high density in the trunk of Xenopus embryos (Chitnis & Kintner 1996), which was interpreted as NeuroD being relatively refractory to Notch inhibition.
ESR 1 expression offers a more striking example for this cell context-dependent effect of Notch signaling. In ICD-expressing embryos, ectopic ESR 1 was induced only in cells outside of the head region (arrows, Fig. 7Bn,o); within the head area, ESR 1 was never induced and the endogenous ESR 1 was even inhibited, sometimes completely by the ICD injection (67% (36); arrowheads, Fig. 7Bn,o). This suggests that Notch signaling, dependent on different cell contexts, regulates these neural genes differently. Since in myogenesis Hairy controls Notch activity, and indeed Hairy 2 is expressed at a much higher level in the head than in the trunk (Fig. 3A; Fig. 7C,D), it is likely that Hairy also regulates Notch signaling in neural development.
There are some significant differences between Hairy 2 and ICD effects on these genes. While injection of Hairy, like ICD, inhibited expression of NeuroD (53% (30)) and ESR 1 (65% (34)) more strongly in the head than in the trunk (arrowheads, Fig. 7Bl,p), ectopic N-tubulin expression was sometimes seen in the trunk (34% (35); arrow, Fig. 7Bh), although in other embryos its expression was lost throughout the injection side (60% (37); arrowhead, Fig. 7Bg). The ectopic N-tubulin expression was never seen in ICD-injected embryos. This is likely because in the trunk, low levels of the endogenous Hairy makes Hairy a permissive, or even weakly enhancing signal for some genes such as NeuroD, Nrp 1, and the neural competence factors Sox2 and Sox3; alternatively, this may be due to possible interactions between Hairy and other signaling cascades, such as BMP (data not shown). In contrast, Notch signaling is more likely to be permissive than enhancing, since another downstream gene, ESR 1, also inhibits neurogenesis (data not shown; see also Schneider et al. 2001).
The cell context-dependent effect of Hairy on expression of these genes suggests that Hairy is inhibitory in the head, and permissive or enhancing in the trunk for the genes studied. These different effects must have an impact on the expression of these endogenous genes. To explore this possibility, double-staining in situ hybridization was performed to examine the expression patterns of NeuroD and ESR 1, respectively, with Hairy 2. Consistent with our observation that Hairy has different effects in different regions of the embryo, in the head, neither NeuroD nor ESR 1 was expressed in the same domain as Hairy 2, while in the trunk, both were coexpressed with Hairy 2 (Fig. 7C,D).
The complementary expression of NeuroD and ESR 1 with Hairy 2 in the head of Xenopus embryos suggests that in that region, Hairy may have a role in restricting the expression domain of these genes. To test this, the endogenous Hairy 2 was inhibited by HΔW and expression of NeuroD and ESR 1 was examined in these embryos. As shown in Figure 7E, misexpression of HΔW led to a weak expansion of the endogenous ESR 1, although no NeuroD expansion was noticeable (data not shown). This suggests that Hairy 2 levels must be higher in individual cells in the head than in the trunk, so that cells only in this region can repress the endognous ESR 1; furthermore, as a cell context signal, Hairy may have a preset hierarchy in controlling expression of its target genes to ensure a well defined response to external stimuli.